ARTICLE IN PRESS

Bioresource Technology xxx (2009) xxxxxx

Contents lists available at ScienceDirect

Bioresource Technology
journal homepage: www.elsevier.com/locate/biortech

Characterization of an anaerobic bafed reactor treating dilute aircraft de-icing uid and long term effects of operation on granular biomass
Juan Marin *, Kevin J. Kennedy, Cigdem Eskicioglu
Dept. of Civil Engineering, 161 Louis Pasteur, University of Ottawa, Ottawa, Canada K1N 6N5

a r t i c l e

i n f o

a b s t r a c t
Successful treatment of dilute ethylene glycol based-aircraft de-icing uid (ADF) was achieved using a four compartment, anaerobic bafed reactor (ABR). Three ADF concentrations (0.04, 0.07, and 0.13% v/ v) were continuously fed at different hydrological retention times (HRTs; 24, 12, 6 and 3 h) with concomitant organic loading rates (OLRs) varying between 0.3 and 6 kg chemical oxygen demand (COD)/m3/d. ABR achieved over 75% soluble COD removal and an average methane production potential of 0.30 0.05 L CH4/g CODremoved at 33 C for the experimental conditions evaluated. The different experimental conditions tested and a four-month summer shut-down simulation had no signicant effect on reactor performance or on the settling characteristics of the granular biomass, which remained almost constant during the study. Biomass specic acetoclastic activity however, changed through the study; increasing two fold for the last three compartments and decreasing almost the same magnitude for the rst compartment compared to inoculum, suggesting that a new distribution of microbial consortia was established in each compartment of the reactor by the end of the study. 2009 Elsevier Ltd. All rights reserved.

Article history: Received 5 June 2008 Received in revised form 3 December 2008 Accepted 3 December 2008 Available online xxxx Keywords: Acetoclastic activity Anaerobic bafed reactor Aircraft de-icing uid Ethylene glycol Granular biomass

1. Introduction In our global society, ying as a means of transportation has become a necessity. Every year northern countries face the additional challenge of providing safe air travel under severe winter conditions; to accomplish this, airports use de-icing/anti-icing uids (ADF) to ensure that critical aircraft surfaces are free of ice, snow or frost formation. These elements can alter the shape of the wings airfoil section and its surface ow characteristics, causing a loss of lift that might prevent the plane from taking off or cause it to crash if the ice develops in ight (Transport Canada, 1994). In North America, the uids used to deice/anti-ice aircraft are usually composed of ethylene glycol (EG) or propylene glycol (PG) combined with water and other ingredients. The formulation is proprietary and depends on the nal use of the product, but in general contains wetting agents, corrosion inhibitors, colorants and thickeners. In Canada, air transportation is responsible for the greatest volume of release of EG-ADF into the environment. According to a survey carried out under the authority of the Canadian Environmental Protection Act (CEPA), an estimated 7700 tons of EG were used in 1996 for aircraft de-icing/anti-icing operations (Environment Canada, 2000). Test results have indicated that 16% of the EG used to deice planes remains on the aircraft, 35% is blown behind the

* Corresponding author. Tel.: +1 613 562 5800x6070; fax: +1 613 562 5173. E-mail address: jmari040@uottawa.ca (J. Marin). 0960-8524/$ - see front matter 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.biortech.2009.11.055

aircraft and about 50% falls to the ground in the vicinity of the aircraft following application (Simpson, 1997). Runoff from aircraft de-icing activities results in the release of large quantities of dilute EG-ADF to the environment. While the high 5 d-biochemical oxygen demand (BOD5) of de-icing uids is their main impact on the environment (Miller, 1979; Sabeh and Narasiah, 1992); there is also evidence that the toxicity of ADF increases due to the presence of additives in the formulation (e.g. triazoles, organic amine bases, etc.). It has been reported that ADF formulations were signicantly more toxic to the water ea Cercerodaphnia dubia, and the fat head minnow, Pimephales promelas than pure EG and PG (Pilliard, 1995). Hence, in order to protect the quality of the receiving waters the collection and treatment of this type of wastewater is mandatory under federal authority. ADF is applied at gate areas which have been specially designed for the collection of such streams, during collection ADF runoff get usually mixed with the pavement de-icing used at the specic airport (sodium acetate, urea, etc.) and other contaminants such as grease, which at higher concentrations could inhibit the anaerobic digestion process. However, due to the equalization process during collection their inhibitory effect is reduced (Marin and Kennedy, 2006). Strength of such wastewater in airport collection systems is highly variable and depends in many factors such as weather conditions, airport size, collection system, etc. (Switzenbaum et al., 2001). Marin and Kennedy (2006), reported COD concentrations below 900 mg/l at a medium size airport.

