Influence of the Membrane Cut-off During Ultrafiltration of Kraft Black Liquor with Ceramic Membranes

Influence of the Membrane Cut-off During Ultrafiltration of Kraft Black Liquor with Ceramic Membranes

0263–8762/03/$23.50+0.00 # Institution of Chemical Engineers Trans IChemE, Vol 81, Part A, November 2003 www.ingentaselect.com=titles=02638762.htm I...

481KB Sizes 3 Downloads 55 Views

0263–8762/03/$23.50+0.00 # Institution of Chemical Engineers Trans IChemE, Vol 81, Part A, November 2003

www.ingentaselect.com=titles=02638762.htm

INFLUENCE OF THE MEMBRANE CUT-OFF DURING ULTRAFILTRATION OF KRAFT BLACK LIQUOR WITH CERAMIC MEMBRANES ¨ NSSON O. WALLBERG and A.-S. JO Department of Chemical Engineering, Lund University, Lund, Sweden

F

ractionation of kraft black liquor by ultraŽ ltration is receiving a great deal of interest at the moment, mainly because the extracted lignin can be used as a biofuel, but also because there are indications that the pulp quality is improved when permeate is used instead of black liquor as the diluent in the cooking process. Fractionation of kraft black liquor calls for membranes that can withstand high temperatures at high pH. In this investigation the performance of two ceramic membranes manufactured by Orelis, France, with cut-offs of 5 and 15 kDa was studied. The temperature during the experiments was 90 C and no adjustment of the pH was made. The retention of cooking chemicals was almost zero for both membranes. The volume of black liquor was reduced by 80% with the 5 kDa membrane and by 90% with the 15 kDa membrane. The average  ux during concentration with the 5 kDa membrane was 45 l m 2 h 1 (transmembrane pressure 400 kPa and circulation velocity 3.6 m s 1) while the  ux of the 15 kDa membrane was 95 l m 2 h 1 (transmembrane pressure 100 kPa and circulation velocity 4.5 m s 1). The lignin recovery was 66% and 28% for the 5 and 15 kDa membranes, respectively. Keywords: kraft black liquor; lignin; ultraŽltration; ceramic membrane; energy savings.

INTRODUCTION A modern kraft pulp mill has an energy surplus in the form of bark and the lignin present in the kraft black liquor. In integrated pulp and paper mills most of the energy surplus can be utilized, but in mills where the recovery boiler places limits on the capacity, the extraction of lignin from the black liquor makes it possible to increase the capacity. In a modern pulp mill 25–50% of the lignin can be extracted without disturbing the operation of the mill. However, in principle, all lignin in the black liquor can be extracted without disturbing the operation of the recovery boiler. If large amounts of lignin are exported the fuel shortage can be covered by burning cutting loss (that is, branches, roots and tree tops) in the bark boiler. This alternative is especially interesting for pulp mills where the recovery boiler constitutes a bottleneck as the capital cost of a bark furnace is lower. Generally, precipitation is used to extract lignin from kraft black liquor. Precipitation by acidiŽ cation using carbon dioxide, sulphuric acid, and waste acid from chlorine dioxide generation has been studied (Ale´n et al., 1979; 1985; Uloth and Wearing, 1989a, b; LoutŽ et al., 1991; ¨ hman and Theliander, 2001). A Davy et al., 1998; O comparison between lignin removal by precipitation and ultraŽ ltration performed by PAPRICAN in the late 1980s indicated a lower cost for precipitation (Uloth and Wearing, 1989b). However, during recent years temperature- and

pH-resistant ceramic membranes have drastically altered the conditions possible for the ultraŽ ltration process. In this work the in uence of membrane cut-off was investigated. Two ceramic membranes with cut-offs of 5 and 15 kDa were included in the investigation. The 15 kDa membrane was used in an earlier investigation where the in uence of temperature on membrane performance during ultraŽ ltration of kraft black liquor was studied (Wallberg et al., 2003). The experiment in this work was performed at 90¯ C and no adjustment of the pH of the liquor was made. The in uence of transmembrane pressure (TMP) and volume reduction (VR) on the  ux and lignin recovery was studied. MATERIAL AND METHODS Black Liquor Kraft black liquor is the residual liquor from kraft pulping. In the kraft process wood chips are treated at high temperature with liquor (white liquor), which contains the active cooking chemicals, OH¡ and HS¡, as well as the counterions sodium and, to a lesser extent, potassium. During the digestion process the cellulose Ž bres are liberated from the polymer, lignin, which binds the Ž bres together. The purpose of the cooking chemicals is to break down the lignin polymer into smaller fragments that are soluble in the cooking liquor. The kraft black liquor

