- Email: [email protected]
Anaerobic fixed-film reactors treating carbohydrate wastewater
Wat. Res. Vol. 20, No. 6, pp. 685-695, 1986 Printed in Great Britain. All rights reserved
0043-1354/86 $3.00 +0.00 Copyright © 1986 Pergamon Journals Ltd
ANAEROBIC FIXED-FILM REACTORS TREATING CARBOHYDRATE WASTEWATER K. J. KENNEDYl'* and R. L. DROSTE2 ~Division of Biological Sciences, National Research Council of Canada, Ottawa KIA 0R6 and 2Department of Civil Engineering, University of Ottawa, Ottawa K I N 9B4, Canada
(Received August 1984) A~traet--Anaerobic downflow stationary fixed-film reactors operated at 35°C, successfully treated synthetic (sucrose-based) wastewater of different concentrations at high organic loading rates and short hydraulic residence times. Waste stabilization was due to the high concentration of active biomass retained in the biofilm. Biofilm biomass concentration increased with organic loading rate reaching a maximum of 8.7 kg VFS m -3 of reactor volume (0.112 kg VFS m -2 support surface). The biofilm was found to be completely active and unaffected by diffusional limitations up to an average thickness of 2.6 mm.
Key words--anaerobic, fixed, biofilm, reactor
INTRODUCTION Over the past decade, interest in anaerobic digestion processes has increased dramatically. This has resulted in the development of several novel reactor configurations (Young and McCarty, 1967; Lettinga et ak, 1980; van den Berg and Lentz, 1979; Switzenbaum and Jewell, 1980; Bachman et aL, 1983) that successfully treat low to medium strength wastewaters (50-20,000 mg C O D l- 1 ) having low suspended-solids content at short hydraulic residence times (HRT). All of these reactors have in c o m m o n the ability to immobilize the anaerobic bacteria, especially the slow growing methanogens within the system, resulting in long solid residence times (SRT) that are independent of the H R T . These anaerobic digestion processes accumulate large quantities of microbial biomass, which in turn enables them to efficiently stabilize the organics in wastewater, with low sludge yields. One of these advanced reactor types is the downflow stationary fixed-film ( D S F F ) reactor developed at the laboratories of the National Research Council of Canada. This reactor immobilizes the biomass on oriented support media that form vertical channels running the length of the reactor. The oriented packing combined with the downflow mode of operation allows the D S F F reactor to successfully treat a variety of wastes of widely different organic and suspended solids concentrations at high loading rates and short H R T (Kennedy and van den Berg, 1982a; van den Berg and Kennedy, 1981). Full-scale D S F F reactors have been constructed and are successfully treating screened pig slurry 1-3 % suspended solid (Maat 1984), cheese plant effluent (Samson et al., 1984) and rum mostos (Szendry, 1983). *To whom all correspondence should be addressed. NRCC 25414.
While there have been several reports on the operation and treatability of a variety of wastewaters with D S F F reactors, no information exists on the relationship between waste concentration and D S F F reactor biomass on process efficiency. This is due in part to the difficulty in measuring the biofilm concentration in fixed biomass reactors. One needs to know the relationship between waste stabilization and reactor biomass so that the reactor can be properly characterized with regard to such parameters as active biofilm thickness, specific substrate utilization and biofilm sloughing. This paper reports on a study to evaluate the effects of substrate concentration on steady state D S F F reactor performance. The study includes an assessment of the effect of biofilm concentration on process efficiency. MATERIALS AND METHODS After the DSFF reactors were started and a mature microbial biofilm had developed, the maximum steady-state conversion of substrate to methane, carbon dioxide and biomass was determined. By using different wastewater concentrations and varying the HRT several steady-state organic loading conditions were evaluated. Parameters measured were mixed liquor and effluent total and soluble chemical oxygen demand (COD), volatile-fatty acids (VA), total insoluble biofilm solids (TFS) and insoluble biofilm volatile solids (VFS), biofilm thickness, pH and biogas production and composition.
Reactor operation The effect of substrate concentration on steady state reactor efficiency was conducted with four DSFF reactors. The operation of DSFF reactors has been described in detail in previous papers (van den Berg and Lentz, 1979; van den Berg and Kennedy, 1981). Briefly, feed was pumped into the top of the reactor through a distribution manifold and effluent was withdrawn from the bottom for disposal and recirculation. The liquid level was maintained 5 cm above the top of the biofilm support media to protect the biofilm and to provide a liquid volume for enhanced mixing of 685
686
K . J . KENNEDY and R. L. DROSTE Table 1. Physical characteristics of DSFF reactor
Parameter Height (cm) Inside dimensions (cm) Liquid volume (1.) Empty bed volume (1.) Surface-area-to-volume-ratio (m2 m -3) Voidage (a) Cross-section area of channels (cm2)
98.0 19.0 x 19.0 22.4 24.4 75* 0.92 7.8
*Average value (both sides of support material included) 80 m2 m 3 and removable supports removed 70 m 2 m -3.
recirculated mixed liquor and incoming feed. Recirculation was at a constant rate of 1201 d a y - t . Biogas exited at the top of the reactor through a wet test gas meter used to monitor the rate of biogas production.
Reactor design The reactors used in this study (Table 1) consisted of a vertical tank (19.0 x 19.0 x 98.0cm internal dimensions) with walls made of 1.25 cm-thick plexiglass and having an active liquid volume of 22.41. Each reactor was filled with oriented needle punched polyester (NPP; 3 m m thick; 280 g m -2, white; Texel Inc., Beauce-Nord, Que., Canada) support material which was sewn on to stainless steel wire frames (Fig. 1) to make straight vertical channels 2.8 x 2 . 8 c m square and 61.0cm long supported 25.0cm from the b o t t o m of the reactor. The surface-area-to-reactorvotume ratio of the packing material was 75 m 2 m -3. The N P P material was selected because of the ease of construction, operation and effectiveness of reactors utilizing this support media (van den Berg and Kennedy, 1981). The bottom of the reactor was tapered to one side at an angle of 45 ° to minimize solids accumulation in the bottom of the reactor. Nine removable preweighed film supports each with a surface area (both sides) of 338 cm 2 were placed in the centre of each reactor (Fig. 1). Two of these supports were removed at the conclusion of each steady state operating condition for determination of biofilm concentration and thickness. Each removable film support made up of 2.0% of the total biofilm surface area in the reactor and their removal and immediate replacement with a new support would not significantly affect reactor performance. The surface-area-to-reactor-volume ratio did not take into account the surface area of the reactor walls. Observations indicated that growth on the walls was negligible. The surface-area-to-reactor-volume ratio was chosen based on the surface area optimization studies of van den Berg and Lentz (1979). Surface-area-to-reactor-volume ratio was based on the average ratio with all removable supports in place (80 m 2 m 3) and with all removable supports absent ( 7 0 m 2 m-3). All reactors in this study were located in a temperature controlled room maintained at 35 + I°C.
being added to the reactors (22.41. added). During startup, reactor feed rates were increased while maintaining the VA concentration in the mixed liquor between 200-500 m g 1-1 (acetic acid equivalents). The controlled startup procedure which took 3-4 m o n t h s has been described in detail by Kennedy and Droste 0985). Following startup, each reactor was to be operated at 5 different steady-state H R T ranging between 0.4 and 7.0 days (Table 3). Steady-state data was obtained by operating each reactor for a m i n i m u m of 30 days at each H R T which corresponds to operation for a m i n i m u m of 8-75 H R T at each steady state condition. Complete steady-state evaluation including startup took I 1 months. After 30 days of operation at a particular steady-state condition, samples were taken on 3 successive days and the average of the three data points reported. If a large discrepancy existed a m o n g any of the data points, the steady-state period was extended and reactors resampled.
Analyses Routine reactor performance was assessed by determining influent feed rate, rate of biogas production, biogas composition and VA concentration on a daily basis while C O D and p H of the influent and effluent were determined on a weekly or biweekly basis (three consecutive days during steady state evaluation). C O D was determined by the colorimetric method of Knechtel (1978). Table
2. Composition of (mgl -I)
Sucrose (NH4)HCO 3 Na(HCO3) KHCO 3 (NH4)2SO4 K2HOP 4 KH2PO4 Yeast extract
20.0 4.0 2.0 2.0 1.00 0.52 0.40 0.20
sucrose
10.0 2.0 4.1 4.6 0.50 0.26 0.20 0.10
5.0 1.0 4.2 4.8 0.25 0.13 0.10 0.05
wastes
2.5 0.5 4.4 5.6 0.130 0.065 0.050 0.025
Sugar waste The soluble carbon source was sucrose. Synthetic wastewater was made by diluting a 20,000 m g 1-t stock sucrose and mineral salts solution. Reactors were operated with approximate C O D concentrations of 2500, 5000, 10,000 and 20,000 m g 1-1 (assuming that l g of sucrose equals 1 g COD) (Table 2). Feed was maintained at 4°C and was not preheated before being added to the reactors. To maintain the alkalinity of the feed at 4.7 kg C a C O 3 m -3 additional alkalinity was added to the 0.25, 0.5 and 1.0% (w/v) feeds from an equimolar mixture of K H C O 3 and N a H C O 3.
Experimental plan The reactors were started up with effluent from a completely-mixed reactor acclimated to synthetic sucrose (10,000 m g 1-~) wastewater. To ensure that all reactors were started equally the methanogenic activity of the inoculum was adjusted to 1.5 g acetic acid utilized 1-~ day -t before
Fig. 1. Photograph of N P P support material in D S F F reactor: (A) removable supports; (B) stainless steel support frame; (C) N P P material.
