Heat Transfer and Heat Transfer Fouling in Kraft Black Liquor Evaporators

i ,(l ELSEVIER

H. Miiller-Steinhagen • A large number of experiments have been performed with New Zealand Department of Chemical and Process Engineering, Forest Products Kraft black liquor to measure heat transfer coefficients University of Surrey, and fouling rates during convective and subcooled flow boiling heat trans- Guildford, Surrey, fer as a function of surface temperature, bulk temperature, velocity, and United Kingdom solids concentration. Results from experiments with two chemical fouling C. A. Branch inhibitors, with Teflon surface coating and in plate and frame heat ex- Chemical and Process Engineering Department, changers, also are presented. The fouling deposits are analyzed with University of Newcastle, respect to appearance, composition, and process conditions for which they Newcastle upon Tyne, were obtained. With the assumption of chemical reaction-controlled foul- United Kingdom ing, a deposition model is developed and compared with the experimental data. © Elsevier Science Inc., 1997

Keywords: evaporator fouling, Kraft black liquor, fouling mitigation, modelling

INTRODUCTION Number 4 Set Liquor Flow Pattern Pulp mills have always been confronted with organic and Weak black liquor with a solids concentration of 15% is inorganic deposition and with corrosion. Fouling in multi- split between effects 5 and 6. The liquor fed to effect 5 is ple effect evaporators for Kraft pulp black liquor is caused flashed and concentrated before being pumped to effect 6, by materials that are insoluble (such as fibers, sand, scale where the process is repeated. This liquor is recombined flakes, etc.), only moderately soluble (CaCO3, silica, alu- with the other feed stream, which has passed through a minum silicates) or highly soluble (2Na2SO4-Na2CO 3) in separate flash and evaporation section in effect 6. The the liquor [1-3]. Modern operation conditions have combined liquor stream is pumped through the effect 5 severely aggravated these problems. Increasing energy and 6 afterheaters before flowing from effects 4 to 1 in a costs, stronger environmental control, and the trend to- standard countercurrent arrangement. The liquor leaving ward higher yields has led to much higher levels of organic an effect passes through that effect's afterheater before and inorganic solids circulating in the water system. being pumped to the next evaporator. Upon exiting effect Therefore, a knowledge of the basic mechanisms leading 1, the liquor is flashed in a two-stage tank. This vapor is to deposition as well as an understanding of the governing used as an extra source of heat in effects 3 and 4. process parameters have become essential for the design Similarly, the clean condensate from effect 1 also is and economical operation of modern pulp mills. flashed, with the vapor providing extra heat for effects 2 At the New Zealand Forest Products' Kinleith pulp and and 4. paper mill, two different evaporator sets are used to concentrate the black liquor prior to combustion in the Number 5 Set Liquor Flow Pattern recovery furnace. They are the no. 4 and no. 5 multiple effect evaporator sets. The no. 4 set is a conventional Weak black liquor is pumped from the storage tanks and, rising film or long tube vertical unit with split feed. The if necessary, feed sweetened to a concentration of 18% no. 5 evaporator set is a very modern falling film design solids with liquor from the intermediate black liquor tank. that incorporates flow switching in the final, high-con- The liquor then enters dedicated flash sections in effects centration effects to minimize the extent of fouling. Fig- 5, 6, and 7. From effect 7, evaporation is carded out ures 1 and 2 show the no. 4 and no. 5 evaporator set flow sequentially, in a backward feed manner. The final evapO- diagrams, respectively. To model the effect of fouling on ration stage is in the first effect. This effect is fundamen- the performance of the two evaporator sets, a modular tally different from the preceding six effects. It consists of computer code has been developed for the steady state three subeffects that are linked in such a way as to performance [4, 5]. accommodate variable liquor flow patterns between the

Address correspondence to Prof. H. Miiller-Steinhagen, Department of Chemical and Process Engineering, University of Surrey, Guildford, Surrey, GU2 5XH, United Kingdom. Experimental Thermal and Fluid Science 1997; 14:425-437 © Elsevier Science Inc., 1997 0894-1777/97/$17.00 655 Avenue of the Americas, New York, NY 10010 PII S0894-1777(96)00143-4 426 H. Miiller-Steinhagen and C. A. Branch

Steam i- Surface t-.J- Condenser :i

~ ToEfl3

ToEfl4 U"; ~'m°E"' f~kimm~l

V Clean Condensate Return KEY I ..... Condensate - - - Vapour ~ Liquor ...... InternalLiquor