1379

¨ NSSON WALLBERG and JO

1380

therefore contains the residues of the cooking chemicals and their counterions and lignin dissolved from the wood chips. It also contains multivalent nonprocess elements that have entered the mill with the wood raw material, extractives (fatty acids, resin acids and neutrals) and hydroxy acids. The kraft black liquor used in this investigation was supplied by So¨dra Cell Va¨ro¨ Pulp Mill, Sweden. The Va¨ro¨ Mill is located on the west coast of Sweden. The mill has an annual capacity of 350,000 metric tons of bleached softwood pulp. The pulp is produced from spruce and pine and bleached without using chlorine. The Kappa number, which is a measure of the lignin content of the pulp, is 27–30 after the cook. The black liquor used in these experiments was withdrawn before the evaporator unit. The pH was 13– 14. The total dry substance (TDS) of the liquor was 16 wt%, the lignin content was about 56 g=l and the content of inorganic material was 37 g=l. The inorganic material consists mainly of cooking chemicals (sodium and sulphur). The cooking chemicals make up more than 90% of the inorganic material. Nonprocess elements (NPE), such as manganese, magnesium and calcium, have their origin in the wood and are not part of the cooking chemicals. Analysis The total dry matter content was determined by drying weighed samples at 105¯ C and measuring the weight of the residue. The organic material was measured by heating the residue from the TDS measurement to 950¯ C, and weighing the sample afterwards. The content of organic material is the difference between the weight of residue before and after heating to 950¯ C. The ash content is the difference between the TDS and the organic material. The ash contentis of interest as the concentrated solution should be used as fuel. The ash content after the combustion is therefore of importance. Lignin contains phenolic groups. The lignin concentration can therefore be measured as the light absorption at a wavelength of 280 nm (Hill and Fricke, 1984; Hill et al., 1988). The UV light absorption was measured using a Shimadzu UV-160 spectrophotometer. Before measurement of the light absorption, the samples were diluted with deionized water. The absorption constant used was 24.6 g l¡1 cm¡1 (Fengel et al., 1981). The concentration of inorganic elements was determined by inductively coupled plasma atomic emission spectroscopy (ICP AES) with a Perkin Elmer Optima 3000DV ICP AES instrument. The samples were pretreated in order to remove the organic content, which would otherwise have interfered with the analysis. The pretreatment involved dilution of 10 ml of the sample with 5 ml concentrated nitric acid. The sample was heated under pressure in a microwave oven for 30 minutes. During this heating process carbon is oxidized, which increases the pressure inside the vessel containing the sample. This excess carbon dioxide is released through a valve and the sample is heated again for 2 hours to remove the remaining organic material. The Ž nal solution is then diluted to 50 ml and analysed. Equipment Two KERASEP membranes produced by Orelis, France, with cut-offs of 5 kDa and 15 kDa were used in the experiments. During the experiments the membranes were housed