Anaerobic fixed-film reactors treating carbohydrate wastewater
=
i
g~
o ~ ¢ ~
8
t~-~
•
e~
687
688
K.J. KENNEDY and R. L. DROSTE
Biogas composition and VA concentrations were determined by gas chromatography methods. Biogas composition was determined by the method of van Huyssteen (1967) using a Porapak T column (6.35 mm x 3.04.3 cm) on a Hewlett-Packard 5710A gas chromatograph equipped with a model 3380A integrator. The column was held at 70°C and helium flowing at 40 ml min-t was used as the carrier gas (40 ml min- ~). Volatile acids were determined by the method of Ackman (1972) using a Hewlett-Packard 572 IA chromatograph equipped with an automatic sampler, a model 3380A integrator, and a Chromosorb 101 glass column (6.35 mm × 365.76cm). The column was kept at 180°C. Temperature of the flame ionization detector was 350°C. Helium passed over formic acid at a flow rate of 15 mlmin J was used as the carrier gas. Volatile acid samples were prepared by adding 0.5 ml of sample to 0.5 ml of an internal standard that contained 1000 mgl -~ of isobutyric acid. The chromatograph was calibrated with a standard that contained 1000mgl ] each of acetic, propionic, butyric and isobutyric acids. The volume of sample injected (on the column packing) was 0.8/al. On the final day of each steady-state condition (third steady state sampling day) two removable biofilms were taken from each reactor for determination of TFS, VFS and biofilm volume. The average thickness of the biofilm was determined by allowing the biofilm to air dry (35°C) for 5 min, then, by gently scraping the biofilm and associated bound water into a graduate cylinder, its volume was determined. The TFS and VFS were then determined according to Standard Methods (APHA, 1975). At each steady-state COD reactor profiles were determined by sampiing from side ports along the height of the reactors. Suspended biomass was calculated assuming 1.42 g insoluble COD equals 1.0 g VSS. RESULTS
AND
a n a e r o b i c sludge blanket-filter ( U B F ) reactor also reported process instability a n d were unable to achieve significant rates of waste stabilization. Effects o f influent wastewater c o n c e n t r a t i o n on soluble C O D removal efficiency of a n a e r o b i c D S F F reactors (as percentage o f initial COD), c o m p a r e d to H R T a n d organic loading rate are shown in Figs 2 a n d 3. Substrate removal efficiency was a function o f organic loading a n d H R T . F o r D S F F reactors treat-
loo 9o 8o _ 7o *~ 60 -~ 50 o o ~ 40 _o .Q -= 30 2o 10
DISCUSSION
I 2
0
I 4
Reactor performance A n a e r o b i c D S F F reactors successfully treated low a n d m e d i u m strength sucrose wastewater at high organic loading rates a n d short H R T . F o r all D S F F reactors operating at various steady state conditions, acetic, propionic a n d butyric acids were the m a i n acids detected in the effluent. These acids were the m a j o r c o m p o n e n t o f the soluble C O D (Table 3). Periodic checks showed little or no sucrose in the effluent indicating that the balance o f the soluble C O D was either longer chain V A or o t h e r soluble microbial byproducts. D S F F reactors treating 0.25 a n d 0.5% sucrose wastewater were the m o s t stable at all H R T tested. A t H R T d o w n to 0.5 days there was little a c c u m u l a t i o n o f V A a n d pH of the reactors remained a b o v e 6.7. With increasing influent waste c o n c e n t r a t i o n there was a decrease in reactor stability. The reactor treating 2.0% sucrose wastewater was unstable a n d failed at an H R T of < 2 d a y s ( 1 0 k g COD m 3day-l; 0.133kg COD m - 2 d a y J). Instability was characterized by a c c u m u l a t i o n of longer chain V A in the mixed liquor and c o n c o m i t a n t decrease in p H to below 6.3. It has been reported that a c c u m u l a t i o n of propionic and butyric acids indicate that the reactor is being o p e r a t e d u n d e r stress ( K e n n e d y a n d van den Berg, 1982b; K e n n e d y et al., 1984). G u i o t a n d van den Berg (1984) treating the same 2.0% sucrose wastewater with a c o m b i n a t i o n
I 6
I 8
HRT (days)
Fig. 2. Effect of HRT on soluble COD removal: influent COD concentration, ( l l ) 2500 mg i-i, (O) 5000 mg ] i (A) 10,000 mg1-1, ((3) 20,000mgl i too 9o 8o ,~
7o
~ 60 ® ~ 50 40
oJ -~ -6 30 o~ 2o
1o o
I 2
I 4
I 6
I 8
I t0
I 12
I 14
I 16
I 18
Loading rate (kg COD m-3day-1) Fig. 3. Effect of organic loading rate on soluble COD removal: (ll) 2500mgl i, (O) 5000mgl i, (A) 10,000mgl i (O) 20,000mgl t.
Anaerobic fixed-film reactors treating carbohydrate wastewater ing 2.5, 5.0 and 10.0 g 1-L sucrose wastewater soluble COD removal efficiencies were in excess of 75% at HRT as low as 1 day. A decrease in HRT from 1.0 to 0.5 days resulted in a 50% decrease in soluble COD removal efficiency for the three reactors discussed above. At a 2-day HRT the reactor treating 2.0% sucrose had a COD removal efficiency of 56%. However, in terms of organic loading, COD removal efficiency of the reactor treating 2.0% sucrose wastewater was approximately the same as the other reactors at similar organic loading rates. It is important to mention that while the three DSFF reactors treating more dilute wastewaters were stable at all HRT tested, the reactor treating 2.0°/'0 sucrose waste was unstable. High sucrose concentration coupled with slight variations in flow rate often caused elevated VA concentrations (increased acidogenic biomass production) and low pH in the reactor. These conditions resulted in an inbalance between acidogenic and methanogenic bacterial species and concomitant reactor instability. Total COD removal efficiency was only slightly less than soluble COD removal efficiency. This reflected the low concentration of suspended biomass associated with DSFF reactors operated at short HRT. The DSFF reactors in this study were operated at a high recirculation rate to ensure complete mixed (CM) conditions. Figure 4 shows the soluble COD concentration along the height of the DSFF reactor treating 0.5% sucrose wastewater at various HRT. Soluble COD removal was constant at any point in the reactor for steady state conditions tested.
3.0 2.8 2.6 2.4 o-
,-~
2.2 E ~
2.0
~
1.8
~
1.6
5
,
L ~ E ~ 0 ~
1.4 1.2 1.(3
.~ 0.8 2 0.6 0.4
m ~
m
i
~
m
0.2 0
I
I
I
I
I
O. 2
0.4
0.6
0.8
1.0
Relative reoctor
height
(bottom-4-
top)
Fig. 4. Soluble COD removal profile for DSFF reactor treating 5000 mg 1- ~sucrose waste water: (m) 4.0 day HRT, (O) 2.0 day HRT, (0) 1.0 day HRT, (A) 0.7 day HRT, (©) 0.5 day HRT. W.R. 20/6--B
689
Similar types of concentration profiles were obtained for VA concentrations. Insoluble COD reactor profiles also showed no significant suspended solids gradients existed in the mixed liquor and that there was little accumulation of biomass in the bottom of the reactor. There was no excess floating sludge which was characteristic of DSFF reactors started up with sucrose wastewater at a concentration > 1.0% (Kennedy and Droste, 1985). These results showed the CM behavior of the system and supported DSFF reactor mixing studies previously reported and discussed (Kennedy and Droste, 1985). The CM regime in the DSFF reactor contrasts with the plug flow (PF) pattern reported for conventional random packed upflow anaerobic filters. In conventional filters the majority of the waste stabilization occurs in the bottom one-third of the reactor where most of the active biomass in the system is found (Young and McCarty, 1967). Biomass characteristics
Suspended biomass made up a small percentage of the total biomass within DSFF reactors (Table 3). An average value of 8.6% of all biomass in DSFF reactors was suspended biomass. In general, suspended biomass increased with increased waste strength and longer HRT. For DSFF reactors treating 0.25 and 2.0% sucrose wastewater, average VSS component of total reactor biomass was < 5% and >20%, respectively. Suspended solids concentration in the effluent is often a measure of effluent quality. Table 3 gives effluent suspended solids for influent waste concentrations as a function of HRT and organic loading rate, respectively. Concentration of suspended solids in the effluent increased only slightly for large decreases in HRT and concomitant increased organic loading rates. Most effluent suspended solids concentrations were < 1.8 kg COD m 3 (1.4 kg VSS m 3). Low biomass suspended solids even at low HRT indicate that microbial washout is not a major problem. Periodic checks of mixed liquor acidogenic and methanogenic activity (Kennedy and Droste, 1985) showed little or no methanogenic activity for the majority of steady-state conditions. Some methanogenic activity was found in the mixed liquor of the DSFF reactor treating 2.0% sucrose at a 7 day HRT; this was not surprising since certain methanogenic bacteria have doubling times of less than 7 days (Wandrey and Aivasidis, 1986). Acidogenic activity was always found in the mixed liquor and varied between 2-8 g COD (g VSS)- i day-I. Low concentrations of suspended solids (biomass) and absence of methanogenic activity in the mixed liquor not only reflects the small biomass yield but also the importance of the biofilm for waste stabilization. The presence of >90% of DSFF reactor methanogenic activity and the majority of the acidogenic activity in the biofilm of DSFF reactors operating at short HRT
690
K.J. K~NNEDVand R. L. DROSTE
during startup has also been reported by Kennedy and Droste (1985). For D S F F reactors treating concentrated wastewater at longer H R T the higher concentrations of suspended biomass (with acidogenic and methanogenic activity) shows that the D S F F reactor is beginning to operate more like a CM suspended growth reactor rather than a fixed biofilm reactor. This would mean that for high strength wastewaters ( > 20,000 mg COD 1- ' ) and long H R T there may not be a significant advantage to using a fixed-film anaerobic process. The role of the microbial film in conventional upflow anaerobic filters is not as important as in D S F F reactors. Dahab and Young (1982) have reported that in anaerobic filters the majority of the microbial activity (biomass) is located in the interstitial spaces between the support media in the lower third of the reactor and not attached to the media. In fact the anaerobic filter behaves more like an upflow anaerobic sludge blanket with media than an attached biomass reactor.