Figure 1. Number 4 evaporator set. subeffects. The liquor flow pattern is switched at regular TEST APPARATUS AND EXPERIMENTAL intervals to ensure that the heating elements are continu- PROCEDURE ally washed with incoming intermediate black liquor. This should prevent the buildup of scale on the heat transfer Heat transfer measurements were performed with the surfaces. Figure 3 shows, however, that the number of experimental rig shown in Figure 4. The fluid was pumped washing cycles required to maintain the design capacity of from a temperature-controlled supply tank through a the high-concentration effect can escalate under certain magnetic flowmeter to the test section, which consists of operating conditions, making plant operation complicated an annulus with a heated core. Details about the test and ineffective. section are given by Wenzel et al. [6]. The stainless steel-

Surface Condenser

i I1AI ~ I1BI ~ I lCl I To effect3 t-r~']"q 17 t'r"~'] l'1 tr"~'"l ~"! I I - ~,i ,,~ ,~i' ~!~!~

Flash " i..~ ...... ~ .... ~ ...... F~ul ~_] "l'ank ~~ [s~ Condensate

CleanreC~r (men sate

KEY

[~ FlashTank ...... Condensate -- Liquor - - -- Vapour

Figure 2. Number 5 evaporator set. Fouling in Kraft Black Liquor Evaporators 427

10 The fouling resistance is calculated from the change in heat transfer coefficient with time: 1 1 [ Rf(t)- a(t) ot o ' (1) .¢: 6 where s 0 is the initial "clean" heat transfer coefficient. ~4 The heat transfer coefficient was calculated from

a (2) A(T w - Tb ) "

More details on the experimental procedure are given by 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 Branch [7]. Date The composition of the fouling deposits was studied by two methods. The SEM/EDS (scanning electron micro- Figure 3. Number 1 effect weak black liquor washes per day. scope/energy dispersive X-ray analysis system) was used to look at the fine structure of the deposit, and, with the aid of the EDAX 9100 system, an elemental analysis of sheathed electrical heater (HTRI design) is equipped with the deposit was obtained. X-ray diffraction was used to four thermocouples located close to the heating surface. identify the different compounds present in a sample. The bulk temperature was measured with thermocouples located in mixing chambers before and after the test section. The flow rate was controlled with a gate valve RESULTS before the test section and with the bypass line, which also Heat Transfer Coefficients ensured good mixing of the liquor in the tank. A 0 to 300 kPa pressure gauge was installed to measure the local Effect of Heat Flux Figures 5a and 5b show the effect pressure in the heated section, which could be altered by of the heat flux on the heat transfer coefficient for liquor varying flow velocity and amount of bypass flow. The concentrations of 15% and 60%, respectively. The param- supply tank was maintained at a predetermined tempera- eter in these diagrams is the flow velocity. One can clearly ture with an internal cooling coil and temperature-con- distinguish two different regimes, depending on the slope trolled heating bands on the external surface. Nitrogen of the curves. In the convective heat transfer regime, the blanketing of the supply tank was used to avoid oxidation heat transfer coefficient is only slightly dependent on the of the liquor during the experiments. The rig was fully heat flux but strongly dependent on the velocity. In con- insulated to minimize heat losses to the ambient air. trast, the heat transfer coefficient increases with increas- Concentrated black liquor was obtained from New Zealand ing heat flux in the nucleate boiling regime but is indepen- Forest Products' Kinleith mill in 60-100-L batches. Be- dent of the velocity. The slight variation of the heat cause of the variability of the liquor, results shown in transfer coefficient with heat flux for the convective heat diagrams in the next section were always obtained from transfer regime as seen in Figure 5b is a result of super- the same sample. posed natural convection currents and changes of the

i~ ~ Meter

Cooling Water

~I Plate Heat Exchanger Steom/ T ~ ,111 ~ I T Hot Woter

Bond Flow Meter Condensate/

i _ / Figure 4. Schematic of experimental Pump rig. 428 H. Miiller-Steinhagen and C. A. Branch

10000 , , , .... , ...... , Effect of Flow Velocity Figure 6 shows three distinctive W regimes with respect to the effect of Reynolds number on 0 0 0 0 0 0 0 00 000 m heat transfer--namely, laminar, transition, and turbulent A o ~ m o o ~ u ta ta ,gl~ flow regimes. For laminar flow, the following approximate relation is obtained: A A NUlam = C1 Re °'45. (3) A A A A &A&& This is consistent with heat transfer with simultaneously developing hydraulic and thermal boundary layers [9]. For turbulent flow, the Nusselt-Reynolds relation can be ap- Flow VCxx~:y proximated by o 100 crns -I IB~ tJquor NUturb = 6' 2 Re °8, (4) T b = 80 °C a 60 OTIS -1 TD.S = 15 20 cms -I which indicates fully developed turbulent flow [10-12]. A

i i , i I ,ill i i i i i iiii i i i transition region exists for Reynolds numbers between 1000 10000 I00000 2000 and 4000. Heat Fhx w "--2" m Correlation of Data The measured heat transfer coefficients were compared with the predictions of a calcula-

