Environ. Sci. Technol. 2006, 40, 6466-6472
Long-Term Performance of High-Rate Anaerobic Reactors for the Treatment of Oily Wastewater JEGANAESAN JEGANATHAN, GEORGE NAKHLA,* AND AMARJEET BASSI Department of Chemical and Biochemical Engineering, University of Western Ontario, London, Ontario, Canada N6A 5B9
Complex oily wastewater from a food industry was treated in three different UASB reactors at different operating conditions. Although all three systems achieved fat, oil, and grease (FOG) and COD removal efficiencies above 80% at an organic loading of 3 kg COD/m3‚d, system performance deteriorated sharply at higher loading rates, and the presence of high FOG caused a severe sludge flotation resulting in failure. Initially, FOG accumulated onto the biomass which led to sludge flotation and washout of biomass. The loss of sludge in the bed increased the FOG loading to the biomass and failure ensued. Contrary to previous findings, accumulation of FOG rather than influent FOG concentrations or volumetric FOG loading rate was the most important factor governing the highrate anaerobic reactor performance. The critical accumulated FOG loading was identified as 1.04 ( 0.13 g FOG/g VSS for all three reactors. Furthermore, FOG accumulation onto the biomass was identified mainly as palmitic acid (>60%) whereas the feed LCFA contained only 30% of palmitic acid and 50% of oleic acid.
Introduction In anaerobic processes, fat, oil, and grease (FOG) is first hydrolyzed to free long-chain fatty acids (LCFA) and glycerol. LCFA is converted into acetate and hydrogen via β-oxidation (1) prior to degradation to methane and carbon dioxide. Glycerol is degraded to 1,3 propandiol (2) and subsequently to acetate and hydrogen. Anaerobic digestion of FOG yields higher biogas production since the fraction of degraded substrate for lipids (0.948) is higher than that of carbohydrates (0.504) and proteins (0.710). In addition, anaerobic treatment of FOG produces less biomass because the fraction of substrate used for cell synthesis for lipids is 0.052 compared to 0.5 and 0.29 for carbohydrates and proteins, respectively (3). High-rate anaerobic reactors, such as upflow anaerobic sludge blanket (UASB) reactors (4-6), hybrid UASB reactors (7), and expanded granular sludge bed (EGSB) reactors (8, 9), are widely used for treating oily wastewaters. Treatment of complex (inhibitory/insoluble) wastewaters (e.g., containing LCFA) in high-rate reactors causes operational problems and in some cases even failure (10, 11). Moreover, most of these literature studies were conducted with synthetic wastewater or low-strength wastewater with synthetically added fats to avoid the complexity and heterogeneity of FOG. * Corresponding author phone: 1 519 661 2111, ext. 85470; fax: 1 519 850 2129; e-mail:
[email protected]. 6466
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 20, 2006
The presence of FOG in wastewater causes two main problems for anaerobic treatment processes: (i) inhibition of methanogenesis due to LFCA (1, 12) and (ii) sludge flotation/washout (9, 11) and hence is limiting for gas production and removal of COD (13). There are contradictory reports on the mode of failure and no definite mechanism has been delineated so far. Some researchers believed that failure is mainly due to the inhibition of methanogens and acetogens (1, 12, 14) by LCFAs. LCFAs disappear from solution and accumulate in solid biomass (1) within 24 h and subsequently, adsorb onto the membrane/cell wall of bacteria which damages the microbial cell transport function or protective function. Koster and Cramer (15) reported that acetoclastic methanogenesis was inhibited by oleic (614 mg/ L), myristic (593 mg/L), lauric (320 mg/L), capric (447 mg/L), and caprylic (972 mg/L) acids. Angelidaki and Ahring (14) showed that oleic acid (100-200 mg/L) and stearic acid (>300 