Minimizing fouling in spiral heat exchangers at a BCTMP mill

peer-reviewedT71Minimizing fouling inspiral heat exchangersat a BCTMP millWINNER OF THENATIONAL STUDENTPROBLEM-SOLVINGCOMPETITIONBy M.R. HaqueAbstract: This paper investigates the cause of calcium oxalate (CaC 2 O 4 ) fouling in spiral heatexchangers. CaC 2 O 4 solubility decreases with lowering temperature. The predicted precipitationrate due to temperature difference was consistent with the observed fouling rate. This observationsuggests that the effluent supersaturates and forms precipitate (scale) upon contact with the coldheat exchanger surface. Suspended particles then accumulate and cause rapid fouling. This studyrecommends using physical separation methods and partial replacement of NaOH with Mg(OH) 2during alkaline bleaching to reduce CaC 2 O 4 content and minimize fouling.HEAT EXCHANGER FOULING or scaling aregeneral terms that involve the depositionof extraneous material on theheat transfer surface of a heatexchanger. Although both depositscommonly occur together, deposits from foulingare known as foulants, whereas deposits fromscaling are known as scales. Scale is a hard mineraldeposit that usually precipitates from thesolution and grows in the same place. Foulantsare often high in organics and are formed fromsuspended solids at a point other than wherethey deposit. Both types of deposits introduce alayer of heat resistance and reduce the operationalcapability of the heat exchanger. In severecases, the deposit can be strong enough to interferewith effluent flow and increase the pressurerequired to maintain flow rate through the heatexchanger.Deposits of inorganic salts are a commonproblem in domestic, commercial and industrialprocesses where ‘dirty’ water is used. Industrializednations worldwide annually spend 30 billion$US to deal with this problem [1]. The pulp andpaper sector contributes significantly to thisexpenditure because of the buildup of inorganicand organic materials in the process water.Calcium carbonate, calcium oxalate and calciumsulfate are the most persistent scaling componentsin the pulp and paper industry. Thesecompounds have low solubility products andtend to deposit where abrupt temperature or pHfluctuations exist [2]. The effect of temperaturechange often makes heat exchanger surfaces idealcandidates for fouling. Frequent fouling ofheat exchangers can cause significant fluctuationsin plant operation and requires both timeand manpower to clean up.In order to take preventive measures to reducethe fouling frequency, it is important to identifythe nature of the scaling component and understandthe fouling mechanism. The mechanismsubsequently exposes the relationship betweenfouling potential and temperature, pH or anyother factor in the process. This thesis investigatesan existing heat recovery system located ina pulp mill which has been plagued with frequentfouling problems.BACKGROUNDHigh Yield BCTMP MillThe undisclosed pulp mill is situated in Quebec(referred to as ‘pulp mill’ henceforth) andemploys a bleached-chemo-thermo-mechanicalpulping (BCTMP) process to convert wood chipsto pulp. This is primarily a mechanical process,with production yields over 80%. Maple, birch oraspen wood are chipped and fed, along withwater, to the refiners, as shown in Fig. 1. In therefiners, chips are converted to fiber under highshear stress and high temperature. The pulpstream is then washed and passed through a twostagealkaline peroxide bleaching stage. Water ispumped to facilitate the flow of bleaching chemicalsand to wash the bleached pulp at the lastpress. The pulp mill operates three highly closedwater loops, as seen in Fig. 1. Water is recycledbetween the latency-washing, refining-washing,and washing-bleaching stages to conserve waterand reduce environmental impact [3]. Water circulation,however, causes accumulation of chemicalssuch as oxalate, extractives, lignin, hemicelluloses,cellulose, calcium and magnesium. Thesechemicals deposit on downstream equipment (ex:heat exchangers) and cause fouling problems [3].Heat Recovery SystemThe heat recovery system was installed in 1994 aspart of the pulp mill’s initiative to recover thermalenergy from the effluent discharged from theBCTMP process. The system plays a critical role inmaintaining the high energy efficiency of themill. The effluent is stored in a storage tank, fromthere it is pumped through the three spiral heatexchangers before being sent to an activatedsludge plant (ASP) for treatment. A fluctuatingeffluent temperature between 70°C to 80‡C representsthe potential low grade thermal energythat is recoverable. The heat recovery system alsohelps maintain the downstream temperaturerequired for ASP operation.The original setup has undergone manychanges to deal with malfunctioning equipmentand constant fouling problems. The latest setupemploys three heat exchangers (HEXs), as shownin Fig. 2. Cold river water at seasonal temperatureM.R. HAQUEDepartment of ChemicalEngineering and AppliedChemistryUniversity of TorontoToronto, ONPULP & PAPER CANADA • 108:4 (2007) • 35