1O13OO tion procedure outlined in Ref. 13. Basically, this calcula- W tion procedure consists of a superposition of convective Block Liquor heat transfer--Shah and London [9] and Gnielinski [10] Tb = 80 °C correlations--and boiling heat transfer--Gorenflo [14] Vav --- 60 cm 5-~ 6 correlation--according to a Chen model [15]. Excellent TDS = 60 ~, agreement between measured results and correlations was found for turbulent convective heat transfer and for the

o boiling regime. The 350 convective heat transfer data could be reproduced with an root-mean-square (RMS) error of 7%. Figure 7 shows a comparison between mea- syrrt~ sured and predicted heat transfer coefficients for sub- o Increashg Heat Flux cooled flow boiling, with an RMS value of 8%. a Decreasin 9 Heat Flux A Repeat of above run Fouling

L , , , lilt , , tOO0 1OOOO 1OOOOO The shape of a typical fouling curve is shown in Fig. 8. Heat w The initial fouling period shows a long delay time, which is known as the initiation period. In this period, nucleation sites for crystal growth are forming. Beyond the delay time, the fouling curve shows a linear increase with time. Figure 5. Heat transfer coefficient as a function of the heat Note that there is a transition region between the initia- flux for (a) 15% black liquor and (b) 60% black liquor. tion and linear periods.

Effect of Liquor Solids Concentration Most chemical physical properties of the solution, both as a result of the pulping mills try to maintain their final liquor solids concentration above 65% to achieve high thermal efficiency increased wall superheat. and low emission of odor in the subsequent recovery boilers. However, it is not uncommon for a mill that is Effect of Solids Concentration The solids concentration experiencing serious fouling problems to be producing has a major effect on the physical properties of the liquor with lower concentration. The effects of concentration on fouling rates are strong. Figure 9 shows a compari- solution. Physical properties of Kraft black liquor can be son of the fouling for three concentrations. The difference found in Ref. 8. For the present investigations, the Prandtl between the fouling rates of the 65% liquor and the 60% number varied between 2 and 500, the major contribution liquor is approximately two orders of magnitude. No foul- being the variation of the viscosity. All measured convec- ing was observed for the 55% liquor. The delay time also tive heat transfer coefficients are shown in Figure 6 as is strongly affected by the liquor solids concentration. It is Nu/Pr 1/3 versus Reynolds number. The fact that all data possible that an excursion into rapid fouling may ulti- can be approximated by a single line indicates the validity mately happen even for the 55% liquor, even though this of a Prandtl exponent of 0.33. To eliminate the effect of has not been observed in experiments lasting as long as 1 wall superheat on the measured heat transfer coefficients, week. the heat flux for these experiments was adjusted such that the value of ~b//Zw was close to 2. Effect of Surface Temperature and Flow Velocity Experi- The measured subcooled boiling heat transfer coeffi- ments have been performed on a 65% black liquor for cients decreased with increasing solids concentration for various surface temperatures and flow velocities. Velocity constant flow velocity and heat flux. effects are shown in Fig. 10 for constant surface and bulk Fouling in Kraft Black Liquor Evaporators 429

400

2 < Pr b < 250

#b//~w=* 2 100

10 Uquor Concentration 0 0% • 15% [] 30 % • 45% tx 60%

i I I I I I I I I I I , , IIII I f t I I III 5 10 100 1000 10000 Figure 6. Nu/Pr 1/3 as a function of Reynolds Number Reynolds number. temperatures. For all velocities, the linear fouling rate Results shown in Fig. 11 illustrate that induction (ini- remained essentially constant. This indicates that the de- tiation) period and fouling rate depend strongly on the position rate is independent of the flow velocity. This surface temperature. The fouling rate increases and the result is characteristic for chemical reaction-controlled delay time decreases with increasing surface temperature. fouling. The delay time is seen to be a function of the It should be noted that convective heat transfer occurred velocity, showing a maximum for a velocity of about 50 for the experiments at low surface temperatures; at high cm/s. It is unclear how the velocity affects the nucleation surface temperatures, nucleate boiling prevailed. The ef- kinetics and, without information on the process of nucle- fect of surface temperature on the delay time may be ation site formation, the experiments do not permit con- expressed by the following correlation: clusions to be drawn on this. However, delay times in industrial applications are normally observed during the -- = k d exp , (5) initial commissioning of new heat transfer equipment. td ~ R'Ts After this period, the surface characteristics generally k d = 2.444 X 1020 min -1, (6) change, and the delay time is either significantly reduced or no longer observed. Eaet = 1.647 × 105 J/mol. (7)

w 20,000 m2K t-- R.M.S. Error = 8 % I

°--03

°--o 10,000 o O

t.-

1- ~ Uquor Concentration 03 "1- o .o~ ~< i~I~< A 15% "o i~ - [] 30% ~-- @ 45% @ =xr~ca "-'- z~ 55% 03 ~'~ • 60%