Biofilm characteristics The zoogleal biofilm (Atkinson and Fowler, 1974) that developed on the NPP support material was relatively porous and filamentous with tentacle like appendages extending up to 1 cm into the mixed liquor surrounding the biofilm. With the majority of the D S F F reactors methanogenic activity in the biofilm, biogas evolution and associated micromixing was intense (Fig. 5) in relation to the long residence time of liquid in the reactor and slow reaction rates of anaerobic microbes. The formation of gas bubbles is readily observed in Fig. 5. Biogas production in individual channels has been reported to be in part responsible for a gas lift mixing effect in D S F F reactors (Hall, 1982). Microscopic observation of
D S F F biofilms (Fig. 6, magnification 57 x ) showed microbes attached both to individual polyester fibres and entrapped by the NPP support material. Further magnification of Fig. 6 (magnification 1452 x ) shows bacteria characteristic of the genus Methanosarcina entrapped within the biofilm. The biofilms in all reactors were similar in that they were 5-6% (w/v of biofilm) TFS of which 75-80% were VFS.
Biofilm performance The relationship between organic loading rate and D S F F reactor biofilm biomass is shown in Fig. 7. Biofilm biomass concentration (in terms of reactor liquid volume) increased linearly with increased organic loading rates up to 10kg C O D m - 3 d a y -1 (0.133kg COD m - 2 d a y -1) and was relatively unaffected by influent substrate concentration. At a loading rate of 10 kg COD m - ~ d a y -l a maximum biofilm concentration and maximum biofilm thickness of 8.7kg VFS m -3 (0.116kg VFS m -2) and 2.6 mm, respectively, were obtained. At steady-state loading rates higher than 10kg COD m - 3 d a y 1, biofilm concentration levels off or at least the increase was very slow. Biofilm development during steadystate D S F F reactor operation differed from that found during reactor startup. During startup, biofilm development was found to decrease with increased substrate concentration for reactors operated at similar organic loading and removal rates. This difference in biofilm development is in part due to the startup protocol which limited VA accumulation in the reactor. Figure 8 shows biofilm concentration for various steady-state organic removal rates. Maximum biofilm concentration or a very tow rate of increase in biofilm biomass concentration was achieved at a steady state organic removal rate of 7.5kg COD m - 3 d a y -1
Fig. 5. Biogas bubble formation on biofilm.
Anaerobic fixed-film reactors treating carbohydrate wastewater
691
Fig. 6. Attached and entrapped biomass in the biofilm: (A) NPP fibre; (B) attached biomass; (C) entrapped biomass.
(0.100kg C O D m - 2 d a y - l ) . This indicates that an equilibrium of sorts was reached between biofilm and mixed liquor growth, attachment and biofilm sloughing. Recent studies have indicated that when the liquid velocity in reactor channels is sufficient no significant plugging should occur (Samson and Kennedy, 1985). If this is in fact the case, eventual plugging of the reactor channels should not occur. The D S F F reactor biofilm (biomass) concentration was less than reported for other advanced anaerobic reactors treating similar wastes at 35°C. Biomass concentrations in anaerobic/expanded beds (Switzenbaum and Jewell, 1980), anaerobic filters (Dahab and Young, 1982), upflow anaerobic sludge blankets (Lettinga et al., 1980) and combination upflow anaerobic blanket-filters (Guiot and van den
Berg, 1984) of 40, 25, 30 and 30kg VSSm 3, respectively, have been reported. Assuming that bioimass of similar specific activity develops in each type of reactor (Table 4), the maximum waste stabilization capacity of D S F F reactors would be less than other advanced anaerobic reactors. However, biomass may still be accumulating in the D S F F reactor at low rate, and over an extended period of reactor operation the biofilm concentration may increase resulting in a higher waste stabilization capacity.
Biofilm diffusion phenomena Figure 9 shows the effect of HRT on biofilm biomass concentration at various steady state conditions. Reactor biofilm concentration and concomitant biofilm thickness increased with decreasing HRT
692
K.J. KENNEDYand R. L. DROSTE 83% for DSFF reactors treating 2.5, 5.0 and 10 g 1-1 sucrose wastewater, respectively. At loading rates greater than 1.0 kg COD kg-] biofilm VFS dayCOD removal efficiency decreased rapidly. Figure 11 shows that the specific biofilm substrate utilization rate increased with increased loading rate reaching a maximum specific utilization rate of 0.9-1.2 kg COD kg -l biofilm VFS day -l at loading rates greater than 5 k g C O D m -3day -~ (0.066 kg COD m-2day-t). Figure 12 shows that the specific biofilm substrate utilization rate increased with decreased HRT and increased waste strength. Henze and Harremoes (1982) reported that anaerobic biofilms may become diffusion limited at biofilm thicknesses > 1 mm. Williamson and McCarty (1976) on the other hand reported that anaerobic biofilms would not be diffusion limited at any thickness.
10 .===---A 8 rn II= 7 tO >
g
g
5
E 3 .2 m 2
yof
4 0
I
I
I
I
I
I
I
I
I
I
2
4
6
8
10
12
14
16
18
20
COD
loading rate (kg
10
m -3 day - 1 )
9
Fig. 7. Effect of organic loading rate on biofilm biomass concentration: influent COD concentration, (11) 2500mgl -t, ( O ) 5000mgl -I, ( A ) 10,000mgl -], ( O ) 20,000 mg 1- t.
and increased waste strength (for the ranges studied). This behavior is similar to results reported for attached biomass in anaerobic expanded beds (Switzenbaum and Jewell, 1980). The steady state relationship between COD removal efficiency and specific biofilm substrate loading rate is shown in Fig. 10. COD balances to be reported in a future paper confirmed the reliability of the results. Because of the low concentration of suspended biomass with little or no methanogenic activity, the DSFF reactors specific loading rates are expressed in terms of biofilm biomass. COD removal efficiency decreased with increased specific biofilm substrate loading rate and was relatively unaffected by waste concentration. At a specific biofilm substrate loading rate of 1.0kg COD kg -t biofilm VFS day -~ COD removal efficiency was 80, 79 and
f
&
8
'E 7 to > 6 5
,
/
o
2
I 2 COD
I 4 removal
I 6
I 8
I 10
I 12
rate (kq m-3 day-~)
Fig. 8. Effect of organic removal rate on biofilm biomass concentration: influent COD concentration, (i) 2500mgl -I, ( O ) 5000mg1-1, ( A ) 10,000mgl -I, ((3) 20,000 mg 1- i
Table 4. Comparison of specific utilization rate for anaerobic reactors treating carbohydrate wastewaters
Wastewater
Reactor*
Empty bed volume (m3)
Operating temperature (°C)
Specific rate ( g C O D g I VSSday-I)
Sucrose (synthetic)
DSFF; (FG)
24.4 x I0 3
35
0.9-1.2
This study
Sucrose (synthetic)
Upflow sludge blanket filter; (SG)
4.3 x 10 3
27
1.0-1.2
Guiot and van den Berg (1984)
Dextrose/protein
Upflow filter; (FSG)
160 × 10 3
35
1.0
Mueller and Mancini (1975)
Sugar beets (soured)
Upflow sludge blanket; (SG)
18 x 10-3
30
0.6-0.9
Lettinga et at'. (1980)
Sucrose (synthetic)
Expanded bed; (FG)
1.0 x 10 3
30
0.44).8
Switzenbaum and Jewell (1979)
*(FG)--fixed growth; (SG)--suspended growth; (FSG)--fixed and suspended growth.
Reference
Anaerobic fixed-film reactors treating carbohydrate wastewater
693
2.0
10
1.9 t.8
9
• 8
•
IE
7
o3 >
6
g
1.7
O3 >
1.6 1.5 1.4 1.3
0
1.2 1,1
5
1.0 9
0,9
4
0,8
3
0,6
0,7
g
0,5
E
o 0.4 / 0.3 @) 0.2 I 0.1
2
_o ill3
1
0
I
I
I
I
2
4
6
8
I
0
2
H R T (doys)
I
I
I
I
1
I
I
I
4
6
8
10
12
14
16
18
COD
Iooding
rote
(kg
m - 3 d o y -1)
Fig. 9. Effect of HRT on biofilm biomass concentration: influent COD concentration, (11) 2500mgl -I, ( 0 ) 5000 mg 1-1, (&) 10,000 mg 1- i, (©) 20,000 mg 1- J.