Please cite this article in press as: Marin, J., et al. Characterization of an anaerobic bafed reactor treating dilute aircraft de-icing uid and long term effects of operation on granular biomass. Bioresour. Technol. (2009), doi:10.1016/j.biortech.2009.11.055

ARTICLE IN PRESS
2 J. Marin et al. / Bioresource Technology xxx (2009) xxxxxx

To date, there are reports that show the efciency of anaerobic treatment for concentrated ADF (Darlington and Kennedy, 1999; Schoenberg et al., 2001; Pham, 2002; Zitomer and Tonuk, 2003); however, for dilute ADF wastewater the number of reports is limited. The anaerobic bafed reactor (ABR) presents certain advantages that make it suitable for the anaerobic treatment of diverse types of wastewaters (Barber and Stuckey, 1999). The ABR is compartmentalized horizontally, using a series of vertical bafes, which force the wastewater to ow over and under the bafes as it travels from the inlet to the outlet. ABR can achieve high volumetric OLRs maintaining at the same time long SRT independent from the HRT. According to Weiland and Rozzi (1991), the most signicant advantage of the ABR is its ability to separate acidogenesis and methanogenesis longitudinally down the reactor, which allows it to behave as a two-phase system (Cohen et al., 1982; Baloch et al., 2007). Barber and Stuckey (1999) reported the successful treatment of different types of dilute wastewaters using ABR technology. However, no studies have evaluated the treatment of dilute ADF wastewater using ABR. Seasonally generated wastewaters may have great variations in strength from the beginning to the end of the season. De-icing/ anti-icing activities take place during the winter months and extend depending on weather conditions. Therefore, it is important to have a system robust enough to handle not only the variations in OLR but also the ability to maintain a good performance after a shut-down period. Anaerobic granular biomass can tolerate starvation conditions with low detrimental effects on its activity (Mulder et al., 2001) and its use has made the anaerobic treatment of diverse types of seasonally generated wastewaters possible. However, in order to take the best advantage of such systems it is important to understand the long term effects that certain types of wastewaters or operation conditions might create on granular biomass. The primary objective of this study was to evaluate the performance (in terms of soluble COD removal) of a mesophilic (33 C) ABR inoculated with granular biomass for the treatment of dilute EG-based ADF wastewater under different combinations of inuent ADF concentration and HRT. In addition, the long term effects of different experimental conditions on the granular biomass were also studied in terms of acetoclastic activity and settling characteristics. The recovery ability of the ABR granular biomass was reassessed after a four-month shut-down period.

2.2. Experimental design To investigate the performance of ABR in terms of COD removal a two-factor factorial design with one observation per cell was employed. The two factors considered were HRT and inuent ADF concentration (4 and 3 levels, respectively). Statistical Analysis System (SAS, V.8) was used for data analysis.

2.3. Experimental procedure Both ABRs were placed in a temperature-controlled room (33 2 C) and inoculated with 10 L of anaerobic granular biomass equally distributed in the four compartments resulting in a total reactor biomass concentration of 10.5 g VSS/L. Inoculum with specic acetoclastic activity (SAA) of 0.20 g Ac/g VSS/d was obtained from a full-scale anaerobic up ow sludge bed (AUSB) reactor treating chemical/thermal mechanical pulp efuent at Lake Utopia Paper Ltd. (New Brunswick, Canada). The carbon source was a formulated ADF wastewater using EG based UCAR XL-54 de-icing uid from Union Carbide. Experiments were conducted at ADF concentrations of 0.04, 0.07 and 0.13% (v/v), corresponding to soluble diluted feed COD concentrations of 300, 500 and 750 mg/L, respectively. The feed solution also had other components to provide the granular biomass with adequate sources of nitrogen, phosphorus, miscellaneous nutrients, and alkalinity was provided in excess for pH control (Table 1). The average COD:N:P ratio of the feed solution was 200:5:1. During acclimation, ADF wastewater was prepared every other day with tap water and kept at 4 C. For the specic runs, feed was prepared as needed, based on HRT and because of the high ow rates it was pre-warmed in order to maintain reactor mixed liquor at 33 C. After inoculation, acclimation was initiated by continuous feeding at a long HRT (80 h) with high strength ADF feed (4500 mg COD/L). Acclimation lasted two months, during which production of volatile fatty acids (VFAs) as well as the pH was closely monitored. If the VFAs reached concentrations higher than 500 mg acetate/L or the pH was lower than 7, the reactors were set in full recycling mode (64 L/d) until the normal conditions of operation were restored (VFAs < 500 mg acetate/ L and pH 7). After acclimation, the feed concentration was gradually reduced and HRT gradually decreased until the base line conditions of 40 h-HRT and feed concentration of 750 mg COD/L were achieved. From this base line, the feeding program of each run with dilute ADF was achieved by either increasing the ow rate 510% each day at constant ADF concentration, or the ADF concentration was increased or decreased by an increment of 0.01% ADF per day at a constant HRT. During run-to-run transitions, the levels of VFAs, pH and efuent soluble COD were closely monitored. Total time for experimentation was approximately 13 months (including the four-month summer shut-down). Reactor performance was assessed according to data obtained during steady-state conditions. For the reactors to be considered at steady-state, the concentration levels of VFAs had to be constant (50 mg/L) as well as the COD removal efciency (10% uctuation) for a period of 5-HRTs. Once