T72scalingFIG. 1. Schematic of pulp flow in BCTMP mill (modifiedfrom Zhang et al [3]).FIG. 2. Effluent flow configuration through the heatexchangers (HEXs) in the recovery system.FIG. 3. Internals of a spiral heat exchanger [5].FIG. 4. Dried hard Scale sample.is used as the coolant liquid. The heated river water is then sentto the mill for use.The three heat exchangers at this mill are refurbished spiralunits manufactured in 1973. The original manufacturer wasAmerican Heat Reclaiming Corporation (AHRC), which hassince been bought by Alfa Laval Inc. Originally one of the heatexchangers was a larger unit, and hence two smaller heatexchangers were arranged in series to have comparable effect.In 2004, however, the larger heat exchanger broke down andwas replaced with a smaller AHRC unit.Spiral heat exchangers are made of two flat plates, which arewrapped around each other to create two concentric channelsof rectangular cross section. The warm effluent and coolingwater flow in countercurrent direction, Fig. 3, maximizing theheat transfer efficiency [4,5].The rectangular single channel geometry means that thecross-sectional area decreases with settling solids, increasingvelocity of the effluent. According to Alfa Laval, the high velocitysweeps and removes the settled solids. This increases theamount of solid that can be present in the effluent without foulingup the heat exchanger [4,5]. For this reason, spiral heatexchangers have been employed in fouling prone applicationssuch as PVC slurry, oleum-processing etc. A decrease in channelspacing, however, decreases the maximum solids concentrationthat can be handled by the heat exchanger. It is possible thatinadequate spacing is a reason for fouling in the AHRC heatexchangers.The spiral heat exchangers have been experiencing severefouling problems since installation. The pulp mill has observedthe formation of porcelain like hard scales on the surface of theheat exchanger. It has also been observed that fiber, particulatematter and other foulants also deposit on the scale, eventuallyblocking up the heat exchanger, Fig. 4.Every time a heat exchanger blocks up, it has to be bypassedand blasted with high-pressure water for scale removal. Thisresults in a loss of heat transfer by the system. It has been estimatedthat $400,000+ has been spent to improve the system withthe intention of reducing fouling rate. Major work has been doneto direct the effluent flow to a storage tank to maintain stable andconstant flow rate. Despite the significant investment and effort,the rapid fouling problem remains. It has also been noted thatHEX 5, as seen in Fig. 5 before and after cleaning, has been foulingup more often than the other two HEXs in the mill. A thoroughunderstanding of the cause of fouling has yet to be determined.Hence, a suitable solution to the problem still awaits.Previous Scale AnalysisIn order to determine the inorganic composition of the scale, X-ray fluorescence spectroscopy was performed after ashing driedscale samples. Table I displays the results of the scale analysisperformed in 1998. It is seen that the inorganic portion of thedeposits is mainly calcium salts. It was also found that calciumoxalate (CaC 2 O 4 ) was the major form of calcium salt present inthe sample. The analysis was showed that 70.4% of the depositwas made of calcium oxalate.Additionally, calcium oxalate deposits have also been foundat other points in the pulp mill. A study of calcium oxalatedeposits has been performed in the disc filter shower and rejectspress samples by Zhang et al [3]. This study provides a mechanismfor the formation of calcium oxalate in a BCTMP mill. Relevantresults from this study have been included in this thesis,which specifically tries to determine the cause of scaling in thespiral heat exchangers.OBJECTIVESThe overall objective of this thesis is to determine the cause offouling in the spiral heat exchangers and suggest practical measuresto reduce formation of scale. Understanding the foulingmechanism would help identify effective solutions. The breakdownof objectives is as follows:a) Characterize the nature of the scale depositsb) Once the scale composition has been identified, speculateon the mechanism of scale formationc) Identify possible upstream processes, operating conditions or36 • 108:4 (2007) • PULP & PAPER CANADA

T74scalingum oxalate in the scale is estimated by thefollowing calculation.CaC 2 O 4 (s)—— CaO(s) + CO 2 (g) + CO(g)(Overall decomposition reaction)Mol. Mass (g/mol) 128 56 44 28Using a basis of 1 mole (128gm) of sample,Theoretical mass loss % during totaldecomposition:Mass of Gas Evaporated—————————— 100% =Total Mass of Sample(44 + 28)gm—————— 100% = 56%128gmActual mass loss % from thermal analysisexperiment, Fig. 11: Step 2 + Step 3:FIG. 6. Positions of sample scale collection.FIG. 7. Pulp mill effluent.Mass of Gas Evaporated—————————— 100% = 49%Total Mass of SamplePercentage of calcium oxalate =49—— 100% = 87.5%56Calcium oxalate has three differenthydrated forms: monohydrate, dihydrateand trihydrate. Monohydrate is the moststable and has the lowest solubility [2].Thus it can be speculated that calciumoxalate monohydrate (CaC 2 O 4 .H 2 O) isthe main component (87.5%) of the heatexchanger deposit. The study by Zhang etal. also showed that the pulp mill disc-filtershower and reject press had calciumoxalate deposition [3]. This suggests thatthe mill has been experiencing calciumoxalate scaling problems elsewhere in thesystem and thus it is likely that in the heatexchangers, the scale type is indeed calciumoxalate.Determining Fouling RateOnce the type of scale has been determined,the next step involves quantifyingfouling rate based on a few simplifiedassumptions. The flow rate trends over aone-year period through HEX 3 & 5 andthrough HEX 4 are plotted in Figs. 12and 13.The flow rate trend shows the extent ofthe fouling problem in the pulp mill.Maximum flow is achieved immediatelyafter high pressure cleaning. Then thescaling material and foulants begin todeposit and slowly reduce the flow rate,eventually plugging up the heat exchanger.The deposit build-up reduces the flowchannel area. This increases the pressuredrop across each heat exchanger andreduces the flow rate through the system.The change in flow rate over time isthus a function of fouling rate. A fasterfouling rate leads to a steep decrease inflow. Subsequently a slower fouling ratetakes longer to affect flow. The change inflow rate over time, the slope in Figs. 12and 13, can thus be used to estimate foulingrate.FIG. 8. Schematic of procedure to measure level of saturation of effluent solution.d(flowrate)—————— = f(Fouling rate)dtThe following assumptions are madein order to quantify the fouling rate.