Fig. 11. Effect of organic loading rate on specific biofilm substrate utilization rate: influent COD concentration, (m) 2500mgl -j, ( 0 ) 5000mgl -t, (A) 10,000mgl -I, (O) 20,000 mg 1- i.
Figure 13 shows the relationship between biofilm thickness and specific biofilm substrate removal rate. For the four waste concentrations studied, specific biofilm substrate utilization rate increased with biofilm thickness reaching a m a x i m u m at a biofilm thickness of 0.9 ram. With increased biofilm thickness, up to 2.6 m m , the specific biofilm substrate
removal rate remained relatively constant. This indicated that the whole biofilm was active for thicknesses up to 2 . 6 m m and was not diffusion limited. If the biofilm was not completely active or if a diffusion limitation existed, there would have been a decrease in the specific biofilm activity with increased biofilm depth. Table 4 compares the biofilm
100
90
2.0
F
A I>,
80 tO > 70
7
1.9 1.8 1.7 1.6 1.5
1.4
~
1.3
"~ 60
a ~
1.2 1.1
o 50 a 0~ 40
~
1.0
i
0.8 0.9 0.7 I
•
.1o ~ 30
L
o o.6 ~ 0.5 O.4 ~- 0.3 ~ 0.2 0.1
g,
2010-
0
\
I 0.4 Specific
I 0.8 Iooding
I l.Z
I t.6 rote
I 2.0
[k O COD
I 2.4
I 2.8
kg - 1 V S doy-1)
Fig. I0. Effect of specific biofilm loading rate on soluble COD removal: influent COD concentration, ( , ) 2500mgl -I, ( 0 ) 5000mgl -I, (&) 10,000mgl -I, (C)) 20,000 mg 1-1.
0
I
I
I
I
2
4 HRT (doys)
6
8
Fig. 12. Effect of HRT on specific biofilm substrate utilization rate: influent COD concentration, (m) 2500 mg l-=, (O) 5000 mg 1-1, (&) 10,000 mg I-I, (O) 20,000 mg 1-t.
694
K . J . KENNEDY and R. L. DROSTE
(4) The majority of the acidogenic biomass and methanogenic biomass activity in D S F F reactors was located in the biofilm. (5) Maximum biofilm substrate utilization rates (0.9-1.2kg COD kg -~ V F S d a y - l ) were similar to rates reported for suspended growth cultures treating carbohydrate wastewater. (6) Anaerobic biofilms are completely active and not diffusion limited up to biofilm thicknesses of 2.6 mm.
2,0 1.9
1.8
~ if)
> ?
121 0
~ o ~
1.7 1.6 1.5 1.4
1.3 1.2 1.1 1.0
o. 7 0
0.9 0.8
~ 0.7 E 06 0.5 ~ 0.4
0 REFERENCES
•
=• *
Ackman R. G. (1972) Porous polymer bead packings and formic acid vapor in GLC of volatile fatty acids. J. Chromat. Sci. 10, 560-565. APHA (1975) Standard Methods for the Examination of ~= 0 3 Water and Waste Water Including Bottom Sediments and t/) 0.2 Sludges, 14th edition. American Public Health Association, New York. 0.1 I I I I I I Atkinson B. and Fowler H. W. (1974) Significance of 0 0.04 0.08 0.12 0.16 0.20 0.24 microbial films in fermenters. Adv. Biochem. Engng 3, Biofilrn thickness (crn) 221-277. Bachman A., Beard V. L. and McCarty P. L. (1983) Fig. 13. Effect o f biofilm thickness on specific biofilm Performance characteristics of the anaerobic baffled reacsubstrate utilization rate: influent COD concentration, ( I ) tor. 3rd International Symposium on Anaerobic Digestion, 2500mgl -I, ( 0 ) 5000mgl -I, (A) 10,000mgl -I, (O) Boston, Mass. 20,000 mg 1- t. Berg L. van den and Kennedy K. J. (1981) Support materials for stationary fixed film reactors for high rate methanogenic fermentations. Biotechnol. Lett. 3, 165-170. substrate utilization rate to other anaerobic systems Berg L. van den and Lentz C. P. (1979) Comparison treating carbohydrate wastes and shows that there between up- and down-flow anaerobic fixed film reactors was little difference between suspended and attached of varying surface-to-volume ratios for the treatment of biofilm biomass. The porous nature of the anaerobic bean blanching waste. Proc. 35th Ind. Waste Conf. Purdue Univ., pp. 319-325. biofilm with its tentacle like appendages (increased surface area for microbial/substrate contact), the lack Dahab M. F. and Young J. C. (1982) Retention and distribution of biological solids in fixed bed anaerobic of an electron donor and high Ks values of anaerobic filters. 1st International Conference o f Fixed Film Biologibacteria (low affinity for substrate; Williamson and cal Processes, Kings Island, Ohio. McCarty, 1976), biofilm micromixing caused by bio- Guiot S. and Berg L. van den (1984) Dynamic performance of an anaerobic reactor combining an upflow sludge gas production and the relatively long residence time blanket and a filter for the treatment of sugar waste. Proc. of substrate in contact with the biofilm (minimum 39th Ind. Waste Conf. Purdue Univ. pp. 705-717. 12 h) all may be responsible for minimizing diffusion Hall E. R. Biomass retention and mixing characteristics in problems and increasing the active depth of anaerfixed film and suspended growth anaerobic reactors. IA WPR Specialized Seminar on Anaerobic Treatment in obic biofilms. It is also evident that high substrate Fixed Film Reactors, Copenhagen, Denmark, pp. removal rates in D S F F reactors were directly related 371-398. to the amount of biomass that could be retained in Henze M. and Harremoes P. (1982) Anaerobic treatment of the reactor's biofilm. waste water in fixed film reactors. IA WPR Specialized Seminar on Anaerobic Treatment in Fixed Film Reactors, Copenhagen, Denmark, pp. 1-94. CONCLUSIONS Huyssteen J. J. van (1967) Gas chromatographic separation of anaerobic digester gases using porous polymer. Wat. (1) The high concentration of biomass retained in Res. 1, 237-242. D S F F reactors permits successful treatment of carboKennedy K. J. and Berg L. van den (1982a) Anaerobic hydrate wastewater at high organic loading rates and digestion of piggery waste using a stationary fixed film short HRT. reactor. Agric. Wastes 4, 151-158. (2) D S F F reactors do not retain as much biomass Kennedy K. J. and Berg L. van den (1982b) Stability and performance of anaerobic fixed film reactors during as other advanced reactors and cannot achieve as hydraulic overloading at 10-35°C. Wat. Res. 16, high organic loading rates. 1391-1398. (3) Steady-state biofilm concentration for D S F F Kennedy K. J. and Droste R. L. (1985) Startup of anaerobic reactors treating a carbohydrate wastewater of differdownflow stationary fixed film (DSFF) reactors. Biotechnol. Bioengng 27, 1152-1165. ent concentrations increased with increased organic loading rate and decreased H R T reaching a maxi- Kennedy K. J., Muzar M. and Copp G. (1985) Stability and performance of mesophilic anaerobic fixed film reactors mum of 8.7kg V F S m -3 (0.116kg V F S m -2) at a during organic overloading. Biotechnol. Bioengng 27, loading rate of 10kg C O D m - 3 d a y -~ (0.133kg 86-93. Knechtel J. R. (1978) A more economical method for the COD m-2 d a y - l ). -
D
/
Anaerobic fixed-film reactors treating carbohydrate wastewater determination of Chemical Oxygen Demand. Wat. Pollut. Control 116, No. 5, 25. Lettinga G., Velsen A. van, Hobma W., Zeeuw W. de and Klapwijk A. (1980) Use of the upflow sludge blanket (USB) reactor concept for biological waste water treatment, especially for anaerobic treatment. Biotechnol. Bioengng 22, 699-734. Maat D. (1984) Design, construction and operation of a full scale downflow fixed film reactor using hog waste substrate. 5th Bioenergy R&D Seminar, Ottawa, Canada. Mueller J. A. and Mancini J, L. (1975) Anaerobic filter-kinetics and application. Proc. 30th Ind. Waste Conf., Purdue Univ. pp. 423-447. Samson R. and Kennedy K. J. (1985) Effect of reactor height on mixing characteristics and performance of the anaerobic downflow stationary fixed film (DSFF) reactor. J. Biotechnol. 2, 95-106. Samson R., Peters R., Hade C. and Berg L. van den (1984) Dairy waste treatment using industrial scale fixed film and
695
upflow sludge bed anaerobic digesters: design and start-up experience. Proc. 39th Ind. Waste Conf., Purdue Univ. pp. 14~49. Switzenbaum M. S. and Jewell W. J. (1980) Anaerobic attached film expanded bed reactor treatment. J. Wat. Pollut. Control Fed. 52, 1953. Szendry M. L. (1983) Bacardi Corporation digestion process for stabilization of rum distillery wastes and production of methane. Institute o f Gas Technology: Energy from Biomass, Lake Beauna Vista, Fla. Wandrey C. and Aivasidis A. (1986) Continuous anaerobic digestion with Methanosarcina barkeri. Ann. N.Y. Acad. Sci. In press. Williamson K. and McCarty P. L. (1976) Model of substrate utilization by bacterial films. J. Wat. Pollut. Control. Fed. 48, 9-24. Young J. and McCarty P. L. (1967) The anaerobic filter for waste treatment. Proc. 22nd Ind. Waste Conf. Purdue Univ. pp. 559-574.