2. Methods 2.1. Apparatus Continuous-ow experiments were performed with two identical Plexiglas ABRs (56 30.5 25.5 cm, LWH) as depicted elsewhere (Marin et al., 2007). Both reactors had four compartments comprised of a down-ow section which ended in a 45 angled portion and an up-ow section. Total reactor volume was 32 L (8 L for each compartment). Biogas sampling ports were connected to a header by 380 PVC, Nalgene tubing and directed to a wet-tip gas meter for measurement. Dilute ADF wastewater owed from the top of the down-ow section of the rst compartment and through the portion of granular biomass retained in that section and then up through the bed section in the up-ow section. The process was repeated for each compartment. Finally, the treated efuent exited from the up-ow section of the fourth compartment. No recycling was used during the regular operation (after acclimation) and no direct control of biomass distribution in the ABR compartments was used.

Table 1 Composition of synthetic ADF-based feed solution. Component (g/L) COD NH4HCO3 NaHCO3 KHCO3 KH2PO4 K2HPO4 (NH4)2SO4 Yeast extract 0.04% ADF (v/v) 0.3 0.06 2.1 2.4 0.006 0.008 0.015 0.003 0.07% ADF (v/v) 0.5 0.1 2.4 2.6 0.01 0.013 0.025 0.006 0.13% ADF (v/v) 0.75 0.15 2.4 2.9 0.015 0.02 0.04 0.008

Please cite this article in press as: Marin, J., et al. Characterization of an anaerobic bafed reactor treating dilute aircraft de-icing uid and long term effects of operation on granular biomass. Bioresour. Technol. (2009), doi:10.1016/j.biortech.2009.11.055

ARTICLE IN PRESS
J. Marin et al. / Bioresource Technology xxx (2009) xxxxxx 3

this state was reached, reactors were kept at that condition for other 5-HRTs. For every steady-state, triplicate measurements of VFAs, pH, efuent COD, compartment COD prole, efuent total suspended solids (TSS) and volatile suspended solids (VSS), were performed; biogas production and composition was measured once and biomass settling characteristics were measured by duplicate. SAA was measured in duplicates only for the inoculums; for every compartment of the ABR at the beginning of the summer shut-down and at the end of the summer shut-down. 2.4. Reactor restart up After a four-month summer shut-down simulation, ABR continuous operation started with ADF inuent concentration of 750 mg COD/L and 60 days HRT. Once base line conditions of 750 mg COD/L and 40 h-HRT were reached four additional runs were performed at dilute ADF inuent concentrations of 750 and 500 mg COD/L, and 24, 12 and 6 h-HRT. In this last part of the study ABR performance was assessed only in terms of COD removal. 2.5. Analytical methods Methane content of biogas was determined with a Hewlett Packard 5710a gas chromatograph (GC) (Agilent, Santa Clara, CA, USA); equipped with a thermal conductivity detector and a 3380A model integrator using the method described by van Huyssteen (1967). The GC column was a Porepak T (6.35 mm 304.3 cm) set at 70 C with a helium gas carrier ow of 40 mL/min. Biogas samples were taken from a port at the top of the reactor, with an airtight syringe and 0.5 mL of biogas was injected into the GC for the determination. VFAs were determined by the internal standard method described by Ackman (1972), using a HewlettPackard 5840A GC (Agilent, Santa Clara, CA, USA), equipped with a ame ionization detector, an auto sampler, a 5840 model integrator and a Chromosorb 101 packed column (304.8 cm 2 mm ID, 80/100 mesh size). The oven temperature was 180 C, the injector temperature was 250 C and the detector temperature was maintained at 350 C. The ow rate of the formic acid saturated helium carrier gas was 15 mL/min. Before injection (10 lL) to the GC, samples were centrifuged at 5000 rpm for 5 min in a micro centrifuge, and the supernatant was diluted with an equal volume of internal standard containing 1000 mg/L isobutyric acid. TSS and VSS determinations were based on procedures in Standard Methods (APHA, 1998). Well-mixed samples were ltered through a pre-weighed GF/C berglass lter (VWR International. Mississauga, ON, Canada) and dried to a constant weight in a 105 C oven. The dried lter was ignited at 550 C in a mufe furnace for 40 min for VSS determination. Soluble COD was determined using colorimetric and titrimetric techniques based on procedures in Standards Methods (APHA, 1998). Samples were centrifuged for 15 min at 10,000 rpm and the supernatant was used for COD determination. For the COD colorimetric technique, a PerkinElmer spectrophotometer (Waltham, MA, USA) was used to measure the absorbance at 600 nm. For the titrimetric technique (COD < 400 mg COD/L), samples were titrated with 0.1 M ferrous ammonium sulphate (FAS). Ferroin indicator solution was used to indicate the end-point of the titration. Biomass acetoclastic activity determination was based on Speece (1996); granular biomass (10 mL) from each compartment of the ABR were anaerobically transferred to 130 mL glass serum bottles and diluted with 30 mL of dened medium to give a biomass concentration of approximately 10 g VSS/L. To the sealed bottles 0.2 mL acetic acid stock solution (Pham, 2002) was injected to give an initial acetate concentration in the range of 12001500 mg/L The acetate consumption rate was measured by monitoring the change in acetate