• The spiral heat exchanger surface isflattened into a rectangular cross section,Fig. 14• Effluent velocity remains constant overtime• The fouling rate is constant over timeFor a constant velocity, the volumetricflow rate is defined asVolumetric Flowrate =Velocity Unblocked AreaThe maximum flow occurs at time = 0,when the deposit thickness (t) = 0 andchannel spacing (b) = 15.875mm (heatexchanger inner channel). As time increases,the t is increased by a constant value(fouling rate) until the channel completelyblocks up. The fouling rate that bestmatches the actual flow rate profile withthe theoretical profile, shown in Fig. 15, isselected as the correct estimated foulingrate. The same treatment is applied to analyzeindividual slopes for all data. The averageresults are presented in Table II.It is seen that the series arrangementof HEX 3 and 5 foul up faster than HEX4. This agrees with mill observations thatthey have to clean HEX 5 more often. It isalso seen that the fouling rate throughHEX 3 & 5 increased after April 2005 andthe average maximum flow after scaleremoval decreased after April 2005, notethe boxed area in Fig. 12.Both observations suggest that the heatexchanger cleanup procedure waschanged in April 2005, inversely affectingthe heat exchanger performance. Thepulp mill confirmed that, before April, anexternal company cleaned the heatexchangers with a high-pressure waterhose. Since April, the cleaning has beendone internally with a lower pressure hoseto reduce operating expenses. The effectof inefficient cleaning thus had significantimpact on fouling rate. This is discussedin detail in a later section.Calcium Oxalate FormationOnce it is confirmed that there is excesscalcium oxalate passing through the heatrecovery system, it is important to determinethe source of the calcium andoxalate ions in the BCTMP process.Zhang et al. have done significant workon the formation of oxalate in a BCTMPmill with a similar configuration to ourpulp mill [3]. They have investigated theBCTMP process and identified 3 sourcesof oxalate: input with wood, refining stageand bleaching stage.38 • 108:4 (2007) • PULP & PAPER CANADA

peer-reviewedT75FIG. 9. Data points available around the Heat RecoverySystem.FIG. 10. Thermograph for decomposition of pure calciumoxalate monohydrate.FIG. 11. Thermograph for decomposition of scale sampledried at 85°C from HEX5. FIG. 12. Flow trend through HEX 3 & 5.Their work shows that the input withwood and the refining stage contributedsmall amounts of oxalate. However, a significantamount was formed during peroxidebleaching. Other researchers havealso confirmed peroxide bleaching as amajor oxalate source in BCTMP mills[9,12]. The mechanism of formation hasbeen described as an attack of hydrogenperoxide (H 2 O 2 ) on oxalate precursors(OP) present with fiber in the pulp solution(Fiber-OP). Fiber-OP exists as Fiber-OP-H + , which is in equilibrium with Ca 2+ions to form Fiber-OP-Ca 2+ . Ca 2+ ionsenter the BCTMP process with hard waterand wood. The oxidation of Fiber-OP withH 2 O 2 produces oxalates. The reactionchain is summarized in Fig. 16 [3].Effect of TemperatureIt is well known that crystals formed in onepart of a system and carried elsewhere areless adherent than those crystals formedon site. Therefore, a change of operatingcondition that leads to precipitation withinthe heat exchangers is what causes stablescale formation. For a heat exchanger,temperature is most likely the key factorthat would influence scaling.Figure 17 shows the solubility curve forcalcium oxalate. The solubility values areobtained from various sources compiledin the solubility handbook [10]. The datapoints can be fit with a straight line, suggestinga linear relationship between temperatureand solubility. Thus, contact witha lower temperature surface would reducethe solubility of calcium oxalate and thenprecipitate solids, causing build up ofscale. The equation of the solubility-temperaturegraph is as follows.Solubility (g/L) = 0.0001Temp(°C)+0.0051The solubility graph can be used todetermine the precipitation amountbased on the temperature difference. Thedata for temperature difference is presentedin Fig. 18.Table III displays the precipitationrates formed across each heat exchanger.Using the temperature-solubility relationship,the temperature difference acrosseach heat exchanger is converted to a solubilitydifference value. Maximum flowrate is assumed to be 2100L/min, fromthe average maximum flows achieved,Figs. 12 and 13.Precipitation Rate = Solubility FlowrateFrom the last column in Table III, itcan be seen that HEX 5 has the highestprecipitation rate. This once again agreeswith the mill observation that HEX 5needs to be cleaned more frequently. Toconfirm if the effect of temperature is significant,a theoretical fouling rate is calculated,based on estimates of foulingrates calculated earlier.Fouling rate (kg/day) = Exposed Surface AreaFouling Rate (mm/day) Foulant DensityExposed Surface Area = 90.9 m 2Fouling Rate = 0.07 mm/day, from TableII, HEX 4Density is assumed to be 1500 kg/m 3based on the visible structure of deposits,which appears to have the same structureas clay. Typical clay density is 1500 kg/m 3 .Using