0043-1354/86 $3.00 +0.00 Copyright © 1986 Pergamon Journals Ltd
ANAEROBIC FIXED-FILM REACTORS TREATING CARBOHYDRATE WASTEWATER K. J. KENNEDYl'* and R. L. DROSTE2 ~Division of Biological Sciences, National Research Council of Canada, Ottawa KIA 0R6 and 2Department of Civil Engineering, University of Ottawa, Ottawa K I N 9B4, Canada
(Received August 1984) A~traet--Anaerobic downflow stationary fixed-film reactors operated at 35°C, successfully treated synthetic (sucrose-based) wastewater of different concentrations at high organic loading rates and short hydraulic residence times. Waste stabilization was due to the high concentration of active biomass retained in the biofilm. Biofilm biomass concentration increased with organic loading rate reaching a maximum of 8.7 kg VFS m -3 of reactor volume (0.112 kg VFS m -2 support surface). The biofilm was found to be completely active and unaffected by diffusional limitations up to an average thickness of 2.6 mm.
Key words--anaerobic, fixed, biofilm, reactor
INTRODUCTION Over the past decade, interest in anaerobic digestion processes has increased dramatically. This has resulted in the development of several novel reactor configurations (Young and McCarty, 1967; Lettinga et ak, 1980; van den Berg and Lentz, 1979; Switzenbaum and Jewell, 1980; Bachman et aL, 1983) that successfully treat low to medium strength wastewaters (50-20,000 mg C O D l- 1 ) having low suspended-solids content at short hydraulic residence times (HRT). All of these reactors have in c o m m o n the ability to immobilize the anaerobic bacteria, especially the slow growing methanogens within the system, resulting in long solid residence times (SRT) that are independent of the H R T . These anaerobic digestion processes accumulate large quantities of microbial biomass, which in turn enables them to efficiently stabilize the organics in wastewater, with low sludge yields. One of these advanced reactor types is the downflow stationary fixed-film ( D S F F ) reactor developed at the laboratories of the National Research Council of Canada. This reactor immobilizes the biomass on oriented support media that form vertical channels running the length of the reactor. The oriented packing combined with the downflow mode of operation allows the D S F F reactor to successfully treat a variety of wastes of widely different organic and suspended solids concentrations at high loading rates and short H R T (Kennedy and van den Berg, 1982a; van den Berg and Kennedy, 1981). Full-scale D S F F reactors have been constructed and are successfully treating screened pig slurry 1-3 % suspended solid (Maat 1984), cheese plant effluent (Samson et al., 1984) and rum mostos (Szendry, 1983). *To whom all correspondence should be addressed. NRCC 25414.
While there have been several reports on the operation and treatability of a variety of wastewaters with D S F F reactors, no information exists on the relationship between waste concentration and D S F F reactor biomass on process efficiency. This is due in part to the difficulty in measuring the biofilm concentration in fixed biomass reactors. One needs to know the relationship between waste stabilization and reactor biomass so that the reactor can be properly characterized with regard to such parameters as active biofilm thickness, specific substrate utilization and biofilm sloughing. This paper reports on a study to evaluate the effects of substrate concentration on steady state D S F F reactor performance. The study includes an assessment of the effect of biofilm concentration on process efficiency. MATERIALS AND METHODS After the DSFF reactors were started and a mature microbial biofilm had developed, the maximum steady-state conversion of substrate to methane, carbon dioxide and biomass was determined. By using different wastewater concentrations and varying the HRT several steady-state organic loading conditions were evaluated. Parameters measured were mixed liquor and effluent total and soluble chemical oxygen demand (COD), volatile-fatty acids (VA), total insoluble biofilm solids (TFS) and insoluble biofilm volatile solids (VFS), biofilm thickness, pH and biogas production and composition.
Reactor operation The effect of substrate concentration on steady state reactor efficiency was conducted with four DSFF reactors. The operation of DSFF reactors has been described in detail in previous papers (van den Berg and Lentz, 1979; van den Berg and Kennedy, 1981). Briefly, feed was pumped into the top of the reactor through a distribution manifold and effluent was withdrawn from the bottom for disposal and recirculation. The liquid level was maintained 5 cm above the top of the biofilm support media to protect the biofilm and to provide a liquid volume for enhanced mixing of 685
686
K . J . KENNEDY and R. L. DROSTE Table 1. Physical characteristics of DSFF reactor
Parameter Height (cm) Inside dimensions (cm) Liquid volume (1.) Empty bed volume (1.) Surface-area-to-volume-ratio (m2 m -3) Voidage (a) Cross-section area of channels (cm2)
98.0 19.0 x 19.0 22.4 24.4 75* 0.92 7.8
*Average value (both sides of support material included) 80 m2 m 3 and removable supports removed 70 m 2 m -3.
recirculated mixed liquor and incoming feed. Recirculation was at a constant rate of 1201 d a y - t . Biogas exited at the top of the reactor through a wet test gas meter used to monitor the rate of biogas production.
Reactor design The reactors used in this study (Table 1) consisted of a vertical tank (19.0 x 19.0 x 98.0cm internal dimensions) with walls made of 1.25 cm-thick plexiglass and having an active liquid volume of 22.41. Each reactor was filled with oriented needle punched polyester (NPP; 3 m m thick; 280 g m -2, white; Texel Inc., Beauce-Nord, Que., Canada) support material which was sewn on to stainless steel wire frames (Fig. 1) to make straight vertical channels 2.8 x 2 . 8 c m square and 61.0cm long supported 25.0cm from the b o t t o m of the reactor. The surface-area-to-reactorvotume ratio of the packing material was 75 m 2 m -3. The N P P material was selected because of the ease of construction, operation and effectiveness of reactors utilizing this support media (van den Berg and Kennedy, 1981). The bottom of the reactor was tapered to one side at an angle of 45 ° to minimize solids accumulation in the bottom of the reactor. Nine removable preweighed film supports each with a surface area (both sides) of 338 cm 2 were placed in the centre of each reactor (Fig. 1). Two of these supports were removed at the conclusion of each steady state operating condition for determination of biofilm concentration and thickness. Each removable film support made up of 2.0% of the total biofilm surface area in the reactor and their removal and immediate replacement with a new support would not significantly affect reactor performance. The surface-area-to-reactor-volume ratio did not take into account the surface area of the reactor walls. Observations indicated that growth on the walls was negligible. The surface-area-to-reactor-volume ratio was chosen based on the surface area optimization studies of van den Berg and Lentz (1979). Surface-area-to-reactor-volume ratio was based on the average ratio with all removable supports in place (80 m 2 m 3) and with all removable supports absent ( 7 0 m 2 m-3). All reactors in this study were located in a temperature controlled room maintained at 35 + I°C.
being added to the reactors (22.41. added). During startup, reactor feed rates were increased while maintaining the VA concentration in the mixed liquor between 200-500 m g 1-1 (acetic acid equivalents). The controlled startup procedure which took 3-4 m o n t h s has been described in detail by Kennedy and Droste 0985). Following startup, each reactor was to be operated at 5 different steady-state H R T ranging between 0.4 and 7.0 days (Table 3). Steady-state data was obtained by operating each reactor for a m i n i m u m of 30 days at each H R T which corresponds to operation for a m i n i m u m of 8-75 H R T at each steady state condition. Complete steady-state evaluation including startup took I 1 months. After 30 days of operation at a particular steady-state condition, samples were taken on 3 successive days and the average of the three data points reported. If a large discrepancy existed a m o n g any of the data points, the steady-state period was extended and reactors resampled.
Analyses Routine reactor performance was assessed by determining influent feed rate, rate of biogas production, biogas composition and VA concentration on a daily basis while C O D and p H of the influent and effluent were determined on a weekly or biweekly basis (three consecutive days during steady state evaluation). C O D was determined by the colorimetric method of Knechtel (1978). Table
2. Composition of (mgl -I)
Sucrose (NH4)HCO 3 Na(HCO3) KHCO 3 (NH4)2SO4 K2HOP 4 KH2PO4 Yeast extract
20.0 4.0 2.0 2.0 1.00 0.52 0.40 0.20
sucrose
10.0 2.0 4.1 4.6 0.50 0.26 0.20 0.10
5.0 1.0 4.2 4.8 0.25 0.13 0.10 0.05
wastes
2.5 0.5 4.4 5.6 0.130 0.065 0.050 0.025
Sugar waste The soluble carbon source was sucrose. Synthetic wastewater was made by diluting a 20,000 m g 1-t stock sucrose and mineral salts solution. Reactors were operated with approximate C O D concentrations of 2500, 5000, 10,000 and 20,000 m g 1-1 (assuming that l g of sucrose equals 1 g COD) (Table 2). Feed was maintained at 4°C and was not preheated before being added to the reactors. To maintain the alkalinity of the feed at 4.7 kg C a C O 3 m -3 additional alkalinity was added to the 0.25, 0.5 and 1.0% (w/v) feeds from an equimolar mixture of K H C O 3 and N a H C O 3.
Experimental plan The reactors were started up with effluent from a completely-mixed reactor acclimated to synthetic sucrose (10,000 m g 1-~) wastewater. To ensure that all reactors were started equally the methanogenic activity of the inoculum was adjusted to 1.5 g acetic acid utilized 1-~ day -t before
Fig. 1. Photograph of N P P support material in D S F F reactor: (A) removable supports; (B) stainless steel support frame; (C) N P P material.