mg/L) inhibited the degradation of acetate, propionate, and butyrate at thermophilic temperature. Nonetheless, Hanaki et al. (1) found in batch assays that glucose fermentation was not affected by the presence of LCFA, if readily biodegradable COD was available. Most of these batch inhibitory studies were based on synthetic LCFAs which allowed the assessment of inhibitory effect of individual LCFAs but did not explore synergistic effects. On the other hand, other researchers concluded that the biomass physically adsorbs fat/lipid causing biomass flotation and washout which also reduces LCFA bioavailability (16) and biogas release. Hwu et al. (11) treated synthetic LCFA mixture in a laboratory-scale UASB reactor and concluded that sludge flotation occurred at relatively low influent concentrations (263 mg LCFA/L or 0.2 g COD/g VSS‚d) below the inhibition levels for methanogens, indicating that clearly in continuous systems, sludge flotation precedes inhibition. However, the effect of FOG/LCFA inhibition or sludge flotation on anaerobic sludge has essentially been studied in batch experiments in serum vials (1, 12, 14-15), or in continuous anaerobic reactors using synthetic LCFAs (7, 1011). So far, there is no report available on the anaerobic treatability of real, high-strength FOG wastewater in a continuous system on a sustained long-term basis which clearly establishes the loading criteria and/or describes the exact mechanism of failure of a high-rate anaerobic reactor treating oily wastewater. Hence, the main objective of this study was to investigate the treatability of complex oily wastewater from a rendering industry in three UASB systems at different operating conditions and delineate the failure mechanism. Evaluation objectives include system performance at various oil loading rates and organic loading rates. COD and FOG mass balances were performed and the accumulation of FOG on the biomass was calculated for each system.
Experimental Section Reactor Setup. System 1 (Figure 1a) essentially consisted of a laboratory-scale UASB reactor which was made of Plexiglas with a working volume of 10 L and had two sections (lower part 50 cm high, 10 cm i.d.; upper part 20 cm high, 20 cm i.d.). System 2 (Figure 1b) also consisted of a laboratoryscale UASB reactor which was made of PVC with a working volume of 15 L (60 cm high, 20 cm i.d.). System 3 (Figure 1c) comprised a packed bed reactor (PBR) and a UASB in series with working volumes of 2 L (PBR 100 cm high, 5 cm i.d.) and 10 L (UASB same as system 1). The internal diameters of the UASBs were carefully selected (g10 cm) to prevent severe biomass flotation due 10.1021/es061071m CCC: $33.50
2006 American Chemical Society Published on Web 09/07/2006
FIGURE 1. Experimental setup. to “piston effect” which generally occurs in small diameters treating FOG wastewaters even at lower LCFA concentrations (10-11). The PBR was operated using sol-gel/alginate beads containing immobilized lipase as media. This pretreatment unit was used to hydrolyze the FOG in the wastewater and the effluent was pumped into the UASB reactor. All three systems were operated at a temperature range of 35 ( 3 °C using water circulation jackets. Reactor Filling. Systems 1 and 3 were seeded with 3 and 5 L, respectively, of anaerobic sludge from a full-scale UASB system treating ethanol wastewater with a volatile suspended solids (VSS) concentration of 47 g/L and specific methanogenic activity (SMA) of 0.55 ( 0.17 g COD-CH4/g VSS‚d, while system 2 was seeded with 6.9 L of anaerobic sludge with VSS of 41 g/L and SMA of 0.52 ( 0.10 g COD-CH4/g VSS‚d. The PBR of system 3 was initially packed with 12 g of immobilized lipase beads that were replaced from time to time. Raw Wastewater. Wastewater used in the entire study was collected from a local rendering industry and stored at 4 °C. The characteristics