concentration over time at 35 C. Biomass settling determination was based on the procedure described by Andras et al. (1989). A 10 mL granular biomass sample, taken from the lower sampling port of the up-ow section of each compartment, was placed in a glass up-ow velocity test tube (L = 20 cm, ID = 1.9 cm); tap water at 35 C was pumped through the tube for 5 min at successively increasing ow-rates. The fractions of biomass exiting from the tube were collected on pre-weighed Whatman 185 mm diameter grade 1 lter paper (Fisher Scientic, Ottawa, ON, Canada) and dried overnight at 105 C. The difference in weight yielded the TSS fractions exited at each ow-rate. Settling curves were determined by plotting the cumulative TSS exiting fractions versus up-ow velocity. The v50 was considered to be the up-ow velocity at which 50% of the granules were washed out. 3. Results and discussion 3.1. ABR performance Arithmetic means and standard deviations of six steady-state measurements from the 15 experimental conditions evaluated during this study are summarized in Table 2. From column 4 in Table 2, it can be observed that ABR COD removal efciency was strongly inuenced by HRT, reaching the highest COD removal (9296%) at 24 h-HRT for ADF inuent concentration of 750 mg COD/L and 68% for the worst case scenario (3 h-HRT and 750 mg COD/L inuent concentration). The same effect (higher COD removals at low OLRs) was observed for all ADF inuent concentrations tested (data not included). Since there was only one observation per cell in the experimental design, variables interaction can not be separated and the error variance can not be estimated, unless the interaction effect is zero. Using the test developed by Tukey the presence of interaction can be detected (Montgomery, 2001). Tukeys test for interaction with 95% condence interval is F0.5,1,5 is 6.6, since Fo < F, it can be concluded that there is no evidence of interaction in the data and from the main effects only HRT is signicant (Table 3). It is also important to notice from Table 2 the low efuent VSS (9.5170.5 mg VSS/L); which were basically cell debris and some disrupted granules. It was only at the lowest HRT (3 h) that a small fraction of well dened granules (average diameter < 1 mm) were observed in the efuent; most likely, those small size granules were washed out of the reactors due to a combination of high ow-rate and biogas production, nevertheless, ABR biomass inventory remained high through the study, indicating that over a wide range of operation conditions ABR can perform successfully, which is desired during treatment of seasonally generated wastewaters. 3.2. Soluble COD and VFAs proles Steady-state COD and VFAs proles were created in order to observe the stabilization of inuent COD through each of the four compartments of ABR. A previous study using an ABR with the identical number of compartments showed that the reactor behaved as four continuous stirred tank reactors in series (CSTRs) (Barriault, 2003). Hence, the efuent from the rst compartment was considered the inuent for the second compartment and so on. Fig. 1 shows the typical COD consumption proles at various HRTs for all ADF inuent concentrations tested. We can observe from Fig. 1c that at lower ADF inuent concentrations, the COD consumption from compartment-to-compartment was more irregular in the rst two compartments compared to the COD consumption proles when treating higher ADF inuent concentrations (Fig. 1a and b). At low ADF inuent concentrations, the COD concentration did not always decrease from one compartment to the

Please cite this article in press as: Marin, J., et al. Characterization of an anaerobic bafed reactor treating dilute aircraft de-icing uid and long term effects of operation on granular biomass. Bioresour. Technol. (2009), doi:10.1016/j.biortech.2009.11.055

ARTICLE IN PRESS
4 Table 2 Typical steady-state data of overall ABR performance*. Run BL CRc 1 4 7 10
* a b

J. Marin et al. / Bioresource Technology xxx (2009) xxxxxx

HRT (h) 40 24 24 12 6 3

OLR (kg COD/m3 d) 0.3 0.75 0.75 1.5 3 6

COD removala (%) 96 1.2 93 1.02 92 1.6 90.4 1.04 77 0.5 68 1.5

Total biogas production (L/d) 2.4 0.3 7 0.5 7.3 0.6 14.2 0.6 14 1.6 36 3.5

Biogas composition (%CH4) 64 68 63 69 69 54

Efuent VSSa (mg/L) 9.5 1.6 12.8 3.1 13.7 1.3 51 10.4 64 5.6 170.5 14.1

Data shown correspond to the experimental runs with 750 mg/L ADF inuent concentrations at the different HRTs. Data indicates arithmetic mean standard deviations of triplicates (ve steady-states). Data indicates arithmetic mean standard deviations of triplicate measurements for ABR run at base line conditions (BL). c Data indicates arithmetic mean standard deviation of triplicates for the 2 ABRs run at the same experimental conditions or control run (CR).