Anaerobic fixed-film reactors treating carbohydrate wastewater
=
i
g~
o ~ ¢ ~
8
t~-~
•
e~
687
688
K.J. KENNEDY and R. L. DROSTE
Biogas composition and VA concentrations were determined by gas chromatography methods. Biogas composition was determined by the method of van Huyssteen (1967) using a Porapak T column (6.35 mm x 3.04.3 cm) on a Hewlett-Packard 5710A gas chromatograph equipped with a model 3380A integrator. The column was held at 70°C and helium flowing at 40 ml min-t was used as the carrier gas (40 ml min- ~). Volatile acids were determined by the method of Ackman (1972) using a Hewlett-Packard 572 IA chromatograph equipped with an automatic sampler, a model 3380A integrator, and a Chromosorb 101 glass column (6.35 mm × 365.76cm). The column was kept at 180°C. Temperature of the flame ionization detector was 350°C. Helium passed over formic acid at a flow rate of 15 mlmin J was used as the carrier gas. Volatile acid samples were prepared by adding 0.5 ml of sample to 0.5 ml of an internal standard that contained 1000 mgl -~ of isobutyric acid. The chromatograph was calibrated with a standard that contained 1000mgl ] each of acetic, propionic, butyric and isobutyric acids. The volume of sample injected (on the column packing) was 0.8/al. On the final day of each steady-state condition (third steady state sampling day) two removable biofilms were taken from each reactor for determination of TFS, VFS and biofilm volume. The average thickness of the biofilm was determined by allowing the biofilm to air dry (35°C) for 5 min, then, by gently scraping the biofilm and associated bound water into a graduate cylinder, its volume was determined. The TFS and VFS were then determined according to Standard Methods (APHA, 1975). At each steady-state COD reactor profiles were determined by sampiing from side ports along the height of the reactors. Suspended biomass was calculated assuming 1.42 g insoluble COD equals 1.0 g VSS. RESULTS
AND
a n a e r o b i c sludge blanket-filter ( U B F ) reactor also reported process instability a n d were unable to achieve significant rates of waste stabilization. Effects o f influent wastewater c o n c e n t r a t i o n on soluble C O D removal efficiency of a n a e r o b i c D S F F reactors (as percentage o f initial COD), c o m p a r e d to H R T a n d organic loading rate are shown in Figs 2 a n d 3. Substrate removal efficiency was a function o f organic loading a n d H R T . F o r D S F F reactors treat-
loo 9o 8o _ 7o *~ 60 -~ 50 o o ~ 40 _o .Q -= 30 2o 10
DISCUSSION
I 2
0
I 4
Reactor performance A n a e r o b i c D S F F reactors successfully treated low a n d m e d i u m strength sucrose wastewater at high organic loading rates a n d short H R T . F o r all D S F F reactors operating at various steady state conditions, acetic, propionic a n d butyric acids were the m a i n acids detected in the effluent. These acids were the m a j o r c o m p o n e n t o f the soluble C O D (Table 3). Periodic checks showed little or no sucrose in the effluent indicating that the balance o f the soluble C O D was either longer chain V A or o t h e r soluble microbial byproducts. D S F F reactors treating 0.25 a n d 0.5% sucrose wastewater were the m o s t stable at all H R T tested. A t H R T d o w n to 0.5 days there was little a c c u m u l a t i o n o f V A a n d pH of the reactors remained a b o v e 6.7. With increasing influent waste c o n c e n t r a t i o n there was a decrease in reactor stability. The reactor treating 2.0% sucrose wastewater was unstable a n d failed at an H R T of < 2 d a y s ( 1 0 k g COD m 3day-l; 0.133kg COD m - 2 d a y J). Instability was characterized by a c c u m u l a t i o n of longer chain V A in the mixed liquor and c o n c o m i t a n t decrease in p H to below 6.3. It has been reported that a c c u m u l a t i o n of propionic and butyric acids indicate that the reactor is being o p e r a t e d u n d e r stress ( K e n n e d y a n d van den Berg, 1982b; K e n n e d y et al., 1984). G u i o t a n d van den Berg (1984) treating the same 2.0% sucrose wastewater with a c o m b i n a t i o n
I 6
I 8
HRT (days)
Fig. 2. Effect of HRT on soluble COD removal: influent COD concentration, ( l l ) 2500 mg i-i, (O) 5000 mg ] i (A) 10,000 mg1-1, ((3) 20,000mgl i too 9o 8o ,~
7o
~ 60 ® ~ 50 40
oJ -~ -6 30 o~ 2o
1o o
I 2
I 4
I 6
I 8
I t0
I 12
I 14
I 16
I 18
Loading rate (kg COD m-3day-1) Fig. 3. Effect of organic loading rate on soluble COD removal: (ll) 2500mgl i, (O) 5000mgl i, (A) 10,000mgl i (O) 20,000mgl t.
Anaerobic fixed-film reactors treating carbohydrate wastewater ing 2.5, 5.0 and 10.0 g 1-L sucrose wastewater soluble COD removal efficiencies were in excess of 75% at HRT as low as 1 day. A decrease in HRT from 1.0 to 0.5 days resulted in a 50% decrease in soluble COD removal efficiency for the three reactors discussed above. At a 2-day HRT the reactor treating 2.0% sucrose had a COD removal efficiency of 56%. However, in terms of organic loading, COD removal efficiency of the reactor treating 2.0% sucrose wastewater was approximately the same as the other reactors at similar organic loading rates. It is important to mention that while the three DSFF reactors treating more dilute wastewaters were stable at all HRT tested, the reactor treating 2.0°/'0 sucrose waste was unstable. High sucrose concentration coupled with slight variations in flow rate often caused elevated VA concentrations (increased acidogenic biomass production) and low pH in the reactor. These conditions resulted in an inbalance between acidogenic and methanogenic bacterial species and concomitant reactor instability. Total COD removal efficiency was only slightly less than soluble COD removal efficiency. This reflected the low concentration of suspended biomass associated with DSFF reactors operated at short HRT. The DSFF reactors in this study were operated at a high recirculation rate to ensure complete mixed (CM) conditions. Figure 4 shows the soluble COD concentration along the height of the DSFF reactor treating 0.5% sucrose wastewater at various HRT. Soluble COD removal was constant at any point in the reactor for steady state conditions tested.
3.0 2.8 2.6 2.4 o-
,-~
2.2 E ~
2.0
~
1.8
~
1.6
5
,
L ~ E ~ 0 ~
1.4 1.2 1.(3
.~ 0.8 2 0.6 0.4
m ~
m
i
~
m
0.2 0
I
I
I
I
I
O. 2
0.4
0.6
0.8
1.0
Relative reoctor
height
(bottom-4-
top)
Fig. 4. Soluble COD removal profile for DSFF reactor treating 5000 mg 1- ~sucrose waste water: (m) 4.0 day HRT, (O) 2.0 day HRT, (0) 1.0 day HRT, (A) 0.7 day HRT, (©) 0.5 day HRT. W.R. 20/6--B
689
Similar types of concentration profiles were obtained for VA concentrations. Insoluble COD reactor profiles also showed no significant suspended solids gradients existed in the mixed liquor and that there was little accumulation of biomass in the bottom of the reactor. There was no excess floating sludge which was characteristic of DSFF reactors started up with sucrose wastewater at a concentration > 1.0% (Kennedy and Droste, 1985). These results showed the CM behavior of the system and supported DSFF reactor mixing studies previously reported and discussed (Kennedy and Droste, 1985). The CM regime in the DSFF reactor contrasts with the plug flow (PF) pattern reported for conventional random packed upflow anaerobic filters. In conventional filters the majority of the waste stabilization occurs in the bottom one-third of the reactor where most of the active biomass in the system is found (Young and McCarty, 1967). Biomass characteristics
Suspended biomass made up a small percentage of the total biomass within DSFF reactors (Table 3). An average value of 8.6% of all biomass in DSFF reactors was suspended biomass. In general, suspended biomass increased with increased waste strength and longer HRT. For DSFF reactors treating 0.25 and 2.0% sucrose wastewater, average VSS component of total reactor biomass was < 5% and >20%, respectively. Suspended solids concentration in the effluent is often a measure of effluent quality. Table 3 gives effluent suspended solids for influent waste concentrations as a function of HRT and organic loading rate, respectively. Concentration of suspended solids in the effluent increased only slightly for large decreases in HRT and concomitant increased organic loading rates. Most effluent suspended solids concentrations were < 1.8 kg COD m 3 (1.4 kg VSS m 3). Low biomass suspended solids even at low HRT indicate that microbial washout is not a major problem. Periodic checks of mixed liquor acidogenic and methanogenic activity (Kennedy and Droste, 1985) showed little or no methanogenic activity for the majority of steady-state conditions. Some methanogenic activity was found in the mixed liquor of the DSFF reactor treating 2.0% sucrose at a 7 day HRT; this was not surprising since certain methanogenic bacteria have doubling times of less than 7 days (Wandrey and Aivasidis, 1986). Acidogenic activity was always found in the mixed liquor and varied between 2-8 g COD (g VSS)- i day-I. Low concentrations of suspended solids (biomass) and absence of methanogenic activity in the mixed liquor not only reflects the small biomass yield but also the importance of the biofilm for waste stabilization. The presence of >90% of DSFF reactor methanogenic activity and the majority of the acidogenic activity in the biofilm of DSFF reactors operating at short HRT
690
K.J. K~NNEDVand R. L. DROSTE
during startup has also been reported by Kennedy and Droste (1985). For D S F F reactors treating concentrated wastewater at longer H R T the higher concentrations of suspended biomass (with acidogenic and methanogenic activity) shows that the D S F F reactor is beginning to operate more like a CM suspended growth reactor rather than a fixed biofilm reactor. This would mean that for high strength wastewaters ( > 20,000 mg COD 1- ' ) and long H R T there may not be a significant advantage to using a fixed-film anaerobic process. The role of the microbial film in conventional upflow anaerobic filters is not as important as in D S F F reactors. Dahab and Young (1982) have reported that in anaerobic filters the majority of the microbial activity (biomass) is located in the interstitial spaces between the support media in the lower third of the reactor and not attached to the media. In fact the anaerobic filter behaves more like an upflow anaerobic sludge blanket with media than an attached biomass reactor.