of the wastewater were analyzed periodically and are given in Table 1 of the Supporting Information. The raw wastewater COD and FOG concentrations for system 1 (operated 2 years earlier) were higher than those for systems 2 and 3 due to the unoptimized oil recovery system. The feed to the reactors was homogenized and diluted to the desired percentage using tap water. Systems 2 and 3 were operated with the same wastewater. However for system 3, Table 1 of Supporting Information shows the values for UASB influent (PBR effluent). Reactor Startup and Operation. The operating conditions are given in the Table 1 as average values for each phase. It must be asserted that the variability in influent loadings among various reactors at similar dilutions stems primarily from the raw wastewater. The pH of each feed was adjusted to 7.2 ( 0.3, the optimum pH for both anaerobic process and lipase treatment using Ca(OH)2 to mitigate odor and prevent Ca2+ leaching from the PBR media, and NaHCO3 to provide buffering capacity. No nutrients were added to the feed since the raw wastewater contained sufficient nutrients for anaerobic degradation. Effluent recirculation ratios of 2 to 7, relative to influent flow, and upflow velocity of 0.05-0.13 m/hr (Table 1a) were employed to achieve 20-30% bed expansion in all 3 systems. The upflow velocities were relatively low compared to the typical range of 1-2 m/h (17). However, upflow velocities as low as 0.02 m/h have been reported (7) for the
treatment of oily wastewater in UASB to prevent sludge flotation. The ratios of TSS at the top of the sludge bed to the bottom of the reactor varied narrowly, averaging 0.52 ( 0.05, 0.63 ( 0.05, and 0.63 ( 0.09 in systems 1, 2, and 3, respectively, thus reflecting comparable degrees of mixing in all 3 systems. Immobilized Lipase Production and Characterization. The Candida rugosa lipase (EC 3.1.1.3), was immobilized on hybrid sol-gel/calcium alginate beads and the detailed immobilization and characterization procedure, lipase assay, and characterization of immobilized lipase are given in supporting materials published elsewhere (18). Analytical Procedures. Regular analyses (grab samples collected each week) from each reactor (influent and effluent) including pH, COD, TSS, VSS, FOG, and alkalinity (as CaCO3) were performed (weekly averaged) using the procedures outlined in Standard Methods (19). Reactor sludge and sludge float also were analyzed. SMA was measured as previously reported (20) using acetate as substrate. Samples filtered through a 0.45 µm GFC glass fiber filter were used for the determination of soluble COD (SCOD) and VFA. The VFA concentration was analyzed by a gas chromatograph (GC Varian 3800, Varian Canada Inc., Mississauga, Canada) with a flame ionization detector equipped with a fused silica column (30 m × 0.32 mm × 5 µm). Nitrogen was used as carrier gas at 5 mL/min and the oven temperature was programmed at 110 °C and increased to 165 °C at a ramp of 20 °C/min. Detector and injector temperature was set at 200 °C. The composition of the biogas was analyzed by a gas chromatograph (GC HP 5890, Technical Lab Services, Ajax, Canada) equipped with a thermal conductivity detector and a steel column packed with 2m Proapak Q (80/100 mesh). Helium was used as the carrier gas at 10 mL/min and the oven temperature was kept at 80 °C. LCFA concentrations were analyzed using the method outlined in AOCS (21) with hexane used as extraction agent instead of heptane (22). The LCFA concentration was analyzed by injecting a 1.5 mL sample into a gas chromatograph (GC Varian 3800) equipped with an auto sampler, FID, and a polar column (15 m × 0.25 mm × 0.25 µm) coated with fused silica (CP-Wax 52 CB). Detector and injector temperature were set at 200 °C and the oven temperature was programmed at 60-200 °C at a ramp of 6 °C/min. Helium was used as carrier gas at a flow rate of 1 mL/min. Split ratio of 1:20 was used at injector. VOL. 40, NO. 20, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6467
