Conversion of paper sludge to ethanol. I: Impact of feeding frequency and mixing energy characterization

Transcription

1 Bioprocess Biosyst Eng (2007) 30:27 34 DOI /s y ORIGINAL PAPER Conversion of paper sludge to ethanol. I: Impact of feeding frequency and mixing energy characterization Zhiliang Fan Æ Lee R. Lynd Received: 26 September 2006 / Accepted: 29 September 2006 / Published online: 9 November 2006 Ó Springer-Verlag 2006 Abstract In this paper, conversion of paper sludge to ethanol was investigated with the objective of optimization of the overall operation costs. Experimental work was undertaken to optimize cellulase loading, and to determine mixing energy requirements. It was found that decreasing feeding frequency (feed additions per residence time) allows the cellulase loading to be decreased at least two fold with no decrease in cellulose conversion but also entails mixing a slurry of higher solids content and lower conversion at the beginning of the operating cycle. The viscosity of paper sludge slurries was found to increase exponentially with decreasing conversion and increasing solid content. In particular, the viscosity (V) was described well by equation V =e (kx X 0 )(S S 0 )+C (V viscosity (cp), X conversion, S solid content (g/l), k, X 0, S 0, C are empirical parameters). Added costs associated with operating at low feeding frequencies (including higher mixing energy and higher capital costs for the motor and for sludge hold tasks) were found to be small Z. Fan Department of Biological Systems Engineering, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, USA L. R. Lynd (&) Chemical and Biochemical Engineering Program, Thayer School of Engineering, Dartmouth College, Hanover, NH 03755, USA Lee.Lynd@Dartmouth.edu L. R. Lynd Biological Science, Dartmouth College, Hanover, NH 03755, USA compared to the economic benefits resulting from reduced cellulase loading. Introduction Paper sludge is a solid by-product of pulping and paper-making operations that is currently disposed of primarily in landfills in the US [1 4]. It is also an attractive biomass feedstock for production of fermentation products such as ethanol [5 11]. Compared to other cellulosic feedstocks, the composition of paper sludge is not typical (carbohydrate composition varies from 20 to 70%) and the scale of its availability is relatively small (around three million dry tons per year) [1, 11]. However, paper sludge also offers distinct advantages as a bioprocessing feedstock, including potential for negative feedstock cost, no requirement for pretreatment to be made amenable to enzymatic hydrolysis, and integration of processes into a preexisting industrial infrastructure at the mill. Because of these features, paper sludge is very attractive as a point-of-entry and proving ground for emergent industrial processes featuring enzymatic hydrolysis [10, 11]. Conversion of paper sludge to ethanol via simultaneous saccharification and fermentation (SSF) has been studied in batch systems at low ethanol concentrations [5 7, 9, 10]. In our previous study, we developed and subsequently modified a unique solids-fed semi-continuous bioreactor system, and reported data for ethanol production from paper sludge using that system using industrial media and a residence time of

2 28 Bioprocess Biosyst Eng (2007) 30: days. Economically recoverable ethanol concentrations (greater than 40 g/l) were produced [11]. For processing of cellulosic biomass featuring dedicated cellulase production, the cost of cellulase production or purchase has long been seen as a major obstacle to cost competitiveness [12 16]. Although recent advances have lowered the cost of cellulose production, it is still a significant component of the cost of cellulose production at cents per gallon (Novozyme press release, April 2005). In addition to efforts to lower the cost of cellulase production, lowered cellulase usage via process optimization is also an important objective. On the other hand, mixing power is a further part of the overall operating cost associated with processing cellulosic biomass, in addition to cellulase production. Paper sludges are particularly difficult to mix and mixing power consumption