in a K01 module (Orelis). The 5 and 15 kDa membranes had 19 and 7 parallel  ow channels, respectively. Data for the two membranes are given in Table 1. A centrifugal pump (NB32=25-20, ABS Pump Production AB, Sweden) regulated with a frequency converter (LUST CDA3000, Lust Antriebstechnik GmbH, Germany) provided the desired cross- ow velocity and feed pressure. The feed pressure and cross- ow velocity were regulated by the frequency converter and a valve on the retentate side after the module. The pressure was measured before and after the membrane module and on the permeate side. The transmembrane pressure was regulated by a valve on the permeate side of the module. The transmembrane pressure is the difference between the average pressure on the feed side and the permeate pressure. Two 200 l feed tanks were connected to the plant. One was used as the feed tank during the experiments and the other was used for cleaning. As the temperature of the black liquor was close to the boiling temperature, evaporation from the tank must be kept to a minimum. This was achieved using a sealed lid with only a small hole for a temperature probe. The feed tank was heated with steam, which condensed inside a coil in the feed tank. When the feed solution had reached the desired temperature this was maintained with a regulated electrical heater inside the tank. Samples of the feed solution were withdrawn through a valve at the bottom of the feed tank. Flux was measured with a balance. The analogue signals were transformed with an AD converter (INTAB AAC-2) before they were transferred to the computer. The data were recorded and analysed with a LABVIEW 6 program. Performance of Experiments The experiments were performed in two consecutive steps. First, the in uence of the TMP was studied at constant concentration (both retentate and permeate were recirculated to the feed tank). The experiment was then continued by bleeding off permeate to a Ž nal volume reduction (VR) of 0.8 for the 5 kDa membrane and 0.9 for the 15 kDa membrane. The volume reduction is the ratio between the permeate volume and the initial feed volume. Initially, the pure water  ux was measured at 50¯ C and 50 kPa. The black liquor (initial feed volume 170 l), which had been preheated in the feed tank to the temperature of the pure water, was circulated in the plant, during heating to 90¯ C. During the heating period the permeate valve was closed. When the solution had reached 90¯ C the permeate valve was opened and the transmembrane pressure was adjusted to 20 kPa (15 kDa membrane) and 50 kPa (5 kDa membrane). The  ux was measured after 30 minutes. The pressure was then increased stepwise to 400 kPa for the 5 kDa membrane and to 200 kPa for the 15 kDa membrane. The TMP was kept constant for 30 minutes at each pressure level. The Ž rst 10 l of permeate, containing mainly water Table 1. Data for the commercial ceramic membranes made of Al2O3-TiO2 (Orelis, France), used in the investigation. Cut-off (kDa) Diameter of each  ow channel (mm) Number of channels Membrane area (m2)

5 3.5 19 0.245

15 6 7 0.155

Trans IChemE, Vol 81, Part A, November 2003

INFLUENCE OF MEMBRANE CUT-OFF

from the pure water  ux measurement, was withdrawn. Thereafter, both retentate and permeate were recirculated to the feed tank. Samples of the feed and the permeate were withdrawn at the different pressure levels. The TMP was then reduced to 200 kPa (5 kDa) and 100 kPa (15 kDa) and the black liquor was recirculated overnight for about 16 hours. The next day permeate was continuously withdrawn until the VR reached 0.6 for the 5 kDa membrane and 0.67 for the 15 kDa membrane. Then the permeate was recirculated overnight (for approximately 16 hours). Concentration was then continued until the volume reduction was 0.8 for the 5 kDa membrane and 0.9 for the 15 kDa membrane. During concentration the TMP was 400 kPa for the 5 kDa membrane and 100 kPa for the 15 kDa membrane. Samples of the permeate and retentate were withdrawn for analysis at regular intervals. The temperature was 90¯ C in all experiments. During the experiments with the 5 kDa membrane the cross- ow velocity was 3.6 m s¡1 and in the experiments with the 15 kDa membrane 4.5 m s¡1. At these velocities the pressure drop due to friction in the  ow channels was about the same, 90 kPa, in both membranes. Because of the emphasis on preserving the same pressure drop in both membranes the Reynolds number in the experiments with the 15 kDa membrane was 2.2 times that in the 5 kDa membrane experiments. The experiments were completed by rinsing the membrane for 1–2 hours with permeate collected during concentration. The temperature and TMP were 90¯ C and 50 kPa. The rinsing solution (the collected permeate) was then replaced by an alkaline cleaning agent, 0.25 wt% Ultrasil 11 (Henkel). Retentate and permeate were withdrawn until these process streams were no longer coloured. Retentate and permeate were then recycled at 50 kPa until the temperature of the cleaning solution had fallen from 90¯ C to 50¯ C. This took about 90 minutes. When the temperature of the cleaning solution had reached 50¯ C the dirty cleaning solution was replaced with new cleaning solution (0.25 wt% Ultrasil 11). Retentate and permeate were recycled for at least 2 hours at 50¯ C and 50 kPa. Finally, the membrane was thoroughly rinsed with deionized water at 50¯ C and 50 kPa. The pure water  ux was measured in order to ensure that the membrane was clean. If the initial pure water  ux was not restored, the cleaning procedure was repeated. RESULTS In uence of TMP on Membrane Performance The difference in permeability of the two membranes included in the study was considerable. The pure water  ux of the 5 kDa membrane was 35 l m¡2 h¡1 and that of the 15 kDa membrane 210 l m¡2 h¡1, at 50 kPa and 50¯ C. This resulted in a signiŽ cantly higher  ux of kraft black liquor for the 15 kDa membrane than for the denser membrane, as can be seen in Figure 1. At 200 kPa the  ux was about 25 and 220 l m¡2 h¡1 for the 5 and 15 kDa membranes, respectively. The lignin retention of the 5 kDa membrane was markedly higher, as can be seen in Figure 2(a). The lignin retention of the 5 kDa membrane was 63% and that of the 15 kDa membrane 33%. The retention of cooking chemicals (which constitute the main part of the inorganic material) was almost the same for the two membranes, whereas the Trans IChemE, Vol 81, Part A, November 2003