TABLE 1. Reactor Operational Parameters and Average Analytical Parameters
system 1 system 2
system 3
system 1 system 2
system 3
a
phase
duration (d)
feed (%)
I II III I II III IV I II III
67 35 64 64 61 40 87 64 61 40
1-15 20 30 5 5 10 20 5 5 10
stage
inf. COD (g/L)
I (11)a II (5) III (10) I (10) II (9) III (6) IV (13) I (10) II (9) III (6)
13.4 ( 2.7 18.1 ( 1.4 26.0 ( 3.4 3.3 ( 0.3 3.0 ( 0.2 4.7 ( 0.3 10.0 ( 1.1 3.2 ( 0.4 3.0 ( 0.2 5.3 ( 0.3
Reactor Operational Parameters HRT recycle flow upflow velocity (d) (L/d) (m/hr)
OLR (kg COD/m3‚d)
FOG LR (kg FOG/m3‚d)
0.08 0.06 0.06 0.05 0.05 0.05 0.05 0.13 0.13 0.13
2.7 ( 0.5 3.6 ( 0.2 5.2 ( 0.7 1.3 ( 0.1 2.4 ( 0.2 3.8 ( 0.2 8.0 ( 0.9 1.3 ( 0.1 2.3 ( 0.1 4.2 ( 0.3
1.0 ( 0.5 1.4 ( 0.2 2.9 ( 0.4 0.5 ( 0.1 1.0 ( 0.1 1.4 ( 0.3 3.4 ( 0.5 0.4 ( 0.1 0.6 ( 0.2 1.0 ( 0.4
Average Reactor Analytical Parameters eff. COD inf. FOG eff. FOG (g/L) (g/L) (g/L)
meth. yield (mL CH4/g CODr)
float volume % (float/feed)
0.32 ( 0.10 0.35 ( 0.05 0.24 ( 0.09 0.48 ( 0.07 0.37 ( 0.06 0.32 ( 0.08 0.24 ( 0.06 0.42 ( 0.10 0.35 ( 0.07 0.18 ( 0.09
0.5 ( 0.2 1.5 ( 0.4 2.9 ( 0.3 0.2 ( 0.1 0.2 ( 0.1 0.2 ( 0.0 0.6 ( 0.1 0.00 0.00 0.4 ( 0.2
5 5 5 2.5 1.25 1.25 1.25 2.5 1.25 1.25
0.3 ( 0.2 1.1 ( 0.3 1.0 ( 0.6 0.5 ( 0.4 0.4 ( 0.1 0.4 ( 0.1 1.1 ( 0.3 0.3 ( 0.2 0.4 ( 0.3 0.7 ( 0.3
14 10 10 30 24 24 24 20 16 16
5.4 ( 1.5 6.9 ( 0.9 14.6 ( 1.8 1.4 ( 0.2 1.2 ( 0.1 1.8 ( 0.4 4.2 ( 0.7 0.8 ( 0.1 0.7 ( 0.2 1.2 ( 0.4
0.1 ( 0.1 0.4 ( 0.2 0.6 ( 0.3 0.2 ( 0.1 0.1 ( 0.1 0.2 ( 0.1 0.3 ( 0.1 0.1 ( 0.0 0.1 ( 0.1 0.3 ( 0.1
Number of samples.
FIGURE 2. Variation of VFA in all three systems
Results Reactor Operation. System 1 was operated at a constant HRT of 5 d throughout the study. Sludge flotation started on day 35 and was about 0.54 ( 0.21% (v float/v wastewater) but increased to 2.88 ( 0.31% in phase III (Table 1b). System 2 was operated at an HRT of 2.5 d for 64 days and then at 1.25 d for the next 188 days. Sludge flotation started on day 51 and was about 0.18 ( 0.14% and increased to 0.61 ( 0.15% in phase IV (Table 1b). Preliminary respirometric studies with immobilized lipase, published elsewhere (18), showed that the COD and FOG reduction were 49% and 45% without pretreatment and 65% and 64% with pretreatment respectively, while growth rate was 3.4 fold higher than without pretreatment with similar Monod half-saturation constants. Hence, the PBR of the third system was packed with immobilized lipase beads and was operated at 4 L/d for 64 days and 8 L/d for another 101 days. The UASB reactor of the third system was operated at the same HRT as system 2. Sludge flotation started on day 123 and was about 0.35 ( 0.18% in phase III (Table 1b), which was much lower than that in the other two systems. VFA, Alkalinity, and pH. Although raw wastewater pH was adjusted to 7.2 ( 0.3, due to fatty acids production in PBR, the effluent pH dropped to 6.2 ( 0.8 and thus was adjusted again to 7.2 ( 0.3 using NaHCO3 before feeding to the UASB. Figure 2 shows the variation of influent and effluent VFA with time in systems 1-3 respectively. Although influent VFA concentrations increased with the decreased influent 6468
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 20, 2006