the above data, theoretical foulingrate for HEX 4 is found to be 9.54 kg/day.HEX 4 was considered because it is theonly HEX in its line. The series arrangementon the other line makes it difficultto relate flow rate and fouling rate.The underlying assumption in estimatingfouling rate based on flow rate gradientsis that deposit builds up uniformlyacross the entire heat exchanger surface.The fouling rate estimated from the flowrate gradient ignores the fact that depositedsolids at a single point along the effluentflow channel are sufficient to constrictthe flow. In reality, the deposit formationis not uniform. This is visible in Fig. 19.The real fouling rate is thus significantlylower than 9.54 kg/day. The overestimatedfouling rate of 9.54 kg/day howeverprovides a frame of reference for comparisonwith the precipitation rate due totemperature difference across HEX 4,which is 4.35 kg/day.From previous calculations, Table III, itwas found that 4.35 kg/day of calciumoxalate precipitated due to the change inPULP & PAPER CANADA • 108:4 (2007) • 39

T76scalingFIG. 13. Flow trend through HEX 4.FIG. 14. Cross section of flattened heat exchanger channel.FIG. 16. Calcium oxalate formation during peroxide bleaching(Adapted from Zhang et al).TABLE II. Summary of fouling rate through heat recoverysystem.HeatHeatExchanger Exchanger3 & 5 4FIG. 15. Comparison of theoretical flow rate trend andactual flow rate trend through HEX 3 & 5.calcium oxalate solubility with decreasing temperature acrossHEX 4. The proximity of 4.35 kg/day precipitation rate and 9.54kg/day theoretical fouling rate suggest that it is possible thattemperature difference drastically affects calcium oxalate fouling.The difference between the precipitation rate and the foulingrate could be also explained by:• The effect of suspended particles. The effluent contains significantamount of suspended particles, fibers, etc., Fig. 7. These suspendedsolids are likely to contribute to fouling and account fora significant portion of the difference between the two numbers.• The analysis uses average values to estimate the temperaturedifference. Temperature fluctuations in the actual process couldresult in faster deposition of solids.The fact that these numbers are not off by orders of magnitudesuggests that temperature difference has a significant effecton the fouling rate. The next step is to characterize the level ofcalcium oxalate saturation, which leads to the following analysis.Effect of Saturation LevelSeveral researchers have highlighted the importance of saturationlevel on the scaling mechanism [11,12,13]. The procedureto determine saturation level and the amount of suspended calciumoxalate in the effluent was detailed in an earlier section ofData Analysis Units Before After Nov 04-April 05 April 05 Apr 05Flow after cleanup (max) l/hr 2100 1230 2100Rate of flowdecrese (L/hr)/day 40 90 20Fouling Rate mm/day 0.14 0.3 0.07Time to completeblockage # of days 53 12 105this paper. Results are presented in Table IV.Table IV shows that the effluent has a high quantity of suspendedcalcium oxalate particles. The presence of suspendedcalcium oxalate particles suggests that the effluent is saturatedwith calcium oxalate. Therefore, the solubility graph of calciumoxalate can be used to determine the dissolved concentration.The amount of suspended calcium oxalate content is 72 timesgreater than the dissolved content. This means that the effluentis highly saturated.It was earlier mentioned that suspended particles from one partof the system carried elsewhere are less adherent than those particlesformed on the same spot. However, once the deposit forms, itis likely that the bombardment of suspended particles will injectadditional mass into the scale and cause rapid fouling, Fig. 20.Undissolved Calcium Oxalate Particle Flow = Particle Desnisty MaximumFlow0.90(g/L) 2100(L/min) ÷ 1000(kg/g) 1440(min/day) =2700(kg/day)Approximately 2.7 tonnes of calcium oxalate in suspendedparticles are passing through the heat exchanger every day. Thisis a rather high number and, even if a fraction of this massattaches to the scale, the contribution to fouling would be sig-40 • 108:4 (2007) • PULP & PAPER CANADA

peer-reviewedT77FIG. 18. Average temperature data across individual heatexchangers.FIG. 17. Change in solubility with solution temperature forcalcium oxalate monohydrate.FIG. 20. Contribution of suspended particles to deposit formation.TABLE III. Precipitation rates due to temperature differenceacross each heat exchanger.Temp Temp T Solubility Max. Precip.In Out Flow RateCelcius Celcius Celcius g/L L/min kg/dayHEX 4 81 67 14 0.0014 2100 4.35HEX 3 81 75 6 0.0006 2100 1.84HEX 5 75 59 16 0.0016 2100 4.72FIG. 19. Close-up of heat exchanger non-uniform deposit.nificant. The overall high calcium oxalate content in effluentmeans that it is highly prone to fouling.Proposed Fouling MechanismSo far the sources of calcium oxalate have been identified andthe effect of temperature and saturation level established. Theresults suggest the following fouling mechanism taking place.The pulp mill contains three highly closed water loops. Thecontinuous recirculation of water builds up organic and inorganicchemicals and thus the dissolved ions are saturated. Saturationof calcium and oxalate ions was confirmed while characterizingthe effluent. The mechanism begins with the calciumand oxalate ions dissolving in the solution.A key factor in scale formation is supersaturation to the pointof crystallization. This is regarded as the primary cause of scaling[8]. Various researchers have shown that the growth rates ofcalcium oxalate crystals have a second order dependence onsupersaturation [8,13]. Supersaturation is defined as a solutionthat contains a higher concentration of dissolved solute than theequilibrium concentration. The driving force of supersaturationis either a differential concentration at a fixed temperature or adifferential temperature at a fixed concentration. However, achange in pH or pressure that can alter the solubility also affectssupersaturation [8].There is a proportional relationship between calcium oxalatesolubility and effluent temperature, as noted earlier. It is thuslikely that a decrease in temperature as the effluent passesthrough the heat exchanger causes a disruption in the thermodynamicequilibrium, which is necessary for supersaturation.Calcium oxalate is a sparingly soluble salt and needs a smalldegree of supersaturation to precipitate [9,12]. Thus, when theeffluent temperature decreases, the solubility decreases and thesolution become supersaturated before precipitating out.The presence of suspended particles suggests that nucleationsites are not a limiting factor for precipitation. Adequate residencetimes due to insufficient effluent velocity mean that theprecipitated calcium oxalate adheres to the heat exchanger surfaceand starts growing crystals.Once the scale has formed, extraneous materials settle directlyon top of the scale, resulting in rapid fouling. The high percentageof suspended particles and the temperature differenceacross the heat exchangers result in fouling. The higher the temperaturedifference, the faster the fouling rate will be. Figure 21shows a summary of the proposed mechanism.Other researchers have also suggested similar reasons for calciumoxalate scaling. Yu and Ni mention that precipitation ofoxalates is strongly dependent on the solubility of CaC 2 O 4 , ionsupersaturation, nucleation sites, precipitation rate and contacttime. They have found that for a high yield mill, the factors thataffect calcium oxalate solubility are also the key parameters thatinfluence the scale growth rate. These factors are mainly temperatureand pH fluctuations [9]. The effect of pH is probablynegligible, as pH is constant across the heat exchanger.Effect of Heat Exchanger Scale Removal EfficiencyIt was shown earlier that the scale removal efficiency affectedfouling rate. It is expected that inefficient scale removal wouldlower maximum flow rate and reduce the time required for plugup. This effect is seen from a detailed study of the flow ratetrend through HEX 3 & 5, Table II and Fig. 12.From Fig. 12, it is apparent that since April 2005 the pulp millunderwent changes that influenced maximum achievable flowPULP & PAPER CANADA • 108:4 (2007) • 41

T78scalingFIG. 21. Proposed calcium oxalate scaling mechanism.FIG. 22. Effect of temperature difference on overall flowtrend through HEX 3 & 5.TABLE IV. Summary of dissolved and suspended calciumoxalate content in effluent.Temperature Total Total Ratio:Dissolved Suspended Total Suspended/Ca-Ox Ca-Ox Total Dissolvedper Liter per Liter(Celcius) (g/L) (g/L)75 0.013 0.90 72TABLE V. Energy exchanged by individual heat exchangersin the current setup.Temp Temp T (C) Flow HeatIn (C) Out (C) (L/min) Exch. (kW)Heat Exch 4 81 67 14 2100 2050Heat Exch 3 81 75 6 2100 880Heat Exch 5 75 59 16 2100 2340FIG. 23. Possible heat exchanger arrangements.rate after cleaning. The fouling rate has also been faster since.The only difference the pulp mill could identify was switchingfrom an external (high pressure) hose for cleaning to an inhouse(lower pressure) hose, in order to reduce cost.The lower maximum flow (~1200 L/min) after April 2005could be explained by the failure of lower pressure cleaning toremove the deposits completely. So part of the heat exchangerwas already blocked when the operation restarted. This howeverdoes not explain the cause of increased fouling rate.Figure 22 shows the overlapping trend of flow rate throughHEX 3 & 5 and temperature difference ((T) across the heatrecovery system. It is apparent that since April, the temperaturedifference across the heat recovery system is higher. This couldbe the cause of the increased fouling rate because it has alreadybeen established thatT ;Fouling RateHowever temperature difference in later dates (in August2005) once again decreases to a low value. The fouling rate,however, remains fast. This could be due to the fact that the temperaturedifference is plotted around the entire heat recoverysystem and not around individual heat exchangers. The lack ofdata makes it difficult to accurately relate the temperature differenceacross HEX 3 and 5 to the flow rate trend through HEX3 and 5. Other reasons for the increased fouling rate could be achange in the bleaching chemistry or other process alterationsnot identified by the mill.At this stage, there is insufficient information to conclude thatthe change in cleaning procedure has affected fouling rate. Themaximum flow rate after cleaning is definitely lower when thecleaning operation is not performed efficiently. It is, however,possible to achieve higher maximum flow rate even with a lowerpressure hose, because of evidence showing over 2000 L/minflow rate from two peaks achieved after April 2005, Fig. 12.If indeed the reduction in plug up time from 53 days to 12days, Table II, is due to reduced cleaning efficiency, then itimplies that a low pressure cleaning has increased cleaning frequencyby 4 times. The result confirms that a solution to thefouling problem would initially require the heat exchanger surfaceto be entirely clean. This reduces the nucleation sites availableto facilitate calcium oxalate precipitation. Also if scale isalready present when flow is started, then the suspended particlescan settle and contribute to fouling right from the onset.Effect of Heat Exchanger ArrangementThe final discussion is on heat exchanger arrangement, whichaffects both fouling and heat transfer rate. The pulp mill currentlyhas 3 spiral heat exchangers with the same configuration.Figure 23 shows the different combinations of heat exchangerarrangement possible.The existing setup, Fig. 23-a, is the first arrangement to be evaluated.Table V provides a snapshot of the energy exchanged inthis setup. It is seen that for HEX 3, T is only 6°C. The low T42 • 108:4 (2007) • PULP & PAPER CANADA