1381

Figure 1. The in uence of transmembrane pressure on the  ux of kraft black liquor during ultraŽ ltration with ceramic membranes with cut-offs of 5 and 15 kDa. The temperature was 90 C and the cross- ow velocity 3.6 and 4.5 m s 1 in the 5 and 15 kDa membranes, respectively.

retention of the multivalent metal ions is signiŽ cantly higher than that of the monovalent Na and K ions, as can be seen in Figure 2(b). The retention of multivalent ions was higher for the 5 kDa membrane, but the difference in retention was not especially marked. The reason for the high retention of multivalent ions is probably that colloids are formed, which are retained by both membranes. In a previous investigation the retention of lignin during ultraŽ ltration with three polymeric membranes was studied (Wallberg et al., 2003). These experiments were performed with black liquor from the same mill as in this work, but at a lower temperature, 60¯ C. The retention of lignin by the three polymeric membranes with cut-offs of 4, 8, and 20 kDa was 80, 67, and 45%, respectively. The retention of lignin by the polymeric membranes was thus higher, possibly because of the lower temperature during these experiments. A decrease in the lignin retention when the temperature is increased was shown in an earlier investigation with the ceramic 15 kDa membrane (Wallberg et al., 2003). Concentration The black liquor was concentrated to a VR of 0.8 with the 5 kDa membrane and 0.9 with the 15 kDa membrane. The  ux of both membranes decreased as the volume reduction increased, but the  ux of the 15 kDa membrane was markedly higher during the entire concentration process, as can be seen in Figure 3. In this study the in uence of VR on  ux is described by a polynomal correlation: J ˆ a ‡ (b ¢ VR) ‡ (c ¢ VR2 ) ‡ (d ¢ VR3 )

(1)

where a, b, c, and d are polynomial coefŽ cients. The average  ux during batch concentration is obtained by integration of equation (1). The average  ux, Jav, is then: „ VR J dVR Jav ˆ 0 VR ³ ´ ± ´ ² ³d b c ˆ a‡ ¢ VR ‡ ¢ VR2 ‡ ¢ VR3 (2) 2 3 4