is sensitive to both the solids content and extent of conversion [11]. In light of these considerations, it is appropriate to consider impacts on mixing energy requirements in the course of optimization to reduce cellulase loading. In this study, experimental work was undertaken to examine the relationship between feeding frequency and cellulase loading, and also to characterize mixing energy requirements for paper sludge in relation to concentration and conversion. We conclude with an examination of cost tradeoffs involving reduced cellulase loading and increased mixing requirements, making use of an economic model developed in a companion paper. Materials and methods Sources of sludge, yeast, and cellulase Paper sludge samples A and B were provided by the Fraser Paper Mill in Gorham, NH, USA. These are both primary clarifier sludges, obtained at different times. Saccharomyces cerevisiae D5A (provided by NREL) was used for simultaneous saccharification and fermentation (SSF). Cellulase (BRC ) was obtained from Iogen (Ottawa, Ontario) and b-glucosidase (Novozyme 188) was from Novo Nordisk (Wilton, CT, USA). Batch and semi-continuous reactor experiments Batch experiments were carried out in 250 ml serum bottles as described by Lynd et al. [10]. Paper sludge was characterized of cellulose composition and insoluble lignin composition using a method described by Fan et al. [11] and was added to a concentration of 20 g/l cellulose. The enzyme loading used was either 10.5 FPU/g cellulose or 5 FPU/g cellulose as indicated with b-glucosidase supplementation at 60 IU/g cellulose. A simple growth medium consisting of corn steep liquor and MgSO 4 was used, with composition and preparation as described previously [10, 17]. Semi-continuous SSF experiments were carried out using sludge A or sludge B in a reactor system described previously [11]. The reactor system was operated with different feeding frequencies (residence time/feeding interval time) as indicated in the text while keeping the residence time constant at 4 days. Other inputs and operating conditions were as described previously [11]. Effluents were analyzed for ethanol concentration, soluble glucan, soluble xylan, solid content, residual glucan, residual xylan, acid insoluble mineral, and acid insoluble volatile lignin content. Material balances for the inputs and outputs to the reactor were conducted as described previously [11]. Measurement of viscosity and density of paper sludge Sludge sample A as received from the paper mill (about 25% solids) was reduced to relatively uniformly sized slurry using a type F203 Kurup coffee grinder (Peoria, IL, USA), the particle size is about 8 ± 2 mm. Samples with varying solid concentrations were prepared by sludge sample A with different amounts of water. Partially hydrolyzed samples were prepared by adding 60 ml cellulase (66 FPU/mL) and 340 ml of water to 700 g of wet sludge (26.5% solids) and incubating for 1 day at 37 C. The resulting slurry was added to different amounts of water to prepare samples with different solids concentrations. The viscosity was measured using a Brookfield RVDV Pro II viscosity meter in a 600 ml beaker (Diameter cm) under different shear rates. Shear rate c (S 1 ) was calculated according to the manual provided by the supplier using Eq. 1. c ¼ 2xR2 C R2 b x 2 ðr 2 C R2 b Þ x angular velocity of spindle (rad/s) = [(2p/60) N] (N = rpm) R C radius of container (cm) = cm R b radius of spindle (cm) x radius at which shear rate was measured. ð1þ

3 Bioprocess Biosyst Eng (2007) 30: The aggregate densities of paper sludge slurries were determined by weighing slurries in a 200 ml gradual cylinder and calculate the density by dividing the weight of slurry by the volume. Estimation of density and viscosity as a function of solids concentration and conversion The average shear rate in an agitated vessel is estimated using Eq. 2 [18] c ¼ kn ¼ kðn=60þ k shear rate constant (r 1 ) n the rotating speed (rad/s) N rpm. ð2þ The shear rate constant is dependent on the type and shape of impeller. We used a k value of 22 r 1 for pitched paddles [19]. We based viscosity measurements and subsequent determination of mixing energy requirements on an impeller