instability caused by high strength of cheese whey in a uasb reactor

Instability Caused by High Strength of Cheese Whey in a UASB Reactor

J. Q. Yan,' K.V. Lo? and K. L. Pinde$*

British Columbia, Canada; ZDepartment Bio-Resource Engineering, and 3Department of Chemical Engineering, The University of British Columbia, Vancouver, Canada

Received October 28, 1992/Accepted October 29, 1992

Laidlaw Environmental Services, Ltd., 7842 Progress Way, Delta,

The anaerobic digestion of cheese whey was studied in a UASB reactor. The profiles of the reactor, i.e., the distributions of the substrate concentration and pH under different operating conditions were developed. From the concentrations of substrates measured at various levels above the bottom of the reactor, two reaction stages, namely acidogenesis and methanogenesis, were distinguished. The instability caused by high influent concentration was interpreted as the accumulation of VFAs in the acidogenic stage beyond the assimilative capacity of the methanogenic stage. A range of stable operating conditions was predicted from the results of the profile measurements. The optimal influent concentration was found to be between 25 and 3 0 g COD/L at an HRT of 5 days for system stability. Other options for stability control were discussed. 0 1993 John Wiley & Sons, Inc. Key words: anaerobic digestion cheese whey UASB reactor

INTRODUCTION

Cheese manufacturing, one of the biggest food industries in North America, is facing a difficult waste disposal problem. Despite the fact that a number of studies6-" during last decade have shown that anaerobic fermentation of cheese whey could be a possible means for waste disposal, the system is suspect due to difficulties frequently encountered in maintaining a stable operation. Numerous researchers have reported unsuccessful experiments. Their lack of success was apparently due to the high organic concentration and the tendency of rapid acidification of the waste. The frustrating unsuccessful experiments have led to the suggestion that, without pH control, the system stability was hard to maintain.' Different pH-control devices have been used in previous ~ t u d i e s . ~ , ~ On the other hand, some nutritional additions were thought to be necessary to maintain the system ~tabi l i ty .~

Improper reactor design and poor start-up procedure may be the reasons for failure in previous research. Efforts have been made to improve treatment efficiency by using an upflow anaerobic sludge blanket (UASB) reactor equipped with a three-phase separator." Results from the prelimi- nary feasibility assessment have shown that the anaerobic

* To whom all correspondence should be addressed.

digestion of cheese whey using a UASB reactor can be an efficient treatment method for diluted cheese whey.

The question then arose as to why high influent concentration makes the operation of the reactor more difficult. What really causes the instability? Motivated by the need to overcome the existing difficulties, this study was to initiated to find the cause of the instability and to develop control strategies for maintaining a stable operation. In the first step of the research, a study of the profiles in a UASB reactor, i.e., the distribution of the substrate concentration and pH under a wide spectrum of operating conditions, was conducted.

From these measurements, several interesting conclu- sions have been drawn. It is believed that this is the first time in anaerobic fermentation studies that the two stages, acidogenesis and methanogenesis, have been shown to occur separately in the same reactor. The results from this study provided a better understanding of the cause of the instability.

EXPERIMENTAL APPROACHES

Reactor Set-Up

A UASB reactor was used in these studies, in an effort to increase the volatile suspended solid concentration in the reactor and increase the treatment efficiency. Second, pH control and nutrient addition were not used, in order to observe the instability and to find its cause.

The flow sheet of the UASB system is presented in Figure 1. The reactor was made of acrylic pipe with an inner diameter of 11.5 cm (4.5 in.) and a height of 168 cm (60 in.). The total volume and working volume of the reactor were 17.5 L and 14.3 L, respectively. A series of sampling ports were fitted at intervals on one side of the reactor to permit sampling for the analysis of the sludge, COD, VFA, and pH. The reactor was operated in an upflow and continuous model.