peer-reviewedT79means that the energy recovered fromHEX 3 accounts for only 20% of the totalenergy exchanged. The low T combinedwith the frequent fouling problem meansthat the current setup is not operating efficiently.In the mill, fouling causes the actualflow rates to be lower than 2100L/minand the heat exchangers often have to bebypassed. Fouling thus creates frequent disruptionsto operation and prevents energyrecovery for a significant time periods.If the three HEXs are operating in parallel,Fig. 23-b, the bulk volume of effluentis diverted through three HEXs instead oftwo. This reduces the flow rate througheach heat exchanger. Spiral heat exchangersoperate via countercurrent flowthrough a single long channel. The lowerflow rate would mean that the mean effluentvelocity is also low. Several researchershave shown that breaking rate for calciumoxalate crystals is proportional to averageshear rate [13,14,15]. Average shear rateincreases with velocity. Thus decreasedvelocity would result in a lower breakingrate and removal of fewer crystals. A lowvelocity also ensures that a higher percentageof precipitated particles have achance to settle on the heat exchangersurface and form stable deposits. This setuppossibly does not have heat transferbenefits because all three streams wouldhave a T slightly higher than 16°C. Thecurrent setup has ~15°C change across theline with HEX 4 and ~22°C (16 +6)change across the line with HEX 3 & 5.The final setup is shown in Fig. 23-c,with two HEXs operating in parallel andthe third heat exchanger set in standbymode. When either operating HEX reachesan advanced stage of fouling, the standbyHEX can be brought into operationwhile scale removal takes place. At maximumflow, the heat exchanged with thissetup is comparable to all three unitsarranged in parallel. There is, however, a20% reduction in heat transfer comparedto the existing setup.The final setup that could have beenconsidered is all three HEXs in series. Butthis scenario is not evaluated based on theassumption that the HEXs would not beable to handle a single flow that is greaterthan 2100L/min. A detailed recommendationbased on the effect of heat exchangerarrangement is provided in the next section.RECOMMENDATIONSThe above sections focused on determiningthe cause of heat exchanger fouling.The high fouling tendency has beenattributed to several factors, such as temperaturedifference, high calcium oxalatecontent and inefficiency of scale removal.Recommendations based on the improvedunderstanding of the heat recovery processare presented in this section.The proposed recommendations are,however, based on limited knowledge ofthe pulp mill process and should be consideredas preliminary. Detailed engineeringstudies/trials have to be conducted todetermine specific economic benefits andthe effectiveness of each recommendation.It is believed that the following suggestionswould improve the robustness of the heatrecovery operation and minimize problemsassociated with fouling. The main recommendationsare presented initially followedby a summary of minor improvements.Operating Heat Exchanger 3 as BackupThe effects of different heat exchangerarrangements have been shown. The currentsetup has been deemed inefficientbecause T across HEX 3 is only 6°C andalso because lack of standby heat exchangersoffer insufficient flexibility in case of ablockage. The overall effect when heatexchangers are bypassed for cleaning isthe flow of effluent with fluctuating temperatureto the ASP plant and loss of heattransfer potential.The recommended setup would be touse two heat exchangers in parallel andconfigure the third heat exchanger instandby, Fig. 23-c. The advantages would be1. Increased operational stability andflexibility. Heat recovery system can beoperational even when one heat exchangeris down for cleaning by diverting flowthrough the standby heat exchanger. Thiswould also allow mill personnel moretime to clean the heat exchangers.2. Over a certain time period, the totalheat exchanged would possibly be higherbecause heat exchangers do not oftenhave to be bypassed.The major disadvantage would be that,at maximum flow, the current setup recoversapproximately 20% more energy dueto a higher (T. The following steps shouldbe taken before proceeding with reconfiguringthe heat exchanger arrangement.1. Manual temperature readings shouldbe taken for longer periods to validateclaims made above.2. Downstream effluent temperaturerequirements (in the ASP plant) must bechecked for lowest temperature requirement.Lowest achievable temperaturewith suggested setup is higher than theexisting setup.3. Alfa Laval Inc. should be contacted tovalidate suggested changes to the process.Converting Feed Storage Tank into a ClarifierThe proposed mechanism suggests thatthe high amount of suspended particles(calcium oxalate, fiber, etc.) contribute tofouling after the scale has started to form.The suspended particles adhere to thescale surface and increase the rate of fouling.It is thus expected that reducing particlecontent would significantly reducethe fouling