speed 60 rpm (n = 1) used in the 1 L lab reactor. This speed has already provided sufficient mixing in lab scale reactor, and impellers with larger size while rotating at the same speed should provide more than enough turbulence to meet the mixing needs at full scale. The viscosity at an average shear rate of 22 s 1 (n = 1) was estimated at different solid concentrations and extents of enzymatic hydrolysis. An empirical equation, obtained by fitting the viscosity data with respect to the solid concentration and conversion was used to estimate the viscosity at any time of the cycle. The density of paper sludge was estimated using an empirical equation, obtained by fitting the density data with respect to the solid concentration, with the solid concentration at any time of the cycle as the input. Estimation of viscosity and density during the SSF cycle Viscosity and density during the reaction cycle were estimated in order to evaluate mixing energy requirements. In the absence of a valid rate law for paper sludge hydrolysis, a first order decline of the fractional cellulose conversion X was assumed, for which XðtÞ ¼X s exp ðln X s ln X e Þt ð3þ T T cycle time (h) X s X e conversion at the beginning of the cycle conversion at the end of the cycle. Solids concentration at any time during the cycle is linearly related to conversion, and was described using the following equation SðtÞ ¼S s þ S e S s X e X s ðxðtþ X s Þ S s S e X s X e ð4þ solids concentration at the beginning of the cycle (g/l) solids concentration at the end of the cycle (g/l) conversion at the beginning of the cycle (g/l) conversion at the end of the cycle (g/l). Estimation of mixing power consumption during SSF The power delivered to the liquid in an industrial reactor described in the accompanying paper was calculated from the following equation: P ¼ N p n 3 D 5 a q N p D a ð5þ power number Diameter of the impeller (D a = m, sized using Aspen Icarus Process Evaluator). Values for power numbers, N p, were found using a correlation of power number versus Reynolds number for a non-baffled marine propeller (pitch = 1.5) [18] with viscosity and density at each time of the reaction cycle as input to calculate Reynolds number. The average mixing energy requirement over the operating cycle was calculated by integrating the power consumption over time and then dividing by the cycle time. Estimation of the economic impact of lowering feeding frequency The economic impact of lowering feeding frequency on an ethanol production facility is evaluated by estimating the extra capital cost associated with running the reactor at lower feeding frequency (higher costs for bigger sludge holding tanks and bigger motors), and then determining the payback time of this additional capital costs as a result of

4 30 Bioprocess Biosyst Eng (2007) 30:27 34 reduced operating expenses (saving in cellulase cost minus the additional electricity expense) using the economic model described in a companion paper. Sludge holding tanks were costed as live bottom storage bins using the Aspen Icarus Process Evaluator. Agitators were costed using Aspen Icarus based on the maximal viscosity and density in the reactor operating cycles. The cost of cellulase can be estimated to be in the range of $ per million IU (corresponding to cellulase costing cents per gallon ethanol produced for a cellulase loading of about 15 IU/g cellulose based on Genencor news release April 2004). In this study, we used a cellulose cost of $0.17 per gallon ethanol produced for a cellulase loading of 10 FPU/g, corresponding to a cellulase cost of $2.4 per million IU. An electricity price of 0.5 cents/kwh, representative of the particular location analyzed in an accompanying paper, and a more representative 5 cents/kwh were used to evaluate electricity costs. Results SSF experiments Composition of sludges A and B Compositional analysis based on a dry weight basis and the solid contents of paper sludge A and B is presented in Table 1. Batch SSF results Table 1 Sludge composition Sludge A (%) Sludge B (%) Solid content Composition based on dry weight Glucan Xylan Mannan Acid insoluble mineral Acid insoluble volatile Acid soluble mineral Total Ethanol