Table 3 ANOVA for no-interaction determination. Source of variation Concentration HRT Non-additivity Error Total Sum of squares 33.35 1362.05 0.5541 39.09 1413.77 Degrees of freedom 3 2 1 5 11 Mean square 16.67 452.01 0.5541 7.81 Fo 2.133 58.07 0.07 Pr > F 0.0448 <0.0001

(a)
sCOD (mg/L)

800 600 400 200 0

750 mg COD/L
concentration at 24 and 12 h-HRTs, the COD increased from compartments 1 to 2. At higher ADF inuent concentrations (500 and 750 mg COD/L), the COD concentration was high in each compartment and COD consumption occurred stepwise all through the four compartments (Fig. 1a and b). Compartment-to-compartment COD consumption and VFAs production proles shown in Fig. 2 suggest that at lower OLRs, only the biomass in the rst two compartments were being stressed (increased propionic acid concentration) while at higher OLRs, the biomass in more compartments was being stressed (Fig. 2c). The production of soluble microbial products (SMP) can be another explanation for the observed compartment-to-compartment COD increase. SMP production associated with anaerobic digestion of dilute wastewaters has been previously reported (Langenhoff et al., 2000). According to Barker and Stuckey (1999), there are many factors for the production of SMP in anaerobic systems, but normally under starvation conditions its production is favored. This effect was expected to be observed in the last compartments of the ABR, which were exposed to the lowest substrate concentrations; instead it was normally seen in the rst two compartments under high organic load. Hence, it may be speculated that SMP were produced in the rst two compartments, but that they were consumed to a certain extent in the last compartments. In general, the trend observed was a gradual COD consumption as the dilute ADF wastewater moved through the reactor. The same trend was observed for the VFAs; they were gradually consumed as the dilute ADF moved from the rst compartment to the last one. For the experimental conditions mentioned above, there was no relationship between the higher inuent COD concentration and the VFA production in the reactor compartments. It can be observed from Fig. 2 that even if an increase in COD occurred for certain compartments, VFAs concentrations were rather low for the same compartments (more evidence of SMP production). In this study it is difcult to address the percentage COD removal of each compartment, because as mentioned before, in some cases COD concentration in the subsequent compartment was higher or the same. However, the four compartments together functioned to stabilize the ADF. At certain OLRs, the rst and second compartments showed signs of stress (low COD removal, high VFAs) and this effect disappeared in the last compartments, but, if the last compartments were stressed, the total performance of the

3 h HRT 6 h HRT 12 h HRT 24 h HRT

2 3 Compartment

(b)
sCOD (mg/L)

500 mg COD/L
500 400 300 200 100 0 1 2 3 4
Compartment 3 h HRT 6 h HRT 12 h HRT 24 h HRT

300 mg COD/L

(c)
sCOD (mg/L)

300 250 200 150 100 50 0 1 2 3 Compartment 4

3 h HRT 6 h HRT 12 h HRT 24 h HRT

Fig. 1. Steady-state COD consumption proles for ABR under different ADF inuent concentrations of (a) 750 mg COD/L, (b) 500 mg COD/L, and (c) 300 mg COD/L at the different HRTs. Figure shows the average values of three COD measurements per ve steady-state conditions and error bars indicate the standard deviations.

next as expected and as reported in other ABR studies (Barriault, 2003; Baloch et al., 2007). Instead, for 300 mg COD/L ADF inuent

Please cite this article in press as: Marin, J., et al. Characterization of an anaerobic bafed reactor treating dilute aircraft de-icing uid and long term effects of operation on granular biomass. Bioresour. Technol. (2009), doi:10.1016/j.biortech.2009.11.055

ARTICLE IN PRESS
J. Marin et al. / Bioresource Technology xxx (2009) xxxxxx 5

(a)
Concentration (mg/L)

350 300 250 200 150 100 50 0

300 mgCOD/L and 24 h HRT

sCOD Acetic acid

consumption of VFAs took place, it can also be stated that in those compartments, methanogenesis took place; its extent compartment-to-compartment was dependent on the experimental conditions. Hence, based on pH, COD and VFA compartment-tocompartment proles, it can be suggested that under high OLRs, the two rst compartments were used as acidogenic units while at lower OLR only the rst compartment was used as the acidogenic unit. 3.3. Biomass settling characteristics The success of high rate anaerobic reactors such as the ABR is related to its ability to maintain a high biomass inventory, which allows for longer SRT independent of the HRT. It is important to determine if the granular biomass can keep its ability to settle under determined reactor operation conditions, since this could eventually result in biomass washout and reactor failure. Table 4 shows typical v50 values obtained for some of the experimental conditions. From Table 4 it can be observed that no detrimental change occurred in the settling characteristics of the granular biomass under the continuous operation of the ABR compared to inoculum. An average v50 of 25.9 2.2 m/h suggested a stable granular biomass under a wide range of operational conditions, the typical S shape of the curves for the determination of v50 was also maintained during all the experimental conditions tested (Fig. 3a and b, for COD inuent concentration of 750 mg/ L and HRTs of 24 and 3 h, respectively). These results are in agreement with other studies on settling characteristics of anaerobic granular biomass treating ADF (Pham, 2002; Barriault, 2003). Before summer shut-down, the v50 for the rst compartment of both ABRs was slightly lower (20.1 1.3 m/h) compared to the remaining compartments and inoculum. Since the rst compartments of both reactors were exposed to the highest OLRs and ADF inuent concentrations, the granular biomass started losing its original coloration and a crust-type formation around the granules appeared. However, no granule disruption was observed. In other studies (Beloch et al., 2007) granule disruption was reported after exposure to a high OLR (13.4 kg COD/m3/d) for a brewery wastewater, which suggests that the type of wastewater and operation conditions have an impact on granular biomass stability hence, long term reactor performance. It is important to notice the high v50 values in the last two compartments (Table 4); if it is considered that methanogenic microorganisms are distributed more favorably in this portion of the reactor, biomass washout could have an adverse effect on total ABR performance. 3.4. Biomass specic acetoclastic activity The specic acetoclastic activity (SAA) test results determined for the inoculum and granular biomasses contained in each compartment at the end of the experimental runs and after summer