dilution, effluent VFAs were under 1000 mg/L in all the systems. Figure 1 of the Supporting Information shows the variation of VFA/alkalinity ratio for all three UASB reactors which was lower than the 0.4 recommended for stable operation (23). Thus even at the critical FOG loadings and beyond, none of the three reactors exhibited a sharp rise in VFA, indicative of toxicity to methanogens. COD and FOG Removal. The influent COD and FOG loading were increased gradually as shown in Table 1a and Figure 3, and Figure 2 of the Supporting Information. The COD and FOG removal efficiency based on the influent and effluent concentrations (without considering the loss in float and sludge) were usually above 90% for system 1 but around 88% and 80%, respectively, for system 2. However, a sudden drop was observed in COD removal efficiencies from 90% to 70% in system 1 on day 163 and 88% to 55% in system 2 on day 251. This drop in COD removal also corresponded to a drop in FOG removal from 90% to 80% (system 1) and 90% to 76% (system 2). Moreover, gas production steeply decreased in all systems as shown in Figure 3. Scrutiny of the COD biodegradation to methane depicted in Figure 3 clearly illustrates that the inhibition, i.e., steep drop in biogas production started around days 60, 160, and 90 for systems 1, 2, and 3, respectively, despite respectable overall COD removal efficiencies. Consequently, the reactors were stopped prior to complete cessation of biogas production in light of the impending failure. The PBR of third system removed about 40% of FOG by hydrolysis to LCFAs and the UASB
FIGURE 3. COD loading rate and COD removal efficiencies: (a) system 1; (b) system 2; (c) system 3. reactor of the system 3 removed more than 80% of the PBR effluent COD and FOG. The data for system 3 given in Figure 2 and Figure 3 of Supporting Information are only for the UASB. However, due to the unstable operation of PBR, the
efficiencies fluctuated considerably, and before complete failure, the PBR was removed and the UASB was fed with synthetic oily wastewater (data not shown) in order to continue the project. VOL. 40, NO. 20, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6469
FIGURE 4. COD balance at different OLR. Overall mass balances for COD (Figure 4) and FOG (Figure 3, Supporting Information) were performed at the different loading rates for all UASB reactors. The influent COD at each loading rate was equated to the summation of COD effluent, sludge float, sludge accumulation, and biodegradation using the weekly data (eqs 1-4 of Supporting Information) within the UASB. As shown in Figure 4, at an OLR of about 2.5 kg COD/m3‚d, all three reactors degraded approximately 75% of COD which corresponded to the observed biogas yield (Table 1). However, increase in loading to 5 kg COD/m3‚d caused FOG accumulation in sludge and increase in foam production which reduced the degradation to 40-50%. A mass balance procedure similar to the one for COD was carried out for FOG (Figure 3 of Supporting Information) and a trend similar to that of COD was observed. Both the COD and FOG were generally balanced within 15%. Methane Yield. Methane yield (Yp) was determined using eq 6 of the Supporting Information at 35 °C and atmospheric pressure. As shown from Table 1, Yp values were within a range of 0.24-0.48 mL CH4/g CODremoved except for phase III of system 3. Float Characteristics. Generally, anaerobic processing of high FOG wastewater is hindered by sludge flotation (24). In this study, float was removed manually from the top of the reactor every week. The average float volume (as % of feed volume) and specific characteristics are given in Table 1b and Table 2 of Supporting Information, respectively. System 1 produced higher percentages of float compared to others and this was also reflected in the COD and FOG balance (Figure 4 and Figure 3, Supporting Information). This could be due to higher influent COD and FOG concentration (Table 1, Supporting Information) to the system 1 and the different raw wastewater characteristics (Table 1).
Discussion Although system 1 operated at high influent concentrations, its operation at an HRT of 5 d reduced the OLR (Table 1) thus rendering it comparable with the other two systems. The average influent and effluent COD and FOG concentration are given in Table 1. Effluent COD and FOG concentrations (