1382

¨ NSSON WALLBERG and JO

If the retention is constant during the process the correlation between concentration and volume reduction follows the relationship (Cheryan, 1998): ³ ´R 1 C ˆ C0 ¢ (3) 1 ¡ VR

where C0 is the concentration at the start of the batch concentration and R the retention. The retention is obtained by regression analysis of the experimental data. Using this method the retention of lignin during concentration was found to be 77% and 45% for the 5 and 15 kDa membranes, respectively. The lignin concentration during volume reduction of kraft black liquor is shown in Figure 4. The lignin concentration increased from about 56 g l¡1 to 185 and 158 g l¡1 with the 5 and 15 kDa membranes, respectively. Although the volume reduction was lower in the experiment with the 5 kDa membrane, the concentration of lignin in the remaining concentrate was higher using this membrane. Concentrations of some substances at the beginning and at the end of the concentration process can be found in Table 2. The higher concentration of Ca, Mg, and Mn in the concentrate of the 15 kDa membrane is due to the higher volume reduction in this experiment. Recovery of Lignin and Cooking Chemicals The amount of material retained by the membrane depends on the retention of the membrane and the volume reduction. The amount retained, or the recovery, is given by: Figure 2. Retention during ultraŽ ltration of kraft black liquor at 200 kPa and 90 C. (a) Retention of lignin, total dry substance (TDS), and inorganic material. (b) Retention of inorganic substances. The corresponding  ux values can be found in Figure 1.

The average  uxes were calculated using the data in Figure 3. The average  ux was 95 l m¡2 h¡1 when concentrating to a VR of 0.9 with the 15 kDa membrane and 45 l m¡2 h¡1 when concentrating to a VR of 0.8 with the 5 kDa membrane.

Recovery ˆ

mr C ˆ (1 ¡ VR) ¢ r m0 C0

where m denotes mass and C concentration. Subscripts r and 0 denote the retentate and the initial feed. Some of the substances present in the original black liquor recovered in the retentate can be found in Figure 5. The higher recovery of the 5 kDa membrane is due both to the higher retention of this membrane and to the lower volume reduction in this experiment. When the retention is independent of the concentration, the percentage of a compound in the concentrated solution can be calculated from the correlation: Recovery ˆ (1 ¡ VR)1¡R

Figure 3. Flux during concentration of kraft black liquor at 90 C. The cross- ow velocity and the TMP of the 15 kDa membrane were 4.5 m s 1 and 100 kPa. The corresponding values for the 5 kDa membrane were 3.6 m s 1 and 400 kPa. The symbols are experimental values and the lines are obtained by regression analysis of the data using a third-order polynomial [equation (1)].

(4)

(5)

The experiment with the 5 kDa membrane was interrupted at VR ˆ 0.8 because of the  ux decline as the concentration increased. At VR ˆ 0.8 the recovery of lignin was 66%, as can be seen in Figure 5. If the volume reduction had been continued to VR 0.9 with constant retention, R ˆ 77%, the recovery would have been 59%, according to equation (5). The recovery of lignin of the 15 kDa membrane at VR 0.9 was 28%. If the retention of the cooking chemicals is zero the amount of cooking chemicals removed from the lignin concentrate is a linear function of the volume reduction, according to equation (5). In this case the recovery should be 20% and 10% at VR ˆ 0.8 and 0.9, respectively. This is in good agreement with the results shown in Figure 5. It can therefore be concluded that the retention of cooking chemicals was almost zero for both membranes. From the point of view of the recovery of cooking chemicals a high volume Trans IChemE, Vol 81, Part A, November 2003

INFLUENCE OF MEMBRANE CUT-OFF

Figure 4. Lignin concentration during concentration of kraft black liquor. The symbols are experimental values and the lines represent values calculated using equation (3). The correlation coefŽ cient of the 5 kDa membrane was 0.995 and that of the 15 kDa membrane 0.991.

reduction is thus desirable. However, a high volume reduction also results in a lower amount of lignin in the concentrated solution. Optimization of both cooking chemical and lignin recovery is thus necessary.