2 3 Compartment

(b)
Concentration (mg/L)

250 200 150 100 50 0 1

300 mgCOD/L and 12 h HRT

sCOD Acetic acid Propionic acid

2 3 Compartment

Concentration (mg/L)

(c) 800
700 600 500 400 300 200 100 0 1

750 mgCOD/L and 3 h HRT

sCOD Acetic acid Propionic acid Butyric acid

2 3 Compartment

Fig. 2. Steady-state COD and VFAs production levels of an ABR with ADF inuent concentrations of (a) 300 mg COD/L and 24 h-HRT (propionic and butyric acids were undetected), (b) 300 mg COD/L, and 12 h-HRT and (c) 750 mg COD/L and 3 hHRT. Figure shows the average values of three COD measurements and error bars indicate the standard deviations for ve steady-state conditions.

reactor decreased, and the accumulation of VFAs started to increase gradually in each compartment. It is known that the ABR and its compartmentalization favor the separation of acidogenesis and methanogenesis, thus, the measurement of methane production from each compartment could have given more indications about the extent of both processes; however, methane production was measured for the total system. Since in general the rst compartment showed the highest COD and VFAs concentrations and lower pH values compared to the three remaining compartments, it can be assumed that this compartment was more acidogenic in nature than the others. In the last three compartments, the

Table 4 Typical settling velocity (v50) for granular biomass after steady-state conditions. Seed granular biomass Run v50 (m/h) 1 5 9 12 Average
a

27.6 0.28 HRT (h) 24 12 6 3 Compartment 1 26.6 26 23 19.2 24.5 2.8a Compartment 2 25.8 27.7 27.6 24.1 26.3 2.5a Compartment 3 27.1 26.8 26.7 25.8 26 1.2a Compartment 4 26.9 28.4 29.8 28.1 27.2 1.2a

CODinuent (mg/L) 750 500 300 300

Mean standard deviations of duplicate measurements for ve steady-state conditions evaluated.

Please cite this article in press as: Marin, J., et al. Characterization of an anaerobic bafed reactor treating dilute aircraft de-icing uid and long term effects of operation on granular biomass. Bioresour. Technol. (2009), doi:10.1016/j.biortech.2009.11.055

ARTICLE IN PRESS
6 J. Marin et al. / Bioresource Technology xxx (2009) xxxxxx Table 7 Typical settling velocity (v50) for granular biomass after summer shut-down conditions*. Compartment 1 v50 (m/h) 17.6 0.49 Average Compartment 2 21.1 0.3 20.7 2.2 Compartment 3 22.5 0.9 Compartment 4 21.9 0.5

(a)
125
Cumulative TSS Washout (%)

C.1

C.2

C.3

C.4

Average

750 mg COD/L & 24 h HRT

100 75 50 25 0 0 10 20 30
Upflow velocity (m/h)
*

Data indicate the mean and standard deviations of two measurements.

40

50

60

(b)
Cumulative TSS Washout (%) 125

C.1

C.2

C.3

C.4

Average

750 mg COD/L & 3 h HRT

100 75 50 25 0

shut-down are presented in Table 5. From Table 5, it can be observed that SAA changed all through the compartments of ABR. A decrease in SAA was observed for the rst compartment possibly due to the acidogenic nature of this compartment. For the remaining compartments, the SAA increased, which indicates that different groups of bacteria were selectively developed in each subsequent compartment. The higher the SAA, the greater the proportion of acetoclastic methanogens compared to other microbes that make up the consortia. Biomass contained in the last compartments had a high SAA (0.3 0.01 g Ac/g VSS/d), which reinforces the fact that ABR behaved as a two-phase system, where the methanogenic portion was localized most preferentially in that part of the reactor.

