THERMAL PYROLYSIS CHARACTERISTICS OF POLYMER
FLOCCULATED WASTE ACTIVATED SLUDGE
C. P. CHU1, D. J. LEE1* and C. Y. CHANG2
1Department of Chemical Engineering, National Taiwan University, Taipei, Taiwan, 10617 and 2Graduate Institute of Environmental Engineering, National Taiwan University, Taipei, Taiwan, 10617
(First received 29 November 1999; accepted in revised form 22 March 2000)
AbstractÐPolyelectrolyte conditioning is a common practice in wastewater management. This paper experimentally elucidated the thermal pyrolysis characteristics of waste activated sludge at a temperature range of 300±900 K (27±6278C) using thermogravimetric analysis (TGA) in inert atmosphere, with especial attention on the eect of polyelectrolyte ¯occulation (using cationic polyacrylamide). On the pyrolysis rate vs temperature plot two maxima were noted. At the heating rate of 88C/min, polyelectrolyte does not in¯uence the pyrolysis process. As higher heating rates (14 and 208C/min), on the other hand, ¯occulation to charge neutralization point would enhance the rate of thermal pyrolysis. A simple two parallel-reaction kinetic model is applied to interpret the experimental data. Possible roles of ¯occulant on sludge pyrolysis are discussed on the basis of change in sludge structures and the hindrance of surface reactions of sludge particles. 7 2000 Elsevier Science Ltd. All rights reserved
Key wordsÐpyrolysis, conditioning, sludge, model
INTRODUCTION
The current methods for disposal of waste activated sludge include land®ll with stabilization and incin-eration. Sludge pyrolysis becomes a potential alternative to treating waste activated sludge from wastewater treatment plant. The pyrolysis process involved the heating of sludge in an inert atmos-phere, from which part of the organic matters could be released from the sludge and recycled (Suzuki et al., 1988; Campbell and Bridle, 1989), which matches the appeal of resource utilization. In ad-dition, heavy metals (except mercury and cadmium that are going to their salts) could be safely enclosed in the solid residues (Kaminsky and Kum-mer, 1989; Kislter et al., 1987). Some researchers employed a ¯uidized bed to study the pyrolysis of sewage sludge (Kaminsky et al., 1982, 1987; Piskorz et al., 1986, Stammbach et al., 1989).
Thermogravimetric analysis (TGA) is capable of providing pyrolysis kinetic data of sludge at evalu-ated temperatures (Caballero et al., 1995). A few publications experimentally elucidated the pyrolysis kinetics of sewage sludge with the help of a TGA
test (Urban and Antal, 1982; Dumpelmann et al., 1991). However, data interpretation usually faces major diculties for distinguishing the weight vs temperature data into a complex reaction scheme (Boldyreva, 1987). For engineering use, a lump-type kinetic model is commonly adopted to correlated pyrolysis data. Dumpelmann et al. (1991) developed a model for pyrolysis of sewage sludge that could predict the maximum weight loss in a ¯uidized bed. Conesa et al. (1997) proposed a kinetic model for the pyrolysis of anaerobically digested and non-digested sewage sludge. Conesa et al. (1998) utilized this proposed kinetic model to interpret sewage sludge pyrolysis data.
Polyelectrolyte ¯occulation is a common practice in wastewater treatment plant to improve the sludge dewaterability (Wu et al., 1997). The dosage of ¯oc-culant leading to the best available dewaterability is usually referred to as the ``optimal dosage'', whose weight fraction is generally low to the original slurry (ten to hundreds of ppm, say). In a dried ®l-ter cake, on the contrary, the residual polyelectro-lyte amount could reach up to 1% by weight. All previous literature works considered pyrolysis of sludge alone. Information regarding the presence of polyelectrolyte ¯occulation on the pyrolysis of waste activated sludge is still largely lacking in the literature, which is the main theme of this work.