The whey, stored in a refrigerator at 4"C, was introduced continuously into the bottom of the reactor by a peristaltic pump through a copper-alloy coil immersed in a water bath to bring the feed temperature to about 34°C. The effluent left the reactor from the top of the settlement chamber and was collected in a plastic container. The seed

Biotechnology and Bioengineering, Vol. 41, Pp. 700-706 (1993) 0 1993 John Wiley & Sons, Inc. CCC 0006-3592/93/070700-07

DIOGAS L

f I - + I

S A M P L I N G

T A P S

WATER SEAL 4

GAS MEIER

Figure 1. Schematic diagram of the UASB reactor.

floculent sludge was originally obtained from the effluent of an anaerobic rotating biological reactor (AnRBC) in the laboratory, and had been stored in a plastic container in a cool room at 4°C for about 2 months. The effluent from the AnRBC was moved to a glass jar at a temperature of 22”C, and was acclimated using 200 mL of raw cheese whey daily for more than 30 days before it was put into the reactor. Then 4 L of this effluent containing 3% TS and 2.1% VS was used as seed. The details of reactor start-up have been given in a previous article.’*

The fermentation temperature was maintained at 33°C using a temperature controller and a number of external electric heater pads wrapped around the reactor. In this set-up, no pH control was used.

Feed Substrate

Cheddar cheese whey used in this study was obtained from the Fraser Valley Milk Producers Association’s cheese producing plant at Abbotsford, British Columbia, Canada. The characteristics of the whey are described in ref. 12. The fresh cheese whey was collected in a 50-L tank and transported from the cheese plant to the large freezer of the Bio-Resource Lab at UBC and stored there at -30°C. About 1 week before it was needed, a portion of the frozen whey was moved into a cold room at 4°C for defrosting. Then it was diluted to the desired COD concentration by mixing with cool tap water, adjusted to a pH of about 7 by using 5% NaOH, and transferred to the small scale

feed tank, where it was kept refrigerated at 4°C and ready for use.

Experimental Design

There were five different operating conditions, during which the influent COD was increased stepwise from 5 to 40 g COD/L at an approximately constant hydraulic retention time (HRT) of 5 days (Table I). The feed rate was 2.9 L per day. The change of organic loading rate (OLR) depended only on the change of influent concentration. It was noticed that influent concentration was more critical than OLR with regard to inhibition of bacteria for this particular type of reactor.12 However, the OLR is used more often in waste treatment systems to correlate treatment efficiency of reactors.

For each influent concentration, samples were taken from the influent, the corresponding effluent and eight sampling ports mounted along the height of the reactor. To obtain reliable samples for analysis, only one port was sampled every hour starting from sample port 10 (top) to port 1 (bottom). For each subsequent increase in the influent concentration, an operating period of 2 to 3 HRTs was maintained to ensure steady state.

Chemical Analyses

Analyses conducted on the influent and effluent were: total solids (TS), total suspended solids (TSS), volatile suspended solids (VSS) and ash according to the “standard methods.” Chemical oxygen demand (COD) was determined by the colorimetric method using an optical fiber instrument. Gas production was measured by a wet gas meter and then corrected to the standard temperature and pressure (STP). Both gas composition and volatile fatty acid (VFA) were analyzed on a Hewlett Packard 5890A gas chromatograph, using an external standard. The GC was equipped with both a flame ionization detector and a thermal conductivity detector with separate columns; a Porapak column was used for gas analysis and a Carbopack C column for VFAs. Total Kjeldahl nitrogen (TKN), and ammonia nitrogen (NH3-N) were determined using a block digestor and a Technicon Auto Analyzer 11.

Table I. Operating conditions.

Organic loading Influent conc. Run time (g COD/L . d) (COD/L) (days)

0.91 1.97 3.54 5.96 7.77

4.6 9.9

17.7 28.7 38.1

30 28 36 31 18

YAN, LO, AND PINDER: INSTABILITY CAUSED BY HIGH STRENGTH OF CHEESE WHEY 701

RESULTS AND DISCUSSION

Profiles of COD, VFA, and pH

COD, VFA, and pH were monitored at influent concentrations of 4.56, 9.93, 17.1, 28.8, and 38.1 g COD/L, (pH was not measured at an influent concentration of 4.56 g COD/L). The pH, acetic acid, propionic acid, and COD profiles are presented in Figures 2, 3, 4, 5, and 6, respectively.