1383

Figure 5. Amount retained in the retentate after concentration of kraft black liquor at 90 C. The volume reduction of the 5 kDa membrane was 0.8 and that of the 15 kDa membrane 0.9. The transmembrane pressures were 400 and 100 kPa for the 5 kDa and 15 kDa membranes, respectively.

volume reduction also affects the  ux of the ultraŽ ltration plant. In Table 3 material balances for the 5 and 15 kDa membranes at various volume reductions can be found. The calculations are applicable to a modern kraft mill producing 2000 tons pulp day¡1, generating 8 m3 black liquor ton

DISCUSSION Cooking liquor is made up of white liquor, containing the main part of the active chemicals, and black liquor, which is used as a diluent to ensure good liquor circulation without introducing extra water. Lignin can be removed from the black liquor returned to the recovery system by ultraŽ ltration and the lignin fraction can be sold as a fuel to replace fossil fuels in other furnaces. The permeate, containing the cooking chemicals, can be returned to the digester, replacing black liquor, as shown in Figure 6. The requirement on a separation process used to fractionate black liquor is that it should separate the cooking chemicals from the lignin. The cooking chemicals represent a substantial value, and must be recovered. High recovery of cooking chemicals also results in levels of inorganic substances low enough to enable the resulting fuel to be burned in any furnace. The Ž nal lignin concentration, as well as the purity of the lignin fraction, depends on the volume reduction. The

Figure 6. Schematic diagram showing the concept of kraft lignin removal by ultraŽ ltration.

Table 2. Composition of feed and retentate at the end of the concentration process during ultraŽ ltration of kraft black liquor at 90 C. The volume reduction of the 5 kDa membrane was 0.8 and that of the 15 kDa membrane 0.9 at the end of the concentration process. 5 kDa membrane

¡1

Lignin, g l TDS, wt% Organic matter, % Na, g l¡1 S, g l¡1 K, g l¡1 Fe, mg l¡1 Ca, mg l¡1 Mg, mg l¡1 Mn, mg l¡1

15 kDa membrane

Original black liquor

Concentrate at VR 0.8

Original black liquor

Concentrate at VR 0.9

56 15 67 27.4 5.4 4.4 3.2 32.6 43.7 12.9

185 26 77 34.5 5.3 4.4 13.8 85.5 188.0 53.1

56 16 67 29.3 5.0 4.0 2.8 36.7 46.2 13.1

158 22 76 30.8 5.2 4.6 12.8 94.0 316.5 76.7

Trans IChemE, Vol 81, Part A, November 2003

¨ NSSON WALLBERG and JO

1384

Table 3. Mass balance for an ultraŽ ltration plant for the removal of lignin from kraft black liquor (KBL). The pulp production is 2000 tons day 1 generating 8 m3 ton 1 of black liquor (667 m3 h 1). The concentrations of lignin, sodium, and sulphur in the black liquor are 55, 30, and 5 g l 1, respectively. In total, 25% of the lignin in the black liquor is removed. Membrane cut-off 5 kDa Volume reduction Retention of lignin (%) Concentration of lignin (g l¡1) Average  ux (l m¡2 h¡1) KBL treated in the UF plant (m3 h¡1) Permeate  ow (m3 h¡1) Membrane area (m2)

0.8 77

liquor is fractionated by ultraŽ ltration the lignin recovery is about 60% for a 5 kDa membrane and 30% for a 15 kDa membrane at VR ˆ 0.9. The retention of cooking chemicals is almost zero for both membranes. The retention of nonprocess elements (multivalent ions) is high for both membranes (>40%).

15 kDa 0.9 77

0.8 45

0.9 45

190

324

114

155

44

40

103

96

241

283

404

591

193

255

323

532

4386

6375

3136

5542

KBL, kraft black liquor.

pulp¡1. The concentration of lignin in the black liquor is 55 g l¡1. The amount of lignin removed is 25% of the lignin originally in the black liquor, which means that 220 metric tons lignin is removed per day. Equation (3) was used to calculate the concentration of lignin in the concentrate, and the average  ux was calculated with equation (2). The need for extra membrane area due to cleaning of the membranes was not considered in these calculations. The membrane area needed for an ultraŽ ltration plant is smaller at lower VR, but the loss of cooking chemicals is higher. The cooking chemicals can be recovered and returned to the digester=recovery system by diaŽ ltration of the lignin concentrate and by precipitation and washing of the lignin. Precipitation and Ž ltration can also be used to increase the dry substance to levels needed in an ordinary furnace (50–80% TDS). The return of the liquor and washing Ž ltrates to the liquor recycling system would increase the load on the evaporators by 5.8%. The impact of the lignin recovery process on the chemical liquor cycle and mass balance of the mill will be examined in future work.