7 2000 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0043-1354/00/$ - see front matter
49 www.elsevier.com/locate/watres
PII: S0043-1354(00)00235-9
*Author to whom all correspondence should be addressed. Tel.: +886-2-2362-5230; fax: +886-2-2362-3040; e-mail: [email protected]
EXPERIMENTAL
Sample
An activated sludge sample was taken on 24 February 1999 from the wastewater treatment plant of Neili Bread Plant, Presidential Enterprise Co., Taoyuan, Taiwan. The true solid density was measured by Accupyc Pycnometer 1330 (Micromeritics) with a relative deviation of less than 0.5%, giving a measurement of 1457 kg/m3. A particular
sizer (LS230, Coulter) determined the particle size distri-bution as a monodispersed distridistri-bution with an average of 110.4 mm. The weight percentage of the dried solid, deter-mined by drying at 1028C, was 1.43%. All tests started within 2 h after sampling.
Cationic polyelectrolyte T-3052, which is obtained from Kai-Guan Inc. Taiwan, was employed as the sludge ¯occu-lant. T-3052 is a cationic polyacrylamide with an average molecular weight of 107and charge density of 2.27 meq/g.
The original sludge was placed in the mixing vessel into which the polymer solution was gradually poured to make doses of 5, 15, 35, 50 and 70 g per kg dried solids (DS), respectively. The stirring was 200 rpm for 5 min followed by 50 rpm for the next 20 min. The zetameter (Zeter-Meter System 3.0, Zeter-(Zeter-Meter Inc., USA) measured the z potentials of ¯occulated sludge ¯ocs. Figure 1 depicts the results. As Fig. 1 reveals, the z potential increased with polymer dose and reached charge neutralization point at around 15±35 g polymer/kg DS. Figure 1 also depicts the corresponding zone settling velocity (ZSV) data of the sludge. The best settling occurs at the dose of charge neu-tralization point, which is referred to as the ``optimal dose'' of sludge. Sludge with a dosage less than 15 g per kg DS is considered underdosed, while exceeding 40 g per kg DS, overdosed.
The ¯occulated sludge was then vacuum-®ltered to remove most of the free water. Then the dewatered samples were dried at 1028C for 24 h. The dried product was crushed into ®ne powders of size 30 mm.
TGA test
Chen et al. (1997) provided the experimental detail for
the TGA test, which was brie¯y summarized herein for the sake of completeness. The thermal analyzer (SETARAM, 77A-92) was employed for recording the thermographs with argon gas (Ar) as the carrying gas. Vacuum-®ltered sludge cake is the testing sample. The cell temperature was ®rst raised from room temperature to 353 K (808C) at a rate of 208C/min, then was kept at 353 K for 1 h to remove most free water. After this stage the cell tempera-ture was raised again to 900 K (6278C) at heating rates of 8 and 148C/min, respectively. The weight±time w±t data represent the TGA curve, whose slopes (dw/dt ) give the derivative thermogravimetric analysis (DTG) curve (Ewing, 1985).
Pore size and surface area measurement
Autopore II 9220 (Micromeretics) measures the pore volume and surface area of the dried sample prior to TGA tests. Mercury was intruded into the sample interior at dierent pressures, from which the incremental pore volume could be estimated. Assuming that the void por-tion of the sample interior consists of cylindrical pores can the pore size and the surface area data be subsequently calculated.
RESULTS AND DISCUSSION
TGA results
Figure 2a±c depicts the a vs T data at three dierent heating rates. At the heating rate of 88C/ min, all data at various polyelectrolyte doses fall into one single curve. At 14 or 208C/min, on the other hand, the curves of doses 15 and 35 g/kg DS shift to the left, indicating a faster pyrolysis. Except for the doses close to charge neutralization, other tests at 14 and 208C/min coincide with the curves at 88C/min. Apparently the sludge pyrolysis rates were enhanced at a high heating rate with a dose to
charge neutralization, which are denoted as the ``speed-up'' cases. Figure 3a±c depicts the results of da/dt vs T curves. Two peaks are observed around 550 K (2778C) and 800 K (5278C), respectively, whose heights increase with heating rate. Probably two distinct reaction mechanisms exist for T is lower or higher than approximately 700 K (4278C). As stated, two gross categories of the organics are proposed to exist in sludge. One could be
decom-posed at 250±3008C, while the other at 550±6008C. Such an observation closely corresponds to the ®nd-ings by Conesa et al. (1998). Furthermore, once the pyrolysis has been speeded up, the second peak diminished.
In sum, we demonstrated that (i) at a low heating rate of 88C/min, the role of cationic polyacrylamide is negligible on pyrolysis rate up to 500±6008C; (ii) at a high heating rate (14 or 208C/min), on the
Fig. 2. The a vs T curves. (a) 88C/min; (b) 148C/min; (c) 208C/min. The dashed curves are model results based on parameters listed in Table 1.
other hand, adding polyacrylamide to charge neu-tralization point would markedly enhance the pyrol-ysis rate. Point (ii) is somewhat surprising since the so-called ``optimal dose'' is generally considered meaningful in an aquatic environment at room tem-peratures, but on the pyrolysis characteristics at high temperatures. We explored this point further in the next section.
Possible roles of polyacrylamide ¯occulation
Figure 4 depicts the pore size data. In compari-son with the results of original sludge reveals that the pore size between 10 and 100 mm increases sig-ni®cantly with the presence of polyacrylamide. For samples of doses 0, 35 and 70 g/kg DS, the intru-sion volumes of mercury were 0.51, 0.61 and 0.66 ml/g. Consequently, the total pore areas are estimated as 22.1, 28.3 and 36.1 m2/g, respectively. As stated, after ¯occulation, the dried and crushed sludge powders would exhibit a greater pore size and surface area. Such an observation may corre-late with the enhancement in thermal decomposition rate since the available area for surface reaction increases at 35 g/kg DS. However, we noted the slow-down of pyrolysis rate at doses exceeding 50 g/ kg DS.
Figure 5 depicts the pyrolysis data of pure poly-acrylamide. The data of original sludge is also shown in the ®gure for the sake of comparison. Notably, the decomposition rate of polyacrylamide is lower than that of original sludge. Polyacryl-amide molecules would stick on sludge surface after
¯occulation. According to the patch model (Wang and Audebert, 1987), surface coverage would reach a value less than 100% at charge neutralization. In overdosing regime, the coverage could approach 100%. As a result, the slow decomposition rate of the polyacrylamide may hinder the decomposition of organic materials from the (covered) sludge par-ticle surface in the overdosing regime. Such an eect may correspond to the slow-down phenomena of thermal decomposition of sludge at dose exceed-ing 35 g/kg DS.
Therefore, polyacrylamide may have two contra-dictory eects to pyrolysis at 14 or 208C/min. With the action of ¯occulation more surface area would be available in the sludge sample that enhances py-rolysis. On the other hand, at a speci®c temperature polyacrylamide would decompose slower than the sludge sample, thereby hindering the thermal pyrol-ysis rate. The compensation of these two factors may cancel with each other that interprets the trends noted in Fig. 2b and c.
However, the above mentioned mechanisms can-not properly interpret the experimental results at 88C/min, that is, polyacrylamide has no eects on thermal pyrolysis at 88C/min (Fig. 2a). It is specu-lated herein that at a low heating rate, the molten liquid generated from the solid surface might retain for a while before evaporation. Surface migration may thereby deteriorate the pore structure. Such a structure change may not occur at a high heating rate. There is, nevertheless, no experimental proof for such a hypothesis.
Kinetic model
Pyrolysis reaction of sludge is signi®cant during 200±6008C. The detailed reaction scheme should be rather complex. Two peaks are noticeable in Fig. 3a±c, except for the cases at the charge-neutral-ized case at heating rates of 14 and 208C/min. A simple kinetic model considering two parallel reac-tions is adopted here for engineering use. The scheme could be stated as follows:
The remaining fraction of activated sludge after py-rolysis represents the inert residual.
TGA data were represented by the dimensionless conversion a de®ned as follows:
Fig. 3. The da/dt vs T curves. (a) 88C/min; (b) 148C/min; (c) 208C/min. The dashed curves are model results based on parameters listed in Table 1.
a wwiÿ w
iÿ wf, 1
where w, wiand wf, are the sample weights at time
t, at 808C, and at the end of test (6008C), respect-ively. Assuming that the pyrolysis kinetic follows an Arrhenius-type expression: da dt A1 exp ÿE1 RT 1 ÿ an1 A 2 exp ÿE2 RT 1 ÿ an2, 2 Fig. 3 (continued)
where Ei is the activation energy, Ai, the
pre-expo-nential factor, ni, the reaction order, and R, the
gas constant.
We herein estimated the kinetic parameters Ai,
Ei, ni i 1, 2 using nonlinear regression at TGA
data. Table 1 lists the ®tting results. The threshold temperature that divides the whole regime into reac-tion 1 and 2 is set at 700 K (4278C).
All schemes of parametric evaluation reveal that the changes in E, A and n are not independent. As
stated, these kinetic parameters tend to compensate each other and leading to so-called ``compensation eects'' (Boldyreva, 1987). Although the kinetic model with dierent sets of ®tting parameters can better describe the apparent pyrolysis behavior, they should be considered as empirical in nature. No information regarding the intrinsic chemical kinetics can be extracted therein.
Despite the ``speed-up'' cases, all ®tting par-ameters exhibit similar values. For example, E1 Fig. 5. Pyrolysis data of pure polyacrylamide and of original sludge.
Table 1. Fitted kinetic parameters for the two parallel-reaction scheme in eq. (2)
Dose (g/kg DS) A1(sÿ1) E1(J/mole) n1 A2(sÿ1) E2(J/mole) n2
(a) Temperature ramp=88C/min
0 454,000 84,800 6.72 385,000 119,000 1.00 5 472,000 85,200 7.66 389,000 125,000 0.80 15 423,000 84,900 7.39 324,000 126,000 0.84 35 712,000 87,700 7.52 407,000 121,000 0.97 50 608,000 86,600 7.73 436,000 123,000 0.89 70 588,000 86,500 7.35 507,000 124,000 0.96 (b) Temperature ramp=148C/min 0 924,000 86,700 7.03 611,000 125,000 0.83 5 963,000 87,600 5.57 21,600 116,000 0.36 15 787,000 86,800 4.63 905 94,700 0.39 35 939,000 88,400 4.62 804 94,200 0.37 50 724,000 86,500 7.48 24,900 113,000 0.49 70 775,000 86,500 7.16 458,000 126,000 0.84 (c) Temperature ramp=208C/min 0 1,130,000 87,500 6.86 216,000 123,000 0.62 5 901,000 86,500 7.30 655,000 129,000 0.67 15 805,000 86,300 4.10 753 87,700 0.57 35 1,200,000 88,700 4.24 542 88,300 0.44 50 1,100,000 87,400 7.95 1,146,000 127,000 0.77 70 1,120,000 87,500 6.62 664,000 125,000 0.90 Optimal dose.
ranges from 85,000 to 91,000 J/mole; n1, 6.7±7.5; E2, around 120,000 J/mole; n2, 0.62±0.96. Conesa et al. (1998) proposed that the products in o-gas stream around 2508C (reaction 1) include methane, carbon dioxide, water, chloromethane and acetic acid; while around 5508C (reaction 2), hydrogen, methane, carbon dioxide, hydrocarbons, alcohols and chloromethane. The present results reveal that the temperature dependence for reaction 1 is weaker than that is for reaction 2 E1< E2, while the
trend in reactant amount dependence reverses n1
n2:
For the speed-up cases, on the other hand, the kinetic parameters for reaction 1 are similar to those of the normal cases except for a higher A1 and a lower n1. Reaction 1 thereby has been enhanced at the speed-up conditions. However, since reaction 2 has been diminished when the speed-up occurs, the kinetic parameters markedly change. For instance, the pre-exponential factor A becomes very low, indicating a low reaction rate. Meanwhile, n2reduces to around 0.4±0.5, including a weak dependence of the pyrolyzed material. Restated, polyelectrolyte ¯occulation to charge neu-tralization point enhances reaction 1 by transform-ing the organics that is originally released at elevated temperature (reaction 2) to the low-tem-perature regime (reaction 1).
Implications to application
Although we cannot conclude the underlying mechanisms that correspond to the observed pyrol-ysis behavior, nonetheless, the polyelectrolyte ¯oc-culation cannot only enhance sludge dewaterability in solution, but also promote the pyrolysis rate in dried solid if the heating rate is high enough. In ad-dition, ovedosing would deteriorate both the sludge dewaterability and the pyrolysis eciency. Such an observation has implication to sludge management practice.
Notably, since both the heat and mass transfer resistances are negligible in a TGA test, the high heating rate corresponds to a far-from-equilibrium environment close to the sample surface. As stated, since the heating rate is rather high, sample would not reach chemical equilibrium with its environment while its (slow) surface reaction controls the pyrol-ysis process. If all the interpretations to the role of polyacrylamide mentioned above are correct, then to mimic a high heating rate environment, sludge pyrolysis may consider high sludge feed rate with recycling. The size of fed material should be small while the retention time should be kept short. One possible alignment is to convey thin sludge ®lter cake into a high-temperature furnace with a high
feeding speed. Part of the product could be recycled back to the feeding point for raising the conversion. In addition, the sludge should be conditioned to charge neutralization prior to dewatering and pyrol-ysis.
REFERENCES
Boldyreva E. V. (1987) Problems of the reliability of kin-etic data evaluated by thermal analysis. Thermochimica Acta 110, 107.
Caballero J. A., Font R., Marcillam A. and Conesa J. A. (1995) New kinetic model for the thermal decomposition of heterogeneous materials. Ind. Eng. Chem. Res. 34, 806.
Campbell H. W. and Bridle T. R. (1989) Conversion of sludge to oil: a novel approach to sludge management. Wat. Sci. Tech. 21, 1467.
Chen G. W., Chang I. L., Hung W. T., Lee S. F. and Lee D. J. (1997) Continuous classi®cation of moisture con-tent in waste activated sludge. J. Envir. Eng. ASCE 123, 253.
Conesa J. A., Marcilla A., Prats D. and Rodriguez-Pastor M. (1997) Waste Manag. Res. 15, 293.
Conesa J. A., Marcilla A., Moral R., Moreno-Caselles J. and Perez-Espinosa A. (1998) Evolution of gases in the primary pyrolysis of dierent sewage sludges. Thermo-chimica Acta 313, 63.
Dumpelmann R., Richarz W. and Stammbach M. R. (1991) Kinetic studies of the pyrolysis of sewage sludge by TGA and comparison with ¯uidized beds. Can. J. Chem. Eng. 69, 953.
Ewing G. W. (1985) Instrumental Methods of Chemical Analysis, 5th ed. McGraw-Hill, New York, pp. 429±432. Kaminsky W., Semel J. and Sinn H. (1982) Orientierende Versuche zur Pyrolyse von Klarschlamm in einer Indir-ekt Beheizeten Wirbelschicht (Rapid communication). Makromol. Chem. 3, 371±375.
Kaminsky W., Augustin T., Bellmann U. and Kruger-Betz M. (1987) Recycling von Klarschlamm, eds K. J. Thome-Kozmiensky and U. Loll, pp. 309±318. EF Verlag, Ber-lin, Germany.
Kaminsky W. and Kummer A. B. (1989) Fluidized bed pyrolysis of digested sewage sludge. J. Anal. Appl. Py-rolysis 16, 27±35.
Kislter R. C., Widmer F. and Brunner P. H. (1987) Beha-vior of chromium, nickel, copper, zinc, cadmium, mer-cury, and lead during the pyrolysis of sewage sludge. Environ. Sci. Technol. 21, 704.
Piskorz J., Scott D. S. and Westerberg I. B. (1986) Flash pyrolysis of sewage sludge. Ind. Eng. Chem. Proc. Dev. 25, 265±270.
Stammbach M. R., Kraaz B., Haggenbucher R. and Richarz W. (1989) Pyrolysis of sewage sludge in a ¯ui-dized bed. Energy Fuels 3, 225±259.
Suzuki A., Nakamura T., Yokoyama S. Y., Ogi T. and Koguchi K. (1988) Conversion of sewage sludge in heavy oil by direct thermochemical liquefaction. J. Chem. Eng. Japan 21, 288.
Urban D. L. and Antal M. J. (1982) Study of the kinetics of sewage sludge pyrolysis using DSC and TGA. Fuel 61, 799±806.
Wu C., Huang C. and Lee D. J. (1997) Eects of polymer dose on alum sludge dewaterability characteristics and physical properties. Colloids Surf. A. 122, 89±96.