Biofilm characteristics The zoogleal biofilm (Atkinson and Fowler, 1974) that developed on the NPP support material was relatively porous and filamentous with tentacle like appendages extending up to 1 cm into the mixed liquor surrounding the biofilm. With the majority of the D S F F reactors methanogenic activity in the biofilm, biogas evolution and associated micromixing was intense (Fig. 5) in relation to the long residence time of liquid in the reactor and slow reaction rates of anaerobic microbes. The formation of gas bubbles is readily observed in Fig. 5. Biogas production in individual channels has been reported to be in part responsible for a gas lift mixing effect in D S F F reactors (Hall, 1982). Microscopic observation of
D S F F biofilms (Fig. 6, magnification 57 x ) showed microbes attached both to individual polyester fibres and entrapped by the NPP support material. Further magnification of Fig. 6 (magnification 1452 x ) shows bacteria characteristic of the genus Methanosarcina entrapped within the biofilm. The biofilms in all reactors were similar in that they were 5-6% (w/v of biofilm) TFS of which 75-80% were VFS.
Biofilm performance The relationship between organic loading rate and D S F F reactor biofilm biomass is shown in Fig. 7. Biofilm biomass concentration (in terms of reactor liquid volume) increased linearly with increased organic loading rates up to 10kg C O D m - 3 d a y -1 (0.133kg COD m - 2 d a y -1) and was relatively unaffected by influent substrate concentration. At a loading rate of 10 kg COD m - ~ d a y -l a maximum biofilm concentration and maximum biofilm thickness of 8.7kg VFS m -3 (0.116kg VFS m -2) and 2.6 mm, respectively, were obtained. At steady-state loading rates higher than 10kg COD m - 3 d a y 1, biofilm concentration levels off or at least the increase was very slow. Biofilm development during steadystate D S F F reactor operation differed from that found during reactor startup. During startup, biofilm development was found to decrease with increased substrate concentration for reactors operated at similar organic loading and removal rates. This difference in biofilm development is in part due to the startup protocol which limited VA accumulation in the reactor. Figure 8 shows biofilm concentration for various steady-state organic removal rates. Maximum biofilm concentration or a very tow rate of increase in biofilm biomass concentration was achieved at a steady state organic removal rate of 7.5kg COD m - 3 d a y -1
Fig. 5. Biogas bubble formation on biofilm.
Anaerobic fixed-film reactors treating carbohydrate wastewater
691
Fig. 6. Attached and entrapped biomass in the biofilm: (A) NPP fibre; (B) attached biomass; (C) entrapped biomass.
(0.100kg C O D m - 2 d a y - l ) . This indicates that an equilibrium of sorts was reached between biofilm and mixed liquor growth, attachment and biofilm sloughing. Recent studies have indicated that when the liquid velocity in reactor channels is sufficient no significant plugging should occur (Samson and Kennedy, 1985). If this is in fact the case, eventual plugging of the reactor channels should not occur. The D S F F reactor biofilm (biomass) concentration was less than reported for other advanced anaerobic reactors treating similar wastes at 35°C. Biomass concentrations in anaerobic/expanded beds (Switzenbaum and Jewell, 1980), anaerobic filters (Dahab and Young, 1982), upflow anaerobic sludge blankets (Lettinga et al., 1980) and combination upflow anaerobic blanket-filters (Guiot and van den
Berg, 1984) of 40, 25, 30 and 30kg VSSm 3, respectively, have been reported. Assuming that bioimass of similar specific activity develops in each type of reactor (Table 4), the maximum waste stabilization capacity of D S F F reactors would be less than other advanced anaerobic reactors. However, biomass may still be accumulating in the D S F F reactor at low rate, and over an extended period of reactor operation the biofilm concentration may increase resulting in a higher waste stabilization capacity.
Biofilm diffusion phenomena Figure 9 shows the effect of HRT on biofilm biomass concentration at various steady state conditions. Reactor biofilm concentration and concomitant biofilm thickness increased with decreasing HRT
692
K.J. KENNEDYand R. L. DROSTE 83% for DSFF reactors treating 2.5, 5.0 and 10 g 1-1 sucrose wastewater, respectively. At loading rates greater than 1.0 kg COD kg-] biofilm VFS dayCOD removal efficiency decreased rapidly. Figure 11 shows that the specific biofilm substrate utilization rate increased with increased loading rate reaching a maximum specific utilization rate of 0.9-1.2 kg COD kg -l biofilm VFS day -l at loading rates greater than 5 k g C O D m -3day -~ (0.066 kg COD m-2day-t). Figure 12 shows that the specific biofilm substrate utilization rate increased with decreased HRT and increased waste strength. Henze and Harremoes (1982) reported that anaerobic biofilms may become diffusion limited at biofilm thicknesses > 1 mm. Williamson and McCarty (1976) on the other hand reported that anaerobic biofilms would not be diffusion limited at any thickness.
10 .===---A 8 rn II= 7 tO >
g
g
5
E 3 .2 m 2
yof
4 0
I
I
I
I
I
I
I
I
I
I
2
4
6
8
10
12
14
16
18
20
COD
loading rate (kg
10
m -3 day - 1 )
9
Fig. 7. Effect of organic loading rate on biofilm biomass concentration: influent COD concentration, (11) 2500mgl -t, ( O ) 5000mgl -I, ( A ) 10,000mgl -], ( O ) 20,000 mg 1- t.
and increased waste strength (for the ranges studied). This behavior is similar to results reported for attached biomass in anaerobic expanded beds (Switzenbaum and Jewell, 1980). The steady state relationship between COD removal efficiency and specific biofilm substrate loading rate is shown in Fig. 10. COD balances to be reported in a future paper confirmed the reliability of the results. Because of the low concentration of suspended biomass with little or no methanogenic activity, the DSFF reactors specific loading rates are expressed in terms of biofilm biomass. COD removal efficiency decreased with increased specific biofilm substrate loading rate and was relatively unaffected by waste concentration. At a specific biofilm substrate loading rate of 1.0kg COD kg -t biofilm VFS day -~ COD removal efficiency was 80, 79 and
f
&
8
'E 7 to > 6 5
,
/
o
2
I 2 COD
I 4 removal
I 6
I 8
I 10
I 12
rate (kq m-3 day-~)
Fig. 8. Effect of organic removal rate on biofilm biomass concentration: influent COD concentration, (i) 2500mgl -I, ( O ) 5000mg1-1, ( A ) 10,000mgl -I, ((3) 20,000 mg 1- i
Table 4. Comparison of specific utilization rate for anaerobic reactors treating carbohydrate wastewaters
Wastewater
Reactor*
Empty bed volume (m3)
Operating temperature (°C)
Specific rate ( g C O D g I VSSday-I)
Sucrose (synthetic)
DSFF; (FG)
24.4 x I0 3
35
0.9-1.2
This study
Sucrose (synthetic)
Upflow sludge blanket filter; (SG)
4.3 x 10 3
27
1.0-1.2
Guiot and van den Berg (1984)
Dextrose/protein
Upflow filter; (FSG)
160 × 10 3
35
1.0
Mueller and Mancini (1975)
Sugar beets (soured)
Upflow sludge blanket; (SG)
18 x 10-3
30
0.6-0.9
Lettinga et at'. (1980)
Sucrose (synthetic)
Expanded bed; (FG)
1.0 x 10 3
30
0.44).8
Switzenbaum and Jewell (1979)
*(FG)--fixed growth; (SG)--suspended growth; (FSG)--fixed and suspended growth.
Reference
Anaerobic fixed-film reactors treating carbohydrate wastewater
693
2.0
10
1.9 t.8
9
• 8
•
IE
7
o3 >
6
g
1.7
O3 >
1.6 1.5 1.4 1.3
0
1.2 1,1
5
1.0 9
0,9
4
0,8
3
0,6
0,7
g
0,5
E
o 0.4 / 0.3 @) 0.2 I 0.1
2
_o ill3
1
0
I
I
I
I
2
4
6
8
I
0
2
H R T (doys)
I
I
I
I
1
I
I
I
4
6
8
10
12
14
16
18
COD
Iooding
rote
(kg
m - 3 d o y -1)
Fig. 9. Effect of HRT on biofilm biomass concentration: influent COD concentration, (11) 2500mgl -I, ( 0 ) 5000 mg 1-1, (&) 10,000 mg 1- i, (©) 20,000 mg 1- J.