10

20

30

40

50

60

Upflow velocity (m/h)

3.5. Reactor reassessment after summer shut-down

C.4 Average

(c)
120
Cumulative TSS washout (%)

C.1

C.2

C.3

100 80 60 40 20 0 0 10 20 30 40 50 60
Upflow velocity (m/h)

Fig. 3. Typical settling curve for granular biomass after steady-state run conditions of (a) 750 mg COD/L and 24 h-HRT, (b) 750 mg COD/L and 3 h-HRT, and (c) after summer shut-down. C.1, C.2, C.3 and C.4 indicate compartment number 1 through 4, respectively. Settling tests were performed in duplicates.

Table 5 Specic acetoclastic activity (gAc/g VSS/d) for granular biomass. At the end of steady-state conditions Inoculum Compartment Compartment Compartment Compartment 1 2 3 4 0.20 0.01 0.04 0.01 0.33 0.02 0.36 0.04 0.30 0.01 After summer shut-down 0.008 0.0098 0.305 0.021 0.290 0.028 0.341 0.056

Results from the ABR reassessment runs are shown in Table 6. Summer shut-down had no signicant effects on ABRs COD removal efciency; which remained well over 67% for the highest OLR (2 kg COD/m3 d) tested. However, granular biomass characteristics were affected: a slight increase in the efuent VSS was observed compared with the initial steady-state conditions and even though most of the granules washed out were discolored and broken, some well dened granules were also observed. This was most likely due to biomass endogenous decay during the shut-down period. Biomass settling characteristics decreased (Table 7 and Fig. 3c), and the effect was more pronounced in the rst compartment, for which v50 decreased from an average 24.5 2.8 to 17.6 0.5 m/h. According to Schmidt and Ahring (1996), granular biomass can be divided into three fractions based on the settling velocities reported: a poor settling fraction (20 m/h), a satisfactory settling fraction (2050 m/h) and a good settling fraction (>50 m/h). Although the granular biomass settling velocity of the rst compartment falls into poor settling category (Table 7), after the summer shut-down, the average settling velocity for the whole reactor was still acceptable (20.7 2.1 m/h). This also indicates that new granular biomass has to be eventually added to the reactor if its sole purpose is for ADF wastewater treatment. A decrease in the SAA was also observed after the summer shut-down, especially for the rst and third compartments (Table 5). It can be speculated that the population of acidogenic bacteria in the consortia increased in those compartments during the summer shut-down.

Table 6 Steady-state data of ABR performance after summer shut-down. Run 1 2 3 4

*

HRT (h) 24 24 12 6

OLR (kg COD/m3 d)) 0.75 1 1.5 2

CODin (mg/L) 750 500 750 500

COD* Removal (%) 90.5 2.05 92.1 1.9 88 2.3 67.6 6.2

Total biogas production* (L/d) 7.8 0.68 6.6 0.57 14.4 0.91 26.8 3.38

Efuent VSS* (mg/L) 18.7 4.0 28.1 4.4 60.6 10.1 73.5 8

Data indicate the mean standard deviations of triplicates (ve steady-state conditions).

Please cite this article in press as: Marin, J., et al. Characterization of an anaerobic bafed reactor treating dilute aircraft de-icing uid and long term effects of operation on granular biomass. Bioresour. Technol. (2009), doi:10.1016/j.biortech.2009.11.055

ARTICLE IN PRESS
J. Marin et al. / Bioresource Technology xxx (2009) xxxxxx 7

4. Conclusions The ABR proved to be effective for the treatment of dilute ADF-based wastewater, reaching an average methane production of 0.30 0.05 L CH4/g COD removed at 35 C and soluble COD removal efciencies over 75% for most of the experimental conditions evaluated. Better COD removal efciencies were achieved at high ADF concentrations and at HRTs longer than 6 h. The different experimental conditions evaluated did not have a signicant effect on the settling characteristic of the granular biomass. The SAA of the granular biomass changed throughout the reactor, indicating that a different distribution of microorganisms occurred in each compartment by the end of this study. This nding supports the idea that the conguration of this reactor can effectively separate acidogenesis and methanogenesis in a single unit, which could have ramications in terms of special treatment applications such as sulfate reduction or maybe denitrication. A four-month summer shut-down period had no signicant impact on ABR COD removal efciency. However, the settling characteristics and SSA of the granular biomass contained in the rst compartment decreased, but without detrimental effect on the whole ABR operation. Overall, ABR seems to be an adequate technology for the treatment of this type of seasonally generated wastewater, offering the advantage of stable performance under a wide range of OLRs and a fast recovery after a shut-down period.

References
Ackman, R.G., 1972. Porous polymer bead packing and formic acid vapour in GLC of volatile fatty acids. J. Chromatogr. Sci. 10, 560565. Andras, E., Kennedy, K.J., Richardson, D.A., 1989. Test for characterization of anaerobic sludge. Environ. Technol. Lett. 10, 463470. APHA-AWWA-WEF, 1998. Standard Methods for the Examination of Water and Waste Water, 20th ed. American Public Health Association, Washington, DC. Baloch, M.I., Akunna, J.C., Collier, P.J., 2007. The performance of a phase separated granular bed bioreactor treating brewery wastewater. Bioresour. Technol. 98, 18491855. Barber, W.P., Stuckey, D.C., 1999. The use of the anaerobic bafed reactor (ABR) for wastewater treatment: a review. Water Res. 33 (7), 15591578. Barker, D.J., Stuckey, D.C., 1999. A review of soluble microbial products (SMP) in wastewater treatment systems. Water Res. 33 (14), 30633082. Barriault, M., 2003. Effect of recycle on treatment of aircraft de-icing uid in an anaerobic bafed reactor. M.A.Sc. Dissertation, Department of Chemical Engineering, University of Ottawa, Ottawa, ON, Canada.