Concentration (g/l) Sludge A (10 FPU/ g cellulose) Sludge A (5 FPU/ g cellulose) Sludge B (10 FPU/ g cellulose) Sludge B (5 FPU/ g cellulose) Reaction Time (hr) Fig. 1 Batch SSF results Batch SSF experiments were carried out over a period of 96 h, and ethanol concentration was monitored as a function of time (Fig. 1). Ethanol was produced from Sludge B faster than from Sludge A at the same enzyme loading. In particular, the rate of ethanol production for sludge A at a cellulase loading of 10.5 FPU/ g was similar to that of sludge B at 5 FPU/g cellulose. Semi-continuous SSF at different feeding frequencies Semi-continuous SSF experiments were conducted using sludges A and B at different feeding frequencies. Figure 2 presents data obtained with sludge B with cellulase loading of 10.5 FPU/g cellulose. After a transient period of about 12 days, steady state was achieved with a mean ethanol concentration of 45.2 g/ L and cellulose conversion of 95.8 wt%. Steady-state material balances for this run are presented in Table 2 for the period from day 18 to day 26. Mass recovery of 90.4% was measured for xylan, and the ethanol yield was 0.49 g ethanol/g glucose equivalent utilized. Several similar runs were carried out to determine the relationship between cellulose conversion and feeding frequency. As shown in Fig. 3, the experimental results indicate that cellulose conversion can be Concentration Fig. 2 SSF of sludge B Time (hr) xylose(g/l) volume(ml) (right axis) cellulose out(g/l) ethanol(g/l) solid out(g/l)

5 Bioprocess Biosyst Eng (2007) 30: Table 2 Steady state material balance for conversion of sludge B to ethanol Flow rate (g/day) Cellulose in ± 1.21 Cellulose out 0.57 ± 0.06 Conversion 96.8% Xylan in 3.54 ± 0.24 Xylose out 1.44 ± 0.09 Soluble xylan out 1.82 ± 0.13 Xylan in solid out 0.11 ± 0.01 Recovery 90.4% Soluble glucan 0.09 ± 0.02 Glucose brought by enzyme 0.40 ± 0.00 Mannan 0.51 ± 0.03 Ethanol 9.12 ± 0.47 Yield 49.4% Conversion 100% 95% 90% 85% 80% 75% Sludge A: 10.5 FPU/ g cellulose Sludge B: 10 FPU/ g cellulose Sludge B: 5 FPU/ g cellulose Conversion, recovery or yield Feeding Frequency Fig. 3 Conversion versus the feeding frequency Sludge A: 20 FPU/ g cellulose increased by decreasing the feeding frequency (residence time/feeding interval) at a fixed cellulase loading. A glucan conversion of 93.4% was achieved at 10 FPU/g cellulose when the feeding frequency was 1.33 and residence time was 4 days, while conversion of 92% [11] required 20 FPU/g cellulose loading when the feeding frequency was 8 with the same type of sludge. In another set of experiments carried out with sludge B, the conversion improved from 79 to 92% using 5 FPU/g cellulose loading and 60 IU/g cellulose b-glucosidase loading when the feeding frequency was reduced from 8 to 1.33 and the residence time was held constant at 4 days. The average cellulose conversion and solid concentration at the beginning and ending of the cycle at feeding frequencies of 1.33, 2 and 4 for sludge A, at frequency 8 reported for the same sludge reported in a previous study [11] are shown in Table 3. Physical properties of paper sludge The density and viscosity of paper sludge as a function of extent of conversion and solid content were characterized in order to evaluate mixing energy requirements. The density of sludge/water mixtures as a function of solid concentration is shown in Fig. 4 for three different sludge conversions. The density of the paper sludge and water mixture increased with increasing solids concentration, but did not differ much with respect to the extent of enzymatic hydrolysis. Thus, it is adequate for our purposes to treat the density of the sludge/water mixture as a function of solid concentration only. Linear regression incorporating this assumption yields: q ¼ 1; 000 þ 0:3026 S q S ð6þ density of the sludge and water mixture (g/l) solid concentration (g/l). The viscosities of the unreacted, 31% reacted, and completely reacted sludges are shown in Fig. 3. The viscosities of unreacted sludge, 31% converted sludge, and completely reacted sludge at various solids concentrations decreased with increasing shear rate. The sludge and water mixture thus exhibited the behavior of a