REFERENCES Ale´n, R., Patja, P. and Sjo¨stro¨m, E., 1979, Carbon dioxide precipitation of lignin from kraft black liquor, Tappi J, 62(11): 108–110. Ale´n, R., Sjo¨stro¨m, E. and Vaskikari, P., 1985, Carbon dioxide precipitation of lignin from alkaline pulping liquors, Cellulose Chem and Technol, 19(5): 537–541. Cheryan, M., 1998, UltraŽ ltration and MicroŽ ltration Handbook (Technomic Publishing Co, Inc., Lancaster, Pennsylvania, USA). Davy, M.F., Uloth, V.C. and Cloutier. J.-N., 1998, Economic evaluation of black liquor treatment for incremental kraft pulp production, Pulp Paper Can, 92(2): 35–39. Fengel, D., Wegener, G. and Feck, J., 1981, Beitrag zur Charakterisierung analytischer und technischer Lignine, Holzforschung, 35(3): 111–118. Hill, M. and Fricke, A.L., 1984, UltraŽ ltration studies on kraft black liquor, Tappi J, 67(6): 100–103. Hill, M.K., Violette, D.A. and Woerner, D.L., 1988, Lowering black liquor viscosity by ultraŽ ltration, Sep Sci Technol, 23(1&2): 1789–1798. LoutŽ , H., Blackwell, B. and Uloth, V., 1991, Lignin recovery from kraft black liquor: preliminary process design, Tappi J, 74(1): 203–210. ¨ hman, F. and Theliander, H., Filtration properties of lignin precipitated O from black liquor, International Chemical Recovery Conference 2001, Whistler, June 11–14 2001. Uloth, V.C. and Wearing, J.T., 1989a, Kraft lignin recovery: acid precipitation versus ultraŽ ltration. Part I. Laboratory test results, Pulp Paper Can, 90(9): 67–71. Uloth, V.C. and Wearing, J.T., 1989b, Kraft lignin recovery: acid precipitation versus ultraŽ ltration. Part II. Technology and economics, Pulp Paper Can, 90(10): 34–37. Wallberg, O., Jo¨nsson, A.-S. and Wimmerstedt, R., 2003, Fractionation and concentration of kraft black liquor lignin with ultraŽ ltration, Desalination, 154: 187–199. Wallberg, O., Jo¨nsson, A.-S. and Wimmerstedt, R., 2003, UltraŽ ltration of kraft black liquor with a ceramic membrane, Desalination, 156: 145–153.

ACKNOWLEDGEMENTS This work is part of an R&D programme titled ‘Future resource efŽ cent pulp mill — FRAM’. We are very grateful to STEM (the Swedish Energy Agency), MISTRA (the Foundation for Strategic Environmental Research), [ ngpannefo¨reningens Foundation for Research and Development, Aga A Linde, Anox, Borregaard LignoTech, Ba¨ckhammars bruk, Fortum Va¨rme, Holmen Paper, Orelis, Stora Enso, Sydkraft, and So¨dra Cell for their support and their commitment to the programme.

CONCLUSIONS The average  ux during concentration of kraft black liquor is 45 (VR ˆ 0.8) and 95 l m¡2 h¡1 (VR ˆ 0.9) for a 5 and 15 kDa ceramic membrane. However, the membrane area of an ultraŽ ltration plant for removing lignin from black liquor will be only about 15% larger if the plant is equipped with the denser membrane because of the higher lignin retention of the 5 kDa membrane. When kraft black

ADDRESS Correspondence concerning this paper should be addressed to Prof. A.-S. Jo¨nsson, Department of Chemical Engineering, Lund University, PO Box 124, SE-221 00 Lund, Sweden. E-mail: ann-soŽ [email protected] The manuscript was received 20 May 2003 and accepted for publication after revision 8 August 2003.

Trans IChemE, Vol 81, Part A, November 2003