Cohen, A., Breure, A.M., Van Andel, J.G., Van Deursen, A., 1982. Inuence of phase separation on the anaerobic digestion of glucose II. Stability and kinetic responses to shock loadings. Water Res. 16, 445449. Darlington, C., Kennedy, K.J., 1998. Biodegradation of aircraft deicing uid in an upow anaerobic sludge blanket (UASB) reactor. J. Environ. Sci. Health A33 (3), 339351. Environment Canada, 2000. Canadian environmental protection act priority substances list-supporting document for the environmental assessment of ethylene glycol. Commercial Chemicals Evaluation Branch, Environmental Canada, Hull, Quebec. Langenhoff, A.A.M., Intrachandra, N., Stuckey, C., 2000. Treatment of dilute soluble and colloidal wastewater using an anerobic bafed reactor: inuence of hydraulic retention time. Water Res. 34 (4), 13071317. Marin, J., Kennedy, K.J., Eskicioglu, C., Hamoda, M.F., 2007. Compartmental anaerobic bafed reactor kinetic model for the treatment of dilute aircraft deicing uid. In: Proceedings of the IASTED Conference on Environmental Modeling and Simulation, August 2022, Honolulu, Hawaii, USA. Marin, J., Kennedy, K.J., 2006. Treatment of propylene-glycol based aircraft deicing uid using an anaerobic sludge blanket reactor (UASB). Technical report for Jacques Withford, Canada. Miller, L.M., 1979. Investigation of selected potential environmental contaminants: ethylene glycols, propylene glycols, and butylenes glycols. USEPA Report #560/ 79-006, Ofce of Toxic Substances, Washington, DC. Montgomery, C.D., 2001. Design and Analysis of Experiments, fth ed. John Wiley & Sons, Inc., Arizona, USA. Mulder, R., Vereijken, T.L.F.M., Frijters, C.T.M.J., Vellinga, S.H.J., 2001. Future perspectives in bioreactor development. Water Sci. Technol. 44 (8), 2732. Pham, T., 2002. Anaerobic degradation of aircraft de-icing uid (ADF) in up-ow anaerobic sludge blanket (UASB) reactors and fate of ADF additives. Ph.D. Dissertation, Department of Chemical Engineering, University of Ottawa, Ottawa, ON, Canada. Pilliard, D.A., 1995. Comparative toxicity of formulated glycol deicers and pure ethylene and propylene glycol to Ceriodaphnia dubia and Pimephales promelas. Environ. Toxicol. Chem. 14, 311315. Sabeh, Y., Narasiah, K.S., 1992. Degradation rate of aircraft deicing uid in a sequential biological reactor. Water Sci. Technol. 26, 20612064. Schmidt, J.A., Ahring, B.K., 1996. Granular sludge formation in upow anaerobic sludge blanket (UASB) reactors. Biotech. Bioeng. 49, 229246. Schoenberg, T., Veltman, S., Switzenbaum, M.S., 2001. Kinetics of anaerobic degradation of glycol-based type I aircraft deicing uids. Biodegradation 12, 5968. Simpson, A., 1997. Aircraft deicing best management plans. Transport Canadas presentation at the AWMAs 90th Annual Meeting and Exhibition, June 813, Toronto, Ontario, Canada. Speece, R.E., 1996. Anaerobic Technology for Industrial Wastewaters. Archea Press, Nashville, TN. Switzenbaum, M.S., Veltman, S., Mericas, D., Wagoner, B., Schoenberg, T., 2001. Best management practices for airport deicing stormwater. Chemosphere 43, 1051 1062. Transport Canada, 1994. The clean aircraft concept. Airworthiness Notice No. B017. Van Huyssteen, J.J., 1967. Gas chromatographic separation of anaerobic digester gases using porous polymer. Water Res. 1, 237242. Weiland, P., Rozzi, A., 1991. The start-up, operation and monitoring of high-rate anaerobic treatment systems: discussers report. Water Sci. Technol. 24 (8), 257277. Zitomer, D.H., Tonuk, G.U., 2003. Propylene glycol deicer biodegradation kinetics: anaerobic complete-mix stirred tank reactors, lter, and uidized bed. J. Environ. Eng. 129 (2), 123129.

Please cite this article in press as: Marin, J., et al. Characterization of an anaerobic bafed reactor treating dilute aircraft de-icing uid and long term effects of operation on granular biomass. Bioresour. Technol. (2009), doi:10.1016/j.biortech.2009.11.055