pseudoplastic non-newtonian fluid regardless of conversion and solid content (Fig. 5). Viscosities at the average shear rate of 22 s 1 were investigated as a function of solids concentration and fractional conversion. As presented in Fig. 6, viscosity increased sharply with increasing solids concentration for sludge with X = 0 and X = By contrast, the increase of viscosity from sludge that had been completely hydrolyzed (X = 1) was slight. It was found that the viscosity of sludge as a function of solid concentration and cellulose conversion was described by the following equation: V ¼ e ðkx 0 X 0 ÞðS S 0 ÞþC V X S viscosity (cp) conversion solid concentration (g/l). ð7þ Values of the parameters k, X 0, S 0, and C were obtained by fitting the viscosity data at different X and

6 32 Bioprocess Biosyst Eng (2007) 30:27 34 Table 3 Initial and final solid concentration and conversion at different feeding frequencies % converted f Starting S (g/l) Ending S (g/l) Starting X (%) Ending X (%) % converted 0% converted Viscosity (cp) Density (g/m3) % converted 31% converted 0% converted Solid Concentration (g/l) Fig. 6 Experimental and calculated viscosity based on Eq. 7 Viscosity data agrees well with values calculated using equation (Fig. 6) Solid Concentration (g/l) Fig. 4 Density of paper sludge slurries versus solid concentration at different conversions Viscosity (cp) Fig. 5 Viscosity versus shear rate S values to Eq. 7 using the least squares method, yielding k ¼ 0:02756; X 0 ¼ 0:04669; S 0 ¼ 8:2853; C ¼ 3:32343 unreacted sludge (4% solids) 31% converted sludge (3.5% solids) completely converted sludge (4.3% solids) Shear Rate (S -1 ) Mixing energy comparison at different feeding frequencies In a companion paper, we present a process design for an ethanol plant processing 15 dry tons of paper sludge per day. The SSF fermenter in that plant has a volume of 450 m 3 with a liquid volume of 400 m 3, and two-side mounted marine propellers (D a = m, and pitch = 1.5). We evaluate the mixing energy requirement of this fermenter if it was operated at different feeding frequencies using conversion and solids concentration data presented in Table 3. As shown in Fig. 7, decreasing the feeding frequency from 8 to 1.33 leads to about a 6% increase in mixing energy requirements. Average Power Consumption (kw/m3) Feeding Frequency Fig. 7 Average power consumption per unit volume versus feeding frequency

7 Bioprocess Biosyst Eng (2007) 30: Table 4 Economic impact of decreasing feeding frequency f = 1.33 f = 8 Difference Annual operating cost Cellulase cost $70,778 $141,556 $70,778 Electricity cost at kwh 1 (electricity cost at 0.05 kwh 1 ) $1,613($16,130) $1,521($15,130) $92($820) Total $70,686( $69,958) Capital cost Agitator $114,017 $110,100 $3,917 Sludge holding tank $181,680 $112,800 $68,880 Total $295,697 $222,900 $72,797 Economic impacts of lowering feeding frequency As described in the companion paper, a sludge holding tank for a feeding frequency of 1.33 needs to hold 260 ton of moist sludge; agitators are 30 KW side mounted marine impellers. If the process is operated at feeding frequency 8, a sludge holding tank needs the capacity to hold 50 ton of sludge, agitators for the SSF fermenter operated were 25 KW. Table 4 presents a comparison of cellulase expense, electricity, and yearly capital expenses associated with decreasing the feeding frequencies from 8 to It may be seen that, the savings from reduced cellulase costs (about $70,778 year 1 ) far outweigh the expense associated with increased electricity (electricity at either 0.5 or 5 cents/kwh) and added capital cost. Discussion Cellulase loading and bioreactor feeding strategy (batch, fully continuous, and intermittent) are both widely recognized as important design variables impacting the cost and performance of processing cellulosic biomass. However it is much less widely recognized that these variables are related, and in particular that required cellulase loading to achieve a given conversion significantly impacted by feeding strategy. The results presented herein provide substantial support for this proposition. Using a single simultaneous saccharification and fermentation system fed with paper sludge at