1,(XX) Figure 7. Measured and predicted 1,000 10,000 W 20,000 subcooled boiling heat transfer co- Predicted Heat Transfer Coefficient m2 K efficients. 430 H. Miiller-Steinhagen and C. A. Branch

2.2 m 2 K ~o V=v o 1450 W/m2K 34. cm s -1 Block liquor kW 2 T b ,= 80=C T s = 108"C 1.8 T.D3 =65~

1.6

1.4 r"

.--4~ 1.2 03 t E I o s 0.6 / 0.4 @

0.2 / o oo °°0,.~,9~,**0,.000~099 .°700°. * ...... , .... ,,.. 25 50 75 100 125 150 175 200 225 250 Figure 8. Typical fouling curve. ~me min

Figure 11 shows no fouling for the run with a surface tions. It is open to speculation whether deposits may start temperature of 87°C. This experiment was stopped after to form in the experiment at 100°C bulk temperature after 24 h. Equations (5)-(7) predict that the run would have a greatly prolonged delay period. had to be continued for 52 h before the first signs of Figure 13 shows a series of experiments obtained for a fouling may have appeared. bulk temperature of 108°C and surface temperatures more than 25 °C higher than the results presented in Figure 12. Effect of Subcooling The effect of bulk temperature is Under these conditions, bubble formation on the heat shown in Figure 12. For a constant surface temperature transfer surface occurs. The fouling curves start with a and velocity, the rate of fouling increases with decreasing constant rate of deposition with no delay time. These bulk temperature, probably as a result of the decreasing fouling curves suggest a different mechanism from that solubility of the scale-forming components. For the three discussed above. For subcooled and saturated boiling, the runs where fouling was observed, the delay time remains number of active bubble nucleation sites is increased with almost constant with bulk temperature, suggesting that increasing wall superheat. Deposition can occur from the delay time depends on surface rather than bulk condi- concentrated regions under a growing bubble, owing to

0.5

mZK Liquor Concentration Black Liquor kW © 55 % v nv= 50 cm/s 0.4 60% tl = 65,000Wire 2 0 65% ¢-0 O~ 0.3 U~ A rr O) .I~- 0.2 A A 0 14.

0.1

Figure 9. Effect of liquor concentration 0 1,000 2,000 3,000 4,000 on fouling rates. Time rain Fouling in Kraft Black Liquor Evaporators 431

2.2 .... i .... i .... i .... , .... , .... , .... , .... g .... j .... rq2 K kW 2 o 1306W/m~ 24 eros-' Blod< tkluor a 1306a=W/m=K 24V,v troll-' I Tb ==.80"C 1.8 *, 1450 W/tln'ZK 34 cm ,,-' 1", = IOW'C * 1665 W/mZK 50 cms-' T~3 =6~P~ 1.6 + 1800 W/m~ 60 cm a-' • 2000 W/m~ 75 ores-I o o 1.4 o

oo / • 1.2 oo • o ° o ~ +o

1 13 ~ ÷e %E ,o O.S 0 0 &4~+

0.6 0 • o + 0.4 0.2 ..- :oo°°i: • o o ~ oo 0 ...... ~v-...... o ..~ -= e~,~ ~~,~ ~~.~.;-,~.~ ~ ~. ~.; ...... 25 5O 75 IO0 125 150 175 2O0 225 25O Time min Figure lO. Effect of flow velocity on fouling rates. the mechanism of microlayer evaporation [16]. Evidence Appearance of Deposit Deposits obtained from the of the microlayer deposition mechanism was found by fouling experiments have a smooth appearance and are prematurely stopping a high heat flux fouling experiment. strongly bonded to the heating surface. In all cases, the A large number of tiny rings were found on the heat deposits had a thin layer of what appeared to be black transfer surface, the size of these rings being of the same liquor immediately adjacent to the heating surface. For all order as the bubble departure diameter. Obviously, the experiments, the deposit was completely soluble in water, deposition rate from microlayer evaporation must in- thus indicating a carbonate-sulfate type scale. It was crease with the number of active bubble nucleation sites noticed that high heat flux fouling experiments generally on the heat transfer surface. For the low wall or bulk gave highly porous scales. temperature experiments or both, few or no nucleation Scanning electron micrographs obtained from two dif- sites would have been active, and hence the microlayer ferent experiments are shown in Figs. 14 and 15. The deposition mechanism would not have been significant. micrographs show the structure of the deposit on the heat This is the reason for the delay times required for crystal transfer surface. Figure 14 shows a uniform crystalline nucleation. In contrast, the large number of active nucle- structure with the growth of the deposit from nucleation ation sites for the high wall and bulk temperature experi- sites. Figure 15 shows a deposit with a much less defined ments leads to remarkable initial deposition rates. crystal structure. Such variations in the morphology are a