Fig. 11. Effect of organic loading rate on specific biofilm substrate utilization rate: influent COD concentration, (m) 2500mgl -j, ( 0 ) 5000mgl -t, (A) 10,000mgl -I, (O) 20,000 mg 1- i.
Figure 13 shows the relationship between biofilm thickness and specific biofilm substrate removal rate. For the four waste concentrations studied, specific biofilm substrate utilization rate increased with biofilm thickness reaching a m a x i m u m at a biofilm thickness of 0.9 ram. With increased biofilm thickness, up to 2.6 m m , the specific biofilm substrate
removal rate remained relatively constant. This indicated that the whole biofilm was active for thicknesses up to 2 . 6 m m and was not diffusion limited. If the biofilm was not completely active or if a diffusion limitation existed, there would have been a decrease in the specific biofilm activity with increased biofilm depth. Table 4 compares the biofilm
100
90
2.0
F
A I>,
80 tO > 70
7
1.9 1.8 1.7 1.6 1.5
1.4
~
1.3
"~ 60
a ~
1.2 1.1
o 50 a 0~ 40
~
1.0
i
0.8 0.9 0.7 I
•
.1o ~ 30
L
o o.6 ~ 0.5 O.4 ~- 0.3 ~ 0.2 0.1
g,
2010-
0
\
I 0.4 Specific
I 0.8 Iooding
I l.Z
I t.6 rote
I 2.0
[k O COD
I 2.4
I 2.8
kg - 1 V S doy-1)
Fig. I0. Effect of specific biofilm loading rate on soluble COD removal: influent COD concentration, ( , ) 2500mgl -I, ( 0 ) 5000mgl -I, (&) 10,000mgl -I, (C)) 20,000 mg 1-1.
0
I
I
I
I
2
4 HRT (doys)
6
8
Fig. 12. Effect of HRT on specific biofilm substrate utilization rate: influent COD concentration, (m) 2500 mg l-=, (O) 5000 mg 1-1, (&) 10,000 mg I-I, (O) 20,000 mg 1-t.
694
K . J . KENNEDY and R. L. DROSTE
(4) The majority of the acidogenic biomass and methanogenic biomass activity in D S F F reactors was located in the biofilm. (5) Maximum biofilm substrate utilization rates (0.9-1.2kg COD kg -~ V F S d a y - l ) were similar to rates reported for suspended growth cultures treating carbohydrate wastewater. (6) Anaerobic biofilms are completely active and not diffusion limited up to biofilm thicknesses of 2.6 mm.
2,0 1.9
1.8
~ if)
> ?
121 0
~ o ~
1.7 1.6 1.5 1.4
1.3 1.2 1.1 1.0
o. 7 0
0.9 0.8
~ 0.7 E 06 0.5 ~ 0.4
0 REFERENCES
•
=• *
Ackman R. G. (1972) Porous polymer bead packings and formic acid vapor in GLC of volatile fatty acids. J. Chromat. Sci. 10, 560-565. APHA (1975) Standard Methods for the Examination of ~= 0 3 Water and Waste Water Including Bottom Sediments and t/) 0.2 Sludges, 14th edition. American Public Health Association, New York. 0.1 I I I I I I Atkinson B. and Fowler H. W. (1974) Significance of 0 0.04 0.08 0.12 0.16 0.20 0.24 microbial films in fermenters. Adv. Biochem. Engng 3, Biofilrn thickness (crn) 221-277. Bachman A., Beard V. L. and McCarty P. L. (1983) Fig. 13. Effect o f biofilm thickness on specific biofilm Performance characteristics of the anaerobic baffled reacsubstrate utilization rate: influent COD concentration, ( I ) tor. 3rd International Symposium on Anaerobic Digestion, 2500mgl -I, ( 0 ) 5000mgl -I, (A) 10,000mgl -I, (O) Boston, Mass. 20,000 mg 1- t. Berg L. van den and Kennedy K. J. (1981) Support materials for stationary fixed film reactors for high rate methanogenic fermentations. Biotechnol. Lett. 3, 165-170. substrate utilization rate to other anaerobic systems Berg L. van den and Lentz C. P. (1979) Comparison treating carbohydrate wastes and shows that there between up- and down-flow anaerobic fixed film reactors was little difference between suspended and attached of varying surface-to-volume ratios for the treatment of biofilm biomass. The porous nature of the anaerobic bean blanching waste. Proc. 35th Ind. Waste Conf. Purdue Univ., pp. 319-325. biofilm with its tentacle like appendages (increased surface area for microbial/substrate contact), the lack Dahab M. F. and Young J. C. (1982) Retention and distribution of biological solids in fixed bed anaerobic of an electron donor and high Ks values of anaerobic filters. 1st International Conference o f Fixed Film Biologibacteria (low affinity for substrate; Williamson and cal Processes, Kings Island, Ohio. McCarty, 1976), biofilm micromixing caused by bio- Guiot S. and Berg L. van den (1984) Dynamic performance of an anaerobic reactor combining an upflow sludge gas production and the relatively long residence time blanket and a filter for the treatment of sugar waste. Proc. of substrate in contact with the biofilm (minimum 39th Ind. Waste Conf. Purdue Univ. pp. 705-717. 12 h) all may be responsible for minimizing diffusion Hall E. R. Biomass retention and mixing characteristics in problems and increasing the active depth of anaerfixed film and suspended growth anaerobic reactors. IA WPR Specialized Seminar on Anaerobic Treatment in obic biofilms. It is also evident that high substrate Fixed Film Reactors, Copenhagen, Denmark, pp. removal rates in D S F F reactors were directly related 371-398. to the amount of biomass that could be retained in Henze M. and Harremoes P. (1982) Anaerobic treatment of the reactor's biofilm. waste water in fixed film reactors. IA WPR Specialized Seminar on Anaerobic Treatment in Fixed Film Reactors, Copenhagen, Denmark, pp. 1-94. CONCLUSIONS Huyssteen J. J. van (1967) Gas chromatographic separation of anaerobic digester gases using porous polymer. Wat. (1) The high concentration of biomass retained in Res. 1, 237-242. D S F F reactors permits successful treatment of carboKennedy K. J. and Berg L. van den (1982a) Anaerobic hydrate wastewater at high organic loading rates and digestion of piggery waste using a stationary fixed film short HRT. reactor. Agric. Wastes 4, 151-158. (2) D S F F reactors do not retain as much biomass Kennedy K. J. and Berg L. van den (1982b) Stability and performance of anaerobic fixed film reactors during as other advanced reactors and cannot achieve as hydraulic overloading at 10-35°C. Wat. Res. 16, high organic loading rates. 1391-1398. (3) Steady-state biofilm concentration for D S F F Kennedy K. J. and Droste R. L. (1985) Startup of anaerobic reactors treating a carbohydrate wastewater of differdownflow stationary fixed film (DSFF) reactors. Biotechnol. Bioengng 27, 1152-1165. ent concentrations increased with increased organic loading rate and decreased H R T reaching a maxi- Kennedy K. J., Muzar M. and Copp G. (1985) Stability and performance of mesophilic anaerobic fixed film reactors mum of 8.7kg V F S m -3 (0.116kg V F S m -2) at a during organic overloading. Biotechnol. Bioengng 27, loading rate of 10kg C O D m - 3 d a y -~ (0.133kg 86-93. Knechtel J. R. (1978) A more economical method for the COD m-2 d a y - l ). -
D
/
Anaerobic fixed-film reactors treating carbohydrate wastewater determination of Chemical Oxygen Demand. Wat. Pollut. Control 116, No. 5, 25. Lettinga G., Velsen A. van, Hobma W., Zeeuw W. de and Klapwijk A. (1980) Use of the upflow sludge blanket (USB) reactor concept for biological waste water treatment, especially for anaerobic treatment. Biotechnol. Bioengng 22, 699-734. Maat D. (1984) Design, construction and operation of a full scale downflow fixed film reactor using hog waste substrate. 5th Bioenergy R&D Seminar, Ottawa, Canada. Mueller J. A. and Mancini J, L. (1975) Anaerobic filter-kinetics and application. Proc. 30th Ind. Waste Conf., Purdue Univ. pp. 423-447. Samson R. and Kennedy K. J. (1985) Effect of reactor height on mixing characteristics and performance of the anaerobic downflow stationary fixed film (DSFF) reactor. J. Biotechnol. 2, 95-106. Samson R., Peters R., Hade C. and Berg L. van den (1984) Dairy waste treatment using industrial scale fixed film and
695
upflow sludge bed anaerobic digesters: design and start-up experience. Proc. 39th Ind. Waste Conf., Purdue Univ. pp. 14~49. Switzenbaum M. S. and Jewell W. J. (1980) Anaerobic attached film expanded bed reactor treatment. J. Wat. Pollut. Control Fed. 52, 1953. Szendry M. L. (1983) Bacardi Corporation digestion process for stabilization of rum distillery wastes and production of methane. Institute o f Gas Technology: Energy from Biomass, Lake Beauna Vista, Fla. Wandrey C. and Aivasidis A. (1986) Continuous anaerobic digestion with Methanosarcina barkeri. Ann. N.Y. Acad. Sci. In press. Williamson K. and McCarty P. L. (1976) Model of substrate utilization by bacterial films. J. Wat. Pollut. Control. Fed. 48, 9-24. Young J. and McCarty P. L. (1967) The anaerobic filter for waste treatment. Proc. 22nd Ind. Waste Conf. Purdue Univ. pp. 559-574.