rate.Different separation techniques (ex:filtration) could be employed to reducethe particle concentration. However, convertingthe feed storage tank into a clarifiercould be another engineering solution.The dimension of the storage tank is8ft (height) by 10ft (diameter). In theexisting setup, effluent is pumped only2.5ft from the bottom. This arrangementinherently pumps a large quantity of fines,fibers and suspended particles to the heatexchangers.Conversion to a clarifier and pumpingthe overflow from the top of the storagetank would drastically reduce the amountof suspended particles pumped throughthe heat recovery system. Detailed engineering(determining particle settlingtime, engineering cost) must be done forthis project. This is expected to result insubstantial reductions in the fouling ratewithin the system.Partially Substituting NaOH with Mg(OH) 2in BleachingAnother potentially effective solutionwould be removing the sources of calciumoxalate. It has been established by severalresearchers, as noted previously, that themain source of oxalate in BCTMP mills isalkaline peroxide bleaching. The pulpmill uses NaOH as the alkali. An investigationby Yu and Ni has shown that partialreplacement of NaOH with Mg(OH) 2 significantlydecreased calcium oxalate formation.Replacing between 30 to 50%NaOH with Mg(OH) 2 resulted in negligiblecalcium oxalate precipitation in laboratoryexperiments. A side benefit was alsoa gain in brightness of pulp [9].The effectiveness of Mg(OH) 2 bleachingis based on the solubility chemistry.When a significant amount of magnesiumis present in the system, the oxalate reactspreferentially with magnesium, ratherthan calcium, to form magnesium oxalate(MgC 2 O 4 ). The lower solubility of magnesiumoxalate means that this form ofthe oxalate would be dissolved in solution[9]. The solubility product (Ksp) forMgC 2 O 4 is 7 x 10-7. The Ksp for CaC 2 O 4is 1.96 x 10-8. Therefore the Ksp for magnesiumoxalate is approximately 35 timeshigher than calcium oxalate at 25°C. Thisindicates substantially higher solubilityand lower tendency to form deposits.The significant reduction of calciumoxalate would reduce the scale formationat other equipment downstream of thebleaching process. A detailed investigationmust be performed to investigate thisreplacement solution before application.The results obtained by Yu and Ni arefrom experiments done on a laboratoryscale. Trials must be conducted to verifythe effectiveness in an industrial scenario.Further RecommendationsThis section details the minor processadjustments that could further improvethe system.• Flow rate trends have shown that evenwith a lower pressure hose, scale can beremoved sufficiently to achieve high flowPULP & PAPER CANADA • 108:4 (2007) • 43

T80scalingrate (>2000L/min). This is not done on aconsistent basis, despite the best efforts ofthe mill personnel. The mill personnelresponsible for cleaning must be suitablyinstructed to spend sufficient time toclean the heat exchangers in a mannerthat removes all visible deposits from thesurface. The lack of deposits when theeffluent flow starts significantly decreasesthe initial fouling rate and substantiallyreduces the time to foul up.• A trial could be arranged to determinethe effect of velocity on scale deposition.The loading of the feed pump could beramped up (if possible) to increase flowrate prior to restarting the heat exchangersafter cleaning. Spiral heat exchangersare designed to remove deposits withvelocity increase. Researchers have alsoshown that increased velocity and shearstrength disrupt agglomeration of calciumoxalate. It is thus possible that a highenough velocity could reduce residencetime sufficiently to prevent scale fromdepositing. However, to prevent equipmentmalfunction, the heat exchangerdesign pressures must not be exceeded.• The application of chemical scaleinhibitors is another potential to reducescale formation. Although limited study ofinhibitors has been included in this thesis,there is a large body of literature thatfocuses on effective chemical anti-scalants.Different groups of chemicals inhibit scaleformation by different mechanisms includingchelation, dispersion and inhibition.Factors that affect selection of an inhibitorinclude water hardness, alkalinity, temperatureand particulate amount [1]. Calciumoxalate scaling is also frequently experiencedin sugar mill evaporators.Poly(acrylic acid) (PAA) and poly(maleicacid) are commonly used in sugar mills forscale inhibition. It has been shown thatPAAs of molecular weight of 2000-4000have optimum effectiveness [16]. Trialscould be implemented in the pulp mill ifthese chemicals are available and have noadverse effect on pulp quality.• All recommendations mentioned arebased on theoretical studies and wouldrequire detailed engineering before application.The dirty nature of the effluent issuch that it should not be expected thatapplying any combination of the aboverecommendations would completely solvethe scaling problem. Scaling to someextent is thus inevitable. It is, however,expected that implementation of theabove recommendations would substantiallyimprove the current fouling problemand definitely increase the plug up time ofthe heat exchangers to more than 2 weeks.CONCLUSIONSThe objectives of this thesis were to determinethe cause of fouling and recommendways to delay or eliminate foulingin the spiral heat exchangers. The objectiveshave been successfully met and thefollowing conclusions reached.1. Calcium oxalate has been identified asthe main scaling component (> 85%).2. The scaling mechanism that has beenproposed begins with calcium and oxalateions dissolving in the mill water. Decreaseof