5 • , - .., .... , .... , .... , .... , .... , ....

kW 4.=, o 8"/'C Block Uquor = 1600 W/m=K 96"C Tb - eO'C = 1665W/m~ IOa'C Vav = 50 cm a-' o 1785W/m=K 112"C Tn.s = 65 Z * 18~0 W/m~ ~O'C 3.5 • "E~ W/m=K L~O'C

2.5 O'l

2 ,o

1.5 e o & oo # o ' A 0.5 &

0 ..... v .... 0 150 300 4,50 600 750 900 1050 1200 1350 1500 Figure 11. Effect of surface temperature on foul- Time rain ing rates. 432 H. Miiller-Steinhagen and C. A. Branch

25 .... , .... , .... , .... , .... , .... , .... , r'R2 K a o T b Block Uquor kW o 1600W/rn2K 70"C - 1465 W/m2K 80°C T, = 108"C A 1540W/mZK 90"C == Vov ,34 OTI S -! * 1265 W/mlK 100"C T.D.S =65~

A ,= A o C N 1.5 m / A. 2 c # : .e , A

0.5

0 " [ ...... i ...... i .... ~ ...... , ._t ...... Figure 12. Effect of bulk temperature on fouling 0 50 too 15o 2oo 25o ~ 350 4oo ,~o 500 rates. Tune min

although the sample did not show it clearly. Elemental analyses of the deposits from Figs. 14 and 15 show that result of the different types of deposition for each sample. the deposits contain mainly sodium and sulfur. The deposit in Fig. 14 was formed with a long delay time, Diffraction experiments were performed by using a which perhaps explains the distinctive crystal growth. In Philips X-ray diffractometer system. This system consists contrast, the deposit in Fig. 15 was formed by very rapid deposition with no noticeable delay time. It is thought that of a PW 1729 X-ray generator, a PW 1050 vertical go- the rapid deposition rate may not allow the crystal struc- niometer, a PW 1771 diffractometer, a 1133 sample spin- ture to develop to the extent of the Fig. 14 deposit. ner (120 rpm), a PW 1752 curved graphite monocrometer, a xenon sealed proportional detector, a PW 1710 diffrac- Analyses EDAX analysis of black liquor provides its tometer control unit, and a PM 8203A online recorder. major elemental composition. A sample of 65% black The wavelength of the X-ray source was 1.791 × 10 10 m. liquor was oven dried for 24 h at 105°C. The major From all the deposits analyzed, the following compounds elements identified are sodium, potassium, sulfur, and were identified: Na2CO3, Na2S • H20, Na2S2, silicon. A large peak near the silicon and sulfur positions Na6CO3(804) 2 (Burkeite), K2Ca(CO3)2, and CaCO 3. This could indicate aluminum. Calcium also may be present, analysis shows that the deposits consist mostly of Burkeite

0.25 m2.K Black Liquor kW T.D.S = 65 % 0.2 Vav-= ~4 cm/s Tb--lo ¢-O

0.15

0.1 [~ ~1 = 43,000 W/m 2 Ts = 131.7 °(3 0 O Cl = 80,000 W/m 2 Ts = 135.8°(3 LL ~1 = 100000W/m2 Ts= 140"2°(3 0.05

D D

I Figure 13. Fouling resistance versus ol [] t I I I time for low subcooling with varying 1O0 200 300 400 500 surface temperature. Time mi. Fouling in Kraft Black Liquor Evaporators 433

Figure 14. Scanning electron micrograph of a deposit. Figure 15. Scanning electron micrograph of a deposit.

and calcium carbonate with quantities of the liquor itself. The fact that the liquor is present in the deposit is not surprising, because most deposits were to some extent porous and the liquor was generally found in the pores.