a 4-day nominal residence time, decreasing feeding frequency is accompanied by increased conversion at constant enzyme loading, and by decreased required enzyme loading at constant conversion. The mechanistic basis for these trends is addressed in a paper in preparation. From an economic point of view, changing the feeding frequency entails a trade-off. As feeding frequency decreases, the cost of cellulase purchase gets lower, but higher cost are incurred for added capital associated with sludge holding tanks and larger agitator as well as electricity to drive agitators. Our analysis, including evaluation of properties of paper sludge relevant to mixing, indicates that cellulase cost savings and the sludge holding tanks are by far the largest of these terms. We find there is a strong economic incentive to go to lower feeding frequencies. In particular, savings associated with reducing the feeding frequency from 8 to 1.33 pay back the added capital cost in about 1 year. This result is rather insensitive to the cost of electricity. Since paper sludge is more difficult to mix than most cellulosic substrates, we suspect that the incentive to operate at low feeding frequencies shown here for paper sludge is likely applicable to other feedstocks as well. Acknowledgment This work was supported by grants from the Consortium for Plant Biotechnology and from the National Institute of Standards. References 1. Agenda A technology vision and research agenda for America s forest, wood, and paper industry. American Forest & Paper Association (1994) 2. Assessment of costs and benefits of flexible and alternative fuel use in the US. Part II evaluation of a wood to ethanol process. Transportation Sector, Technology Report. US Department of energy DOE/EP-000 (1993) 3. Chemical analysis and testing laboratory analytical procedures. National Renewable Energy Laboratory, Golden (1995) 4. Solid waste management and disposal practices in the US paper industry. Technical Bulletin No. 793, NCASI, New York (1999) 5. Duff SJB, Moritz JW, Anderson KL (1994) Simultaneous hydrolysis and fermentation of pulp mill primary clarifier sludge. Can J Chem Eng 72: Duff SJB, Moritz JW, Casavant TE (1995) Effect of surfactant and particle size reduction on hydrolysis of deinking sludge and nonrecyclable new print. Biotechnol Bioeng 45: Jeffries TW, Schartman R (1999) Bioconversion of secondary fiber fines to ethanol using counter-current enzymatic saccharification and co-fermentation. Appl Biochem Biotechnol 77 79:

8 34 Bioprocess Biosyst Eng (2007) 30: Katzen R, Fowler DE (1994) Ethanol from lignocellulosic wastes with utilization of recombinant bacteria. Appl Biochem Biotechnol 45/46: Lark N, Xia Y, Qin CG, Gong CS, Tsao GT (1997) Production of ethanol from recycled paper sludge using cellulase and yeast, Kluveromyces Marxianus. Biomass Bioenergy 12: Lynd LR, Lyford K, South CR, van Walsum P, Levenson K (2001) Evaluation of paper sludge for amenability to enzymatic hydrolysis and conversion to ethanol. TAPPI J 84: Fan Z, South C, Lyford K, Munsie J, van Walsum P, Lynd LR (2003) Conversion of paper sludge to ethanol in a semicontinuous solids-fed reactor. Bioprocess Biosyst Eng 26: Vonsivers M, Zacchi G (1995) A technoeconomic comparison of 3 processes for the production of ethanol from pine. Biores Technol 51: Gregg DJ, Saddler JN (1997) A review of techno-economic modeling methodology for a wood-to-ethanol process. Appl Biochem Biotechnol 63 65: Sheehan J, Himmel M (1999) Enzymes, energy, and the environment: a strategic perspective on the US Department of Energy s Research and Development Activities for Bioethanol. Biotechnol Prog 15: Lynd LR, Cushman JH, Nichols RJ, Wyman CE (1991) Fuel ethanol from cellulosic biomass. Science 251: Himmel ME, Ruth MF, Wyman CE (1999) Cellulase for commodity products from cellulosic biomass. Curr Opin Biotech 10: Kadam KL, Newman MM (1997) Development of a low-cost fermentation medium for ethanol production from biomass. Appl Microbiol Biotechnol 47: Perry RH, Green DW (1997) Perry s chemical engineers handbook, 7th edn. McGraw-Hill 19. Lai KP, Steffe JF, Ng PKW (2000) Average shear rates in the rapid visco analyser (RVA) mixing system. Cereal Chem 77: