Fuel 82 (2003) 465–472 www.fuelfirst.com
An experimental study of oil recovery from sewage sludge by low-temperature pyrolysis in a fluidised-bedq
Lilly Shen, Dong-Ke Zhang*
School of Chemical Engineering, Curtin University of Technology, GPO Box U1987, Perth, WA 6845, Australia
Abstract
Pyrolysis of activated sewage sludge was investigated under inert conditions in a fluidised-bed to study the effects of temperature and gas residence time on the product distribution and composition with an aim to maximise the oil yield. The temperature was varied from 300 to 600 8C and the gas residence time from 1.5 to 3.5 s. Three groups of products were produced, namely, a non-condensable gas (NCG) phase, a solid phase (char) and a liquid phase (oil). A maximum of 30% oil yield (wt% daf of sludge fed) was achieved at a pyrolysis temperature of 525 8C and a gas residence time of 1.5 s. Higher temperatures and longer gas residence times favoured the formation of NCG, suggesting that secondary cracking reactions had occurred. The oil obtained was analysed using GC – MS and H NMR to determine the oil’s composition and structure, a unit structure of the oil was proposed which consisted of aromatic rings connected by hydrocarbons with –OH functional groups attached. q 2002 Elsevier Science Ltd. All rights reserved.
Accepted 9 September 2002; available online 22 October 2002
Keywords: Fluidised-bed reactor; Sewage sludge devolatilisation; Oil structure
1. Introduction
oil is achieved, lower emissions of NOx and SOx also lower operating costs when compared to incineration [7]. Tests have shown that the oil obtained from sewage sludge pyrolysis can be used directly in diesel fuelled engines and is comparable to low-grade petroleum distillates from commercial refineries [8].
Sewage sludge like most organic wastes is abundant in volatile matter and thus there exists an opportunity to convert this stored energy into a usable energy source such as for heating or perhaps as a liquid fuel substitute and the production of speciality chemicals [1]. Presently, most of the sewage sludge waste is disposed off in landfills, ocean or via incineration [2,3]. The disposal of dewatered sewage sludge in sanitary landfills is not very satisfactory because of the inherent chemical energy lost and associated health problems [4]. Ocean dumping disturbs, at least locally, the ecology of the biosphere and should be avoided [5]. Incineration provides a large volume reduction of sewage sludge and results in improved thermal efficiency. However, the scrubbing costs of the product gases for air pollution control are usually very high [3].
In the past decade, pyrolysis of sewage sludge in an oxygen-free atmosphere at relatively low-temperatures (ca. 500 8C) has generated significant interest [6]. Particular interest is shown in this process as a high recovery of liquid
q Published first on the web via Fuelfirst.com_http://www.fuelfirst.com
* Corresponding author. Tel.: þ 61-8-9266-7581; fax: þ61-8-9266-3554. E-mail address: dkzhang@che.curtin.edu.au (D.K. Zhang).
Many researchers have investigated the effects of operating parameters such as temperature and residence time on product distribution [9,10]. However, many have focused on fixed beds [11] and rotary reactors [9]. Very few have performed low-temperature pyrolysis of sewage sludge in fluidised-bed reactors [12]. In the few studies that have used fluidised-bed reactors, the oil was lumped together without further classification of their composition or structure [13,14]. In this study a fluidised-bed reactor was used to determine the oil recovery and product distribution of sewage sludge under moderate temperatures (300 – 600 8C) and varying gas residence time (1.5 – 3.5 s). A GC –MS was used to investigate the molecular distribution and structure of the oil and H NMR analysis was used to detect the type of functional groups in the oil and verify the GC –MS results. H NMR analysis has been previously used to determine tar composition from coal pyrolysis and has shown to be useful in identifying the type of H-bonds in coal derived oils [15].
0016-2361/03/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 1 6 - 2 3 6 1 ( 0 2 ) 0 0 2 9 4 - 6
L. Shen, D.-K. Zhang / Fuel 82 (2003) 465–472 466
Table 1 Proximate and ultimate analyses of the activated Western Australian sewage sludge
Proximate analysis (wt%) Moisture (as fed) Ash (as fed) Volatile matter (daf) Fixed carbon (daf) 3 22.6 61.3 16.1
Ultimate analysis (wt% daf) C H N S O, Cl (by diff.) 41.8 5.4 4.3 1.14 47.36
a fluidised bed with an internal diameter of 50 mm. The sludge is fed from the sludge hopper by a screw feeder, 40 cm above the distribution plate where nitrogen is used as the fluidising gas. Nitrogen is also fed through the hopper and the nozzle of the screw feeder to assist the flow of the solid. The product gas leaves from the top of the reactor and through a cyclone the solid residual (char) is separated and collected in a char container situated below the cyclone. The product gas is then condensed in a shell-and-tube stainless steel condenser to ambient temperature using tap water as the cooling medium. The condensed oil product is collected in a container below the condenser and the non-condensable gases (NCG) are sampled through a gas trap.
2. Experimental
2.1. Sample preparation
PLC controllers using a visual basic interface through a PC, controls the process. The computer stores all the data from sensors, such as, thermocouples and pressure trans- mitters. The heating is provided by electric means as the usage of NCG to provide the process heat is not feasible on this scale, however, is an alternative for larger scale plants. A pre-heater providing 230 W is used to heat the fluidising gas to a pre-set temperature between 300 and 600 8C. The pre-heater is supplied by a 15 A-power source with ceramic beads packed into the heater for better heat transfer. Three other heaters, each providing 120 W with 10 A are used in the reactor to compensate the heat loss through the reactor wall.
An activated sewage sludge from Western Australia was used for this study to observe the effects of pyrolysis temperature and gas residence time on product distribution and their structures. Table 1 shows the analyses of the sewage sludge sample. The dewatered sludge was crushed and sieved to a particle size of 212 –355 mm and dried at 1058C for 24 h. The dry sample was then kept in an airtight container to prevent re-absorption of moisture before experimentation.
2.3. Experimental procedure
2.2. Experimental apparatus
Fig. 1 shows the schematic of the apparatus in which pyrolysis of the sewage sludge is continuously performed in
During an experimental run, the sludge hopper was filled with sewage sludge and the fluidised bed was loaded with quartz sand in the size fraction of 400– 555 mm (Umf:
Fig. 1. A schematic diagram of the fluidised-bed pyrolysis reactor apparatus.
L. Shen, D.-K. Zhang / Fuel 82 (2003) 465–472 467
Table 2 Summary of operating conditions for the experiments conducted
Test no. Pyrolysis temperature (8C) Test duration (min) Gas residence time (s)
Fig. 2. Effect of bed temperature on product yields of oil, NCG and char for sewage sludge expressed in percentage weight daf basis of sludge fed with a constant gas residence time of 1.5 s.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 302 401 451 476 502 526 552 601 524 527 525 523 522 301 298 302 301 297 88 91 90 86 89 93 87 89 92 86 87 93 94 87 90 93 96 89 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 2.0 2.5 2.75 3.0 3.5 2.0 2.5 2.75 3.0 3.5
2 –3 samples taken during a run from the gas-trap and analysed using a GC. Gas yields were calculated from the gas analysis and the total gas flow through the reactor at the time of sampling. Averages of the samples were taken and are reported in this paper. The operating conditions are also summarised in Table 2.
3. Results and discussion
42 mm s21 at 400 8C). Nitrogen, flowing at five times the minimum fluidising velocity was turned on to purge the apparatus from gaseous residues and oxygen. Mains tap water was turned (at 10 l min21) on to provide the cooling medium for the condenser. The operating par- ameters, such as temperature (300 –600 8C), feed rate of sludge (3.3 – 4 g min21) and fluidising velocity were set by the operator using the PC and the signals were then sent to the controllers. After the system reached steady state the screw conveyor (runs by a variable speed motor rotating at 50 –120 rpm) is automatically switched on and the sludge is uniformly fed to the reactor. The process was terminated by stopping the feed and by controlled cooling of the apparatus. Material balances were taken of sludge consumed and all products collected. It was found that the material balances are generally within 92 –95%.
Fig. 2 shows the effect of pyrolysis temperature on oil, char and NCG yields with a constant gas residence time of 1.5 s. The maximum oil yield obtained was 30% daf of the feed at 525 8C. In order to understand this trend, organic bonds that might have been broken at different temperatures during the pyrolysis were obtained from the literature [16, 17] and summarised in Table 3. From Table 3 it can be seen that at the temperature range of 500 –575 8C carboxylic, phenolic, ether oxygen and cellulosic bonds are broken [16, 17]. The temperature range, however, for maximum oil yield is on the upper end of the temperature band for decomposition of carboxylic and phenolic groups. Above 600 8C most of the carboxylic and phenolic bonds have been broken and this is when the oil yield is observed to drop. Suggesting that carboxylic and phenolic breakage from
Table 3 Temperature ranges for different groups of compounds to decompose
Groups of compounds decomposed Temperature range (8C)
Moisture [16] Carboxylic [16] Phenolic [16] Ether oxygen [16] Cellulosic [17] Oxygen containing compounds [16] up to 150 150 –600 300 –600 up to 600 up to 650 150 –900
Oil samples were retrieved from the apparatus with acetone. The cyclone, condenser and other equipment where oil may have deposited were washed with acetone as soon as the experiment was finished to recover the maximum amount of volatile released. The oil is analysed by a GC – MS operating at 300 8C for 40 min so that sufficient separation of the liquid could be achieved. The GC –MS was able to determine the compounds and their molecular structure in the oil ranging from C4 –C40. The GC – MS has an extensive library and the chromatograms of the compounds in the oil were compared with the compounds in the library. The library then generates all the information available of the compounds, including their structures and molecular weights. H NMR analysis was also performed on the oil samples to determine the type of functional groups in the oil. Gas samples were taken in 30 min intervals with
L. Shen, D.-K. Zhang / Fuel 82 (2003) 465–472 468
it suggests that
By convention used for coal oils [23], heavy oils are defined as species with molecular weight greater than 150. In determining the quality of the oil produced, the selectivity of the light oils is of great importance. In most cases, light hydrocarbons are preferred as they can be easily used as fuels. It was found that the species with MW less than 150 in the oil consisted of compounds with carbon numbers less than 9 and of aromatic nature. Fig. 3 shows the selectivity of light oil to heavy oil for light oils temperatures. The selectivity of different increase significantly after 450 8C and stays relatively constant before 450 8C. Therefore, the rate of secondary cracking reactions is only significant after 450 8C at short residence times.
Fig. 3. The selectivity of light oils to heavy oils obtained from sewage sludge pyrolysis at different temperatures with a gas residence time of 1.5 s.
the maximum volume percent
the feed generates the oil. Table 3 was obtained for brown coal, however, this information is not yet available for sludge. Thus this information can only be indicative of the possible bonds that might also be present in the sludge [18, 19]. The char yield, on the other hand, decreased steadily with increasing temperature, while the NCG yield increased continuously.
The oil yields increase with increasing temperature initially as sludge is subjected to more energy, stronger bonds break and thus an increase in larger compounds are observed. The decrease in oil yields above 525 8C is believed to be a result of secondary decomposition reactions which break the oil into lighter, gaseous hydrocarbons. As a consequence the NCG yields also increase. Char yields are expected to decline as more volatiles are released. These results are similar to the literature observations made in both coal decomposition and wood decomposition [13,20 – 22].
Fig. 4 shows the effect of pyrolysis temperature on different NCG expressed as volume percent of purge gas, gas detected after the condenser in Fig. 1. The gases detected by the GC included CH4, C2H4, C2H6, C3H6, C3H8, CO and CO2. Nitrous oxide (N2O) was also detected, however, is 131 ppm at 600 8C and thus the trend cannot be represented clearly on Fig. 4. The yields of all the gases increased steadily with temperature with the exception of CO, a behaviour that is typical of the pyrolysis of peat, lignite as well as cellulosic biomass [23 – 26]. It can be observed that methane has the highest yield, while propylene and propane the lowest among all hydrocarbons. Carbon dioxide has the highest yield of 6.3% of the non-hydrocarbon gases at a 600 8C. Horne and William [24] has found similar concentrations of CO and CO2 under similar operating conditions, where CO and CO2 were found to be 6.8 and 6.7%, respectively, for a biomass. Presently, there is no available data in the literature for nitrous oxide yields for comparison. However, it has been found that N2O is generally formed in reducing atmospheres and low temperatures in fluidised bed reactors [27]. This coincides with the operating conditions of this study.
Fig. 4. Effect of bed temperature on NCG yields for sewage sludge expressed in percentage volume of purge gas with a residence time of 1.5 s.
L. Shen, D.-K. Zhang / Fuel 82 (2003) 465–472 469
All the gas species increase with increasing gas residence time and the rate of increase is similar for all gases. Methane again has the highest conversion while propane has the lowest out of the hydrocarbons and CO2 again is the dominant gas. CO2 had a maximum yield of 7.2%. After 2.75 s the conversions of the hydrocarbons level off. This may be associated with secondary decomposition in the light hydrocarbons itself, where propane is broken down to smaller species. This phenomenon has been observed in previous studies where C3 hydrocarbons start to decrease with increasing pyrolysis temperature and residence time, as seen in both coal [26] and biomass [21] pyrolysis.
Fig. 5. Effect of gas residence time on product yields of oil, NCG and char for sewage sludge expressed in percentage weight daf basis of sludge fed at a constant pyrolysis temperature of 525 8C.
When the yields of all the products are calculated, 92– 95% daf basis of the feed mass can be accounted for. The discrepancy is a result of many factors and one of the major reasons is that a portion of solid char is lost along with the the NCG due to the inability of the cyclone. Also at light operating temperatures used in the experiment, hydrocarbons in the oil would be expected to evaporate as soon as it was formed. When compared to previous studies [24], the total products accounted for compared favourably. The molecular distributions of the species in oil obtained at varying temperature and gas residence time are shown in Fig. 7. In this figure, peaks are seen at a molecular weight of 360. When the gas residence time increases, the curve shifts towards the left, suggesting that smaller species are formed when the time increases, apparently a result of secondary reactions occurring.
The effect of gas residence time was also studied and the trends of product yields are shown in Fig. 5 for a constant pyrolysis temperature of 525 8C. Oil and char yields are both at maximum at the shortest residence time of 1.5 s. This result is in agreement with other work carried out with a variety of biomass and coal feeds [28, 29]. The decrease of char yield is accompanied by evolution of more volatile matter from the sewage sludge. Oil yields decrease from a maximum of 30 to 21% (daf of feed), which can be explained by the occurrence of secondary cracking reactions, breaking up larger species in the volatiles evolved. Henceforth the yields of NCG increase.
The NCG represents all the gases identified in the gas trap. The effect of gas residence time on C1 – C3 hydro- carbons, CO and CO2 are presented in Fig. 6. The maximum yield of N2O is found to be 92 ppm at 3.5 s. This trend is not presented in Fig. 6 as the values are too small in scale.
Table 4 shows all the species that had been identified, in the oil at 525 8C. Not all of the by the GC –MS, compounds were identified in all of the experiments conducted. For example, at 300 8C the heavier compounds (MW . 700) were not detected as shown in Fig. 7. These identified compounds were made up of aromatic rings linked by long straight hydrocarbon chains. Table 5 shows
Fig. 6. Effect of gas residence time on NCG yields for sewage sludge expressed in percentage volume of purge gas at a constant pyrolysis temperature of 525 8C.
L. Shen, D.-K. Zhang / Fuel 82 (2003) 465–472 470
Fig. 7. Effect of pyrolysis temperature and gas residence time on sewage sludge oil composition. (a) Effect of temperature with residence time ¼ 1.5 s. (b) Effect of residence time when temperature ¼ 525 8C.
Table 4 Composition of sewage sludge oil obtained at 525 8C and a gas residence time of 1.5 s
Species wt% (of oil) Species wt% (of oil)
the structures in the oil from GC – MS analysis. It must be noted that compounds below C20 can be unequivocally identified, nevertheless, for compounds greater than C20, there is less accuracy due to the many more possibilities for a match.
0.17 0.38 0.39 0.44 0.85 1.0 1.2 1.4 1.9 2.3 2.9 3.3 3.7 4.3 4.8 5.4 5.9 6.3 6.6 6.7 6.4 6.3 5.8 5.3 4.9 3.9 3.0 2.1 1.2 0.82 C4H8O C5H10O C6H10O2 C7H14O C8H16O2 C9H16O2 C10H18O2 C11H22O C12H20O2 C13H24O C14H22 C15H24 C16H30O2 C17H30O3 C18H32O3 C19H36O2 C20H36O3 C21H38O2 C22H38O3 C24H42O3 C25H46O4 C27H48O5 C28H50O6 C30H54O5 C31H56O7 C32H58O8 C35H60O10 C38H66O12 C39H68O14 C42H72O16
Fig. 8 shows a typical H NMR spectrum of the oil. The identified functional groups and their retention bands are presented in Table 6. The spectra indicate that there are aromatics, hydrocarbons, and – OH bonds in the oil reflecting the functional groups detected by the GC – MS (Table 5). In observing the structures of the oil species from GC – MS analysis and with the confirmation of functional groups from the H NMR spectra, a common unit structure for the oil compounds is proposed, as shown in Fig. 9, which consists of 1 – 3 aromatic rings connected by straight chain alkanes. It is essential to note that H NMR cannot be used alone to identify the structure of oil, it can only assist in other analytical methods such as GC – MS,
L. Shen, D.-K. Zhang / Fuel 82 (2003) 465–472 471
Table 5 Typical structures of main species in the sludge oil as detected by GC–MS
which is used here. As oil is assumed to be large fragments of the original sludge and by knowing the unit structure of oil, it enables an indication of the types of large clusters and bridges present in the original sludge. This may be useful in assisting the development of the sewage sludge structure, a similar methodology has been approached for coal [30].
In the application of the oil as a fuel, it is preferred that the compounds in the oil are straight chain hydrocarbons as they have a high heating value and also lower viscosity. These are important properties of fuels when used in transport and stationary engines. Thus the oils obtained in this study would need to be upgraded. It has been suggested that char produced from pyrolysis can have a catalytic effect on the oil [31], converting it to straight chain hydrocarbons with lower viscosity. The exact component in the char acting as a catalyst is unknown, however, it has been connected to the heavy metals in the char in the form of salts. Such examples of heavy metals used are nickel which has been used to crack coal oils in hydrogenation processes [32]. Bridle [9] has used this technology in the OFS (oil from sludge) process and it has proven to be successful by having a second reactor and using char as a catalyst. An alternative to the present apparatus used in this study is perhaps an addition of a second fluidised bed with char as the bed material and the vapours produced from the first bed bubbling through to crack the heavier compounds.
Fig. 8. H NMR spectra of sewage sludge oil at a pyrolysis temperature of 525 8C and a gas residence time of 1.5 s.
L. Shen, D.-K. Zhang / Fuel 82 (2003) 465–472 472
References
Table 6 Observed H NMR signals and assignment of protons
[1] Australian Water and Waste-Water Association. Nineteenth Federal Retention band Symbol Assignment of protons Conference, Canberra; 1989. [2] Engineering and Water Supply Department. Sewage sludge manage- ment plan, South Australia, Metropolitan Adelaide; 1994. 6.2– 9.2 1.7– 4.0 HAr Ha
[3] Procto and Redfern. Development of a methodology to investigate the cost effectiveness of various sludge management systems. Final report (DSS-UP-205) to Environmental Canada; 1987. 1.0– 1.7 Hb
0.7– 1.0 Hg Aromatics a-CH2 O–CH2 a-CH3 b-CH3 b-CH2 g-CH3 [4] Campbell Environmental Pty Ltd. Attempted odour removal from sewage sludge oil. Laboratory Report no. 88J380, Chemistry Centre; 1989.
[5] Calcutt T, Frost R. Water Pollut Control 1987;90:235 –49. [6] Dumpelmann R, Richarz W, Stammbach MR. Can J Chem Engng 1991;69:962–83.
[7] Avenell CS, Sainz-Diaz C, Griffiths A. J Fuel 1996;75:1767–74. [8] Campbell Environmental Pty Ltd, Upgrading and evaluation of sludge derived oil as a diesel fuel. Chemistry Centre; 1990. [9] Bridle TR. Environ Lett 1982;3:151 –6. [10] Piskorz JD, Scott S, Westerberg IB. I&EC Process Des Dev 1986;25: 265 –70. Fig. 9. A schematic diagram of the common unit structure found in the oil product of sewage sludge from GC–MS analysis.
4. Conclusions
[11] Scott DS, Piskorz J. Can J Chem Engng 1984;60:666–74. [12] Bellmann U, Kummer AB, Kaminsky W. In: Ferrero GL, Maniatis K, Buekens A, Bridgwater AV, editors. Pyrolysis and gasification. London: Elsevier; 1989. [13] Stammbach MR, Karazz B, Hagenbucher R, Richarz W. Energy Fuels 1989;3:255–9. [14] Kistler RC, Widmer CF, Brunner PH. Environ Sci Technol 1987;21: 704 –8.
[15] Collin PJ, Tyler RJ, Wilson MA. Fuel 1980;59:479–86. [16] Murry JB. The yields and properties of volatile and chars from Yallour brown coal. Report MR 191. State Electricity Commission of Victoria; 1969.
[17] Maniatis K, Buekens A. In: Ferrero GL, Maniatis K, Buekens A, Bridgwater AV, editors. Pyrolysis and gasification. London: Elsevier; 1989.
[18] Stammbach MMR, et al. CHIMICA 1988;42:634– 44. [19] Bechers W, et al. J Anal Appl Pyrol 1999;50:17– 30. [20] Scott DS, Piskorz J, Bergougnou MA, Graham R, Overend RP. Ind Engng Chem Res 1988;27:8–15.
[21] Scott DS, Piskorz J. Can J Chem Engng 1982;60:666–74. [22] Scott DS, Piskorz J, Radlein D. Ind Engng Chem Process Des Dev
A maximum oil yield of 30% (wt% daf basis of sludge fed) was achieved at a bed temperature of 525 8C and a gas residence time of 1.5 s. The oil yield reaches a maximum at 525 8C and decreases with increasing residence time. These trends are consistent with previous studies for sewage sludge. There is clear evidence that secondary cracking reactions occur after 450 8C and for longer gas residence times. Also, the selectivity of the lighter oils increases at high temperatures as a consequence of cracking reactions. The results from the oil analysis using GC – MS and H NMR suggests that the structure of sewage sludge oil is made up of a group of aromatic clusters with one to three aromatic rings connected by long straight chain hydro- carbons with hydroxyl groups.
1985;24:581–8.
Acknowledgements
[23] Ugner PE, Suuberg EM. Fuel 1984;63:606–11. [24] Horne PA, William PT. Fuel 1996;75:1051 –9. [25] Funazukuri T. Ind Engng Chem Process Des Dev 1986;25:172– 81. [26] Cliff DI, Doolan KR, Maackie JC, Tyler RJ. Fuel 1984;63:394 –400. [27] Anthony EJ, Cuenca MA. Pressurized fluidized-bed combustion. Glasgow: Blackie; 1995.
[28] Kislter RC, Wldner F. Environ Sci Technol 1987;21:704–8. [29] Konar SK, Boocock DGB, Mao V, Lui J. Fuel 1994;75:642–6. [30] Fletcher TH, Kerstein AR, Pugmire RJ, Solum MS, Grant DM. Energy Fuels 1992;6:414–31.
[31] Lu GQ, Low CF, Lua AC. Fuel 1995;74:344–8. [32] Churin E. Upgrading of pyrolysis oils by hydrotreatment. London: Elseiver; 1991.