Modeling of Black Liquor Fouling Modeling of the fouling phenomena requires the following assumptions about the fouling process: 1. Only the deposition rate is modeled at this stage. A correlation for the delay time was presented earlier in the section on the effect of surface temperature and flow velocity. 2. On the basis of the above results, it can be assumed that the deposition process is controlled by chemical reaction at the heat transfer surface. 3. There is a positive concentration driving force between the bulk and surface. 4. The model is developed for constant heat flux boundary conditions on the heat transfer surface. With the first three assumptions, the following equation can be written:

am = kr(c b - Cs )n. (8) at Equation (8) is a general order rate equation for a chemical reaction. The surface reaction rate constant k r is assumed to obey the Arrhenius equation: 434 H. Miiller-Steinhagen and C. A. Branch combined with a pseudosteady-state simulation model [4, With the use of Eq. (18), optimal operating times can be 5] for the performance of the multiple effect evaporators obtained. For assumed operating conditions of tclea. = 120 at the New Zealand Forest Products Kinleith mill. min, c~ = 2000 W/(m 2 K), T~ = 135°C and T b = 120°C, Equations (11) and (17) can be used to evaluate the Eqs. (16) and (18) predict an optimum operating time of effects of process parameters on fouling in an industrial 220 min, which is close to plant practice. Most modern evaporator. In most industries, it is important to maintain black liquor evaporators have a parallel setup in the the product solids concentration. Under fouling condi- effects where fouling occurs. For these effects, when one tions, this is usually done by increasing the steam pres- evaporator is down for cleaning, the other is still online. sure. For this case, a constant heat flux is maintained and Hence a continuous switching between evaporators is Eq, (11) applies. From Eq. (11), the fouling is minimized occurring. For the optimal case discussed above, the paral- by maintaining a low surface temperature and a small lel effects would have to be switched twice per 8-h opera- temperature difference. The velocity will have little effect. tor shift. The initiation of a cleaning cycle is dependent on the available steam pressure and the tolerable drop in solids REDUCTION OF FOULING concentration. Some evaporators are already operating at the maxi- Chemical Treatment to Reduce Scale Formation mum steam temperature. For these situations, the fouling continually reduces the surface temperature and Eq. (17) There are a number of chemical treatments for fouling in applies. Although the reaction rate is independent of black liquor systems, most being specific to calcium car- velocity, an increase in velocity for convective heat trans- bonate scaling. Two proprietary chemicals were tested to fer will increase the heat transfer coefficient, which will find their performance in preventing fouling from New reduce the overall rate of fouling. However, for nucleate Zealand Forest Products' Kinleith black liquor. The re- boiling conditions, the heat transfer coefficient is indepen- suits cannot be extrapolated to other liquors. dent of velocity and as such there is no velocity effect. The two chemicals investigated were coded $62-2 and Again, low surface temperatures and small temperature $89-2. Recommended dosage for once-through processes differences minimize the fouling rate. For evaporators is 25 ppm. Based on the recommended dosage, a 65% operating with a fixed steam temperature, the fouling will solids liquor that caused considerable fouling was tested strongly affect the product solids concentration. After a with each of the chemical additives. minimum acceptable concentration has been reached, a Figure 16 shows a comparison of fouling from untreated cleaning cycle must be initiated. and $62-2 treated liquor. It was found that the recom- Epstein [18] analyzed scaling governed by Eq. (17). The mended dosage only marginally reduced the fouling; the analysis optimizes the maximum daily production operat- severe characteristic fouling rate was still observed in the ing cycle, assuming a fixed cleaning time, tclean. The result final stages. Large improvements were noted for the 100- of his analyses is given in Eq. (18) for the optimum ppm and 200-ppm runs. Severe fouling was delayed until evaporator time as a function of the down time for clean- nearly seven times the untreated fouling time. In the ing. 200-ppm treatment, no fouling escalation was observed for the 52-h duration of the run. From these results, it can be opt ~ 2/clean concluded that the $62-2 chemical treatment may be used evp = tclean + ot [°~Rfl (18) to reduce the fouling potential of Kinleith Kraft black liquor, it if is still economical at five to ten times the at J,=o initially recommended dosage.

m2K 1.4 B~k Uq.or $62-2 Conc~raUon kW T:D.S = 85 % E] 0 ppm 1.2 Vav= 30 cm/s 0 25 ppm tl = 75,000 W/m 2 z~ 50ppm Tb= 100°C • 100 pprn J • 200 ppm (~ 0.8 n- ~ 0.6 LI.. 0.4 oo° 0.2

Figure 16. Fouling resistance versus OJ time for varying $62-2 concentra- 0 5OO 1,000 1,500 2,000 2,500 3,000 tions. Time mln Fouling in Kraft Black Liquor Evaporators 435

Figure 17 shows the effect of chemical treatment for CONCLUSIONS the same 65% solids black liquor with various $89-2 concentrations. The chemical treatment with $89-2 began Measurements have been presented for forced convective with the recommended dosage of 25 ppm. Severe fouling and subcooled flow boiling heat transfer to Kraft black was then observed earlier than in the untreated case. The liquor with dissolved solids concentrations between 15% same trend was observed for further increases in concen- and 65%. All experimental data can be predicted with tration. The results clearly show that $89-2 greatly aggra- good accuracy by using well-known correlations. vates fouling. Water-soluble deposits from Kraft black liquor are very sensitive to the liquor solids concentration. For high concentration liquors (> 65%), the rate of deposition is rapid. Deposition rates increase with increasing surface temper- Surface Coating with Teflon ature and decreasing bulk temperature. The fouling pro- Attempts were made to reduce the stickiness of the de- cess is insensitive to variations in velocity, at least over the posit by coating the heat transfer surface with PTFE. observed range of flow rates. The heat transfer mecha- Figure 18 shows a strong sawtooth character. The time nism (convection versus nucleate boiling) considerably until severe fouling occurred was significantly prolonged. affects the mechanism of fouling. The investigated de- When the fouling has grown to a resistance of about 0.15 posits are porous and contain fibers. The rate of deposi- m 2 K/kW, the deposit sloughs. The average period of tion affects the crystal morphology, with high rates of removal is estimated at 15 min. Note that, beyond 300 deposition resulting in less well defined crystal structure. min, the sloughing diminishes. This suggests that the The modeling of the deposition process with a chemical reaction fouling model allows discussion of the effects of a continual removal of deposit may wear down the Teflon number of process parameters on the fouling rates. Under coating. A practical Teflon coating must have better dura- both constant heat flux and constant sensible heat transfer bility than the coatings used in this investigation. modes, fouling is reduced by minimizing the surface temperature and the surface-to-bulk temperature difference. The fouling model can be used to estimate optimal clean- Plate and Frame Heat Exchanger ing cycles of black liquor evaporators. Chemical treatment of high concentration liquors can Plate heat exchangers provide a large wall shear stress reduce their fouling potential. However, it is advisible to and high heat transfer coefficients. This may reduce depo- perform preliminary laboratory experiments under compa- sition by increasing deposit removal rates and reducing rable conditions, to prevent adverse effects. Surface coat- surface temperature. A number of fouling experiments ing with Teflon reduces the overall fouling rate. Plate and were performed with an a-Laval PO1 plate heat ex- frame heat exchangers are less prone to fouling from changer for a 68% solids black liquor, which rapidly Kraft black liquor than are tubular heaters. fouled the annular test section for comparable heat fluxes [19]. Each experiment lasted 24 h. For the investigated velocities (0.65 m/s and 1.0 m/s) and surface tempera- The authors are indebted to PAPRO NZ and to New Zealand Forest tures (114.5°C and 128.5°C), no fouling was observed. Products Ltd. for continuous support of the investigations.

m2K 1.4 Black Uquor $89-2 Concentration kW T.D.S = 65 % [] 0 ppm 1.2 Vav-- 30 ern/s 0 25 ppm q = 75,000 W/m 2 A 50 ppm • 100 ppm 1 Tb= 100°C

0-- Figure 17. Fouling resistance ver- 0 50 100 150 200 250 sus time for varying $89-2 concen- Time rain trations. 436 H. Mi]ller-Steinhagen and C. A. Branch

m2K 1,4 Black Uquor kW © [] T.D.S = 65 % 1.2 Vav= 30 cm/s O q = 30,000 W/m 2 e.-O Tb= IO0°C t~ Nffl 0.8 0 with Teflon rr [] [] without Teflon e- 0.6

0 0 IJ_ 0,4 []