temperature across the heat exchangerreduces the solubility of calcium oxalate,which results in supersaturation followedby precipitation. The precipitated calciumoxalate adheres to the heat exchangersurface and crystal growth begins. Thisleads to the formation of scales. Once thescale has formed, suspended particles settledirectly on top of the scale and contributeto rapid fouling.3. The effluent has been characterized ashaving 72 times more calcium oxalate inthe suspended solids than in the solution,making it highly saturated with calciumoxalate and prone to fouling.4. The precipitation rate due to temperaturedifference and the observed foulingrate across HEX 4 were 4.35 kg/day and9.54 kg/day respectively. These rates aresufficiently close to imply that temperaturedifference affects scaling rate. A highertemperature difference would causemore frequent scale deposition in theheat exchangers.5. The 15°C temperature difference andhigh calcium oxalate content means thatscaling in the heat exchangers to someextent is inevitable.6. It is recommended that decreasing calciumoxalate content would reduce thedeposition rate. Partial replacement ofNaOH with Mg(OH) 2 in the alkalinebleaching stage would form soluble magnesiumoxalate instead of calcium oxalate.7. Application of a separation technique,such as conversion of the feed storagetank into a clarifier, would also reduce calciumoxalate particle content passingthrough the HEXs.8. A temperature difference of only 6°Cacross HEX 3 means that rearranging HEX3 from continuous operation to standbymode will increase both operational flexibilityand heat transfer efficiency.9. Implementation of any or all of the mentionedrecommendations might not completelyeliminate the fouling problem. However,the modifications would significantlyimprove the existing process and increasestability and efficiency and reduce fouling.LITERATURE1. MACADAM, J., PARSONS, S. Calcium carbonatescale formation and control. Reviews in Env Sci andBioTech 3: 159-169 (2004).2. HE, Z., NI, Y., ZHANG, E. Comparison of OxalateFormation from Spruce TMP: Conventional PeroxideBleaching Process Versus the P M Process. Journal ofPulp and Paper Science 29(12): 391-394 (2003).3. ZHANG, J. X., YU, E., NI, Y., ZHOU, Y.,JOLIETTE,D. Calcium and Oxalate In Maple BCTMP Line, Proc.of TAPPSA African Pulp and Paper Week, Durban,South Africa (2002).4. TROM, L. Use spiral plate exchangers for variousapplications. Hydrocarbon Processing 74(5):73-80 (1995).5. Alfa Laval-Spiral Heat Exchangers,Industrial/Application brochure available in manufacturer’swebsite, www.alfalaval.com, AccessedDecember (2005).6. KUTAISH, N., AGGARWAL, P., DOLLIMORE, D.Thermal analysis of calcium oxalate samples obtainedby various preparative routes. Thermochemica Acta297:131-137 (1997).7. KOCIBA, K., GALLAGHER, P. A study of calciumoxalate monohydrate using dynamic differential scanningcalorimetry and other thermoanalytical technique,Thermochimica Acta 282/283 :277-296 (1996).8. COWAN, J.C., WEINTRITT, D.J. Water-formed ScaleDeposits. Gulf Publishing Co. Houston TX. (1976).9. YU, L., NI, Y. Partition of soluble and precipitatedoxalate and its implication on decreasing oxalaterelatedscaling during peroxide bleaching of mechanicalpulps. To be Published (2006).10. SEIDELL, A., LINKE, W.F. Solubility of inorganicand metal organic compounds. Third edition, Vol 1, D.Van. Nostrand Company Inc. New York (1952).11. YU, H., SHEIKHOLESLAMI, R. Effect of ThermodynamicConditions on Fouling of Calcium Oxalateand Silica. AIChE Journal 51(2):641-648 (2005).12. ULMGREN, P., RADESTROM, R. Solubility andMechanisms of Precipitation of Calcium Oxalate in D(Chlorine Dioxide Stage) Filtrates. Journal of Pulp andPaper Science 27(11): 391-396 (2001).13. ZAUNER, R., JONES, A. G. Determination ofnucleation, growth, agglomeration and disruptionkinetics from experimental precipitation data: the calciumoxalate system. Chemical Engineering Science55(19): 4219-4232 (2000).14. COLLIER, A.P., HOUNSLOW, M. J. Growth andAggregation Rates for Calcite and Calcium OxalateMonohydrate. AIChE Journal 45(11): 2298-2305 (1999).15. YU, H., SHEIKHOLESLAMI, R. Modeling of CalciumOxalate and Amorphous Silica Composite Fouling.AIChE Journal 51(4):1214-1220 (2005).16. DOHERTY, W.O.S., et al. Inhibition of CalciumOxalate Monohydrate by Poly(acrylic acid)s with DifferentEnd Groups. Journal of Applied Polymer Science91(3):2035-2041 (2005).Résumé:La présente communication analyse la cause de l’encrassement par l’oxalate de calcium(CaC 2 O 4 ) retrouvé dans les échangeurs spirales. La solubilité du CaC 2 O 4 diminue lorsque latempérature baisse. Le taux de précipitation prévu en raison de la différence de température correspondaitau taux d’encrassement constaté. Cette observation suggère que l’effluent se sursatureet forme un précipité (tartre) au contact de la surface froide de l’échangeur de chaleur. Les particulesen suspension s’accumulent alors et entraînent un encrassement rapide. Nousrecommandons d’utiliser des méthodes de séparation physiques et aussi de remplacer une partiedu NaOH par du Mg(OH) 2 pendant le blanchiment alcalin afin de réduire la teneur en CaC 2 O 4et l’encrassement.Reference: HAQUE, M.R. Minimizing Fouling in Spiral Heat Exchangers at a BCTMP Mill.Pulp & Paper Canada 108(4):T71-80 (April, 2007). Paper presented at 92nd Annual Meeting inMontreal, QC, February 6-10, 2006. Not to be reproduced without permission of PAPTAC.Manuscript received on June 23, 2006. Revised manuscript approved for publication by the ReviewPanel on January 15, 2007.Keywords: SCALING, FOULING, SPIRAL HEAT EXCHANGERS, CALCIUMOXALATE.44 • 108:4 (2007) • PULP & PAPER CANADA