Figure 18. Effect of Teflon coating 0 100 200 300 400 500 on fouling. Time rain

NOMENCLATURE Ad deposit thermal conductivity, W/(m K) a heat transfer surface area, m 2 Pd deposit density, kg/m 3 /x dynamic viscosity, Pa s C1,2 constants v kinematic viscosity, m2/s c b concentration in bulk, kg/m 3 Cp specific heat capacity, J/(kg K) Other c s concentration at surface, kg/m 3 ( ) function of d diameter, m e exponent in Eqs. (11) and (14) J j-factor REFERENCES activation energy, J/mol Eact 1. Frederick, W. J., and Grace, T. M., Preventing Calcium Carbon- K Arrhenius constant, (kg/m 2 min)/(kg/m3) e ate Scaling in Black Liquor Evaporators. South. Pulp Paper kc fouling rate constant, kg(m 3 K) Manuf., August, 22-24, September, 21-29, 1979. kd reaction rate constant, rain 1 2. Frederick, W. J., and Grace, T. M., Scaling in Alkaline Spend kr reaction rate constant, (kg/m 2 min)/(kg/m3) e Pulping Liquor Evaporators. In Fouling of Heat Transfer Equip- m deposit mass per area, kg/m 2 ment, E. F. C. Somerscales and J. G. Knudsen, Eds, pp. 587-601, Hemisphere, Washington, DC, 1981. Nu Nusselt number (= ad/A) 3. Grace, T. M., Solubility Limits in Black Liquor Evaporators. Pulp NUlam Nusselt number for laminar flow Paper 41(13), 42-43, 1967. NUturb Nusselt number for turbulent flow 4. Bremford, D., and Miiller-Steinhagen, H. M., Multiple Evapora- /'/ exponent in Eq. (8) tor Performance for Black Liquor I: Simulation of Steady State Pr Prandtl number (= /XCp/A) Operation for Different Evaporator Arrangements. APPITA J. q. heat flux, W/m 2 47(4), 320-326, 1994. 5. Bremford, D., and Miiller-Steinhagen, H. M., Multiple Effect Q heat flow rate, W Evaporator Performance for Black Liquor II: Simulation of Dy- R universal gas constant, 8.314 J/(mol K) namic Operation and Methods of Increasing the Final Liquor Re Reynolds number (= vd/t,) Concentration. APPITA 49(5), 337-346, 1996. Rf fouling resistance, m 2 K/kW 6. Wenzel, U., Hartmuth, B., and Mi~ller-Steinhagen, H. M., Heat Th bulk temperature, °C Transfer to Ternary Mixtures of Acetone, Isopropanol, and Wa- deposit--fluid interface temperature, °C ter under Subcooled Flow Boiling Conditions I: Experimental L Results. Int. J. Heat Mass Transfer 37(2), 175 183, 1994. heat transfer wall temperature, °C Tw 7. Branch, C. A., Fouling During the Evaporation of Kraft Black t time, min Liquor. PhD Thesis, Univ. Auckland, Auckland, New Zealand, /clean cleaning time, min 1992. to delay time, min 8. Branch, C. A., and Mi~ller-Steinhagen, H. M., Physical Properties Uav average flow velocity, m/s of Kraft Black Liquor. APPITA J. 44(5), 339-341, 1991. 9. Shah, R. K., and London, A. L., Laminar Flow: Forced Convec- Greek Symbols tion in Ducts. pp. 284-319, Academic Press, New York, 1978. heat transfer coefficient, W/(m 2 K) 10. Gnielinski, V., W~cirmefibertragungin Rohren, VDI-Wiirmeatlas. 5th Ol0 clean heat transfer coefficient, W/(m 2 K) ed., VDI-Verlag, Diisseldorf, 1986. Fouling in Kraft Black Liquor Evaporators 437

11. Dittus, F. N., and Boelter, L. M. K., Heat Transfer in Automobile 16. Jamialahmadi, M., Bl6chl, R., and Miiller-Steinhagen, H., Bubble Radiators. Vol. 2, p. 443, Univ. California Press, 1930. Dynamics and Scale Formation During Boiling of Aqueous CaSO4 12. Petukhov, B. S., Heat Transfer and Friction in Turbulent Pipe Solutions. Chem. Eng. Process 24, 15-26, 1989. Flow with Variable Physical Properties. In Advances in Heat 17. McCabe, W. L., and Robinson, C. S., Evaporator Scale Forma- Transfer, J. P. Hartnett and T. F. Irvine, Eds., pp. 504-564, tion in Tubular Heat Exchanger. Ind. Eng. Chem. 16(5), 478-479, Academic Press, New York, 1970. 1924. 13. Miiller-Steinhagen, H., and Jamialahmadi, M., Subcooled Flow 18. Epstein, N., Optimum Evaporator Cycles with Scale Formation. Boiling Heat Transfer to Solutions and Mixtures. Engineering Can. J. Chem. Eng. 57, 659-661, 1979. Foundation Conf. on Flow Boiling, Banff, Canada, 1995. 19. Branch, C. A., Miiller-Steinhagen, H., and Seyfried, F., Heat 14. D. Gorenflo, Behiiltersieden. I/'DI-W'drmeatlas, Sect. Ha, 4th ed. Transfer to Kraft Black Liquor in Plate Heat Exchangers. AP- VDI-Verlag, Diisseldorf, 1984. PITA J. 44(4), 270-272, 1991. 15. J. C. Chen, Correlation for Boiling Heat Transfer to Saturated Fluids in Convective Flow. Ind. Eng. Chem. Process Design Dev. 5, 322-329, 1966. Received April 17, 1996; revised August 15, 1996