Below 4 cm, the pH was in the range from 4 to 5, VFA concentrations were high (up to 2895 mg/L of acetic acid), and COD reduction was between 17% and 49%, depending on the activity of the sludge and the influent concentration. Above 12 cm, the pH increased to 6.4 and the volatile fatty acid concentration decreased to 100 mg/L or less. More than 60% of total COD reduction occurred between 4 cm and 12.5 cm above the reactor bottom ex- cept for an influent concentration of 38.1 g COD/L. The two completely different sets of pH, COD, and VFAs values demonstrated that two separate reaction phases, acidogenesis and methanogenesis, were established in the reactor. The acidogenic phase occurred in the bottom, below 4 cm, which was indicated by a lower pH, higher VFAs and a lower COD reduction (Fig. 3). Above 12 cm, methanogenesis took place. This was demonstrated by a higher pH, lower VFAs, and a high COD removal.

Up to now, in anaerobic digestion studies, phase separation has been accomplished in two separate reactors by

controlling the pH and dilution rate. This is the first time, as far as the authors know, that the occurrence of two phases in the same reactor were reported. It was thought that phase separation would occur for all substrates in plug-flow reactors. Interestingly enough, when using a UASB reactor to treat baker’s yeast wastewater in the same laboratory, two distinct phases were not observed.

The different results could be due to the inherent chemical characteristics of the substrates. Anaerobic digestion is a biological process in which a series of parallel and consecutive reactions take place. From an oversimplified point of view, it has been accepted that only two major steps, acidogenesis and methanogenesis, are considered essential, and generally, the second one is extremely slow and, therefore, is the rate-controlling step. If the two major steps remain in balance, the intermediate products, i.e., VFAs, would not be detected. Therefore, two phases would not be seen in one reactor. It is only possible to observe the two phases in cases in which the reaction rates of the two steps are very different. In other words, the observation of two phases in one reactor is only possible for some particular substrates, such as cheese whey, which is easily converted into short chain acids by acidogenic bacteria. When more fatty acids are formed than can be converted, an accumulation of VFAs occurs and the pH drops. The accumulation of VFAs in the first step being faster than the assimilative capacity in the second step creates a distinct acidogenic phase. The baker’s yeast wastewater contained high concentrations of hard-to-degrade organic material

I I I

Influent

-c COD.9.9 g/l

+ COD47.1 * COD.28.8

-El- COD.38.1

3

I I I I I I

4 12 24 34 50 60 90 110

Reactor Height (cm) Figure 2. Profile of pH.

702 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 41, NO. 7, MARCH 25, 1993

Acetic Acid (mg/l) 10000

1000

100

Influent COD

4.56 g/l --c

+ 9.93 g/l

* 17.7 g/l

-€I- 28.80 g/l

++ 38.1 g/l

10 ' I I I I I I

4.0 12.5 25 37.5 50 62.5 87.5 112

Reactor Height (cm) Figure 3. Profile of acetic acid.

that could not be easily acidified. In contrast, whey has a tendency toward rapid acidification. The observation of two phases indicates that the anaerobic system using whey is not maintained in dynamic balance even at low organic loading rates, even though the overall system, from the effluent analyses, appeared to be very stable at influent concentrations below 28 g COD/L.

With an increase in influent concentration, the VFAs and COD in the reactor gradually increased. The curves for COD, acetic acid, propionic acid, and pH distribution as a function of the height did not change markedly until an influent concentration of 38.1 g COD/L was used. In this condition, the acidogenic as well as the methanogenic zones extended upward. Much higher COD, acetic, and

[ropionic Acid (mg/l) , 10000

t \\- \ I \ k

1 ° k \7\ t \ \ \

1

0.1 I I I I I I

4.0 12.5 25. 37.5 50. 62.5 87.5 112

Reactor Height (cm) Figure 4. Profile of propionic acid.

Influent COD

4.56 g/l

+ 9.93 g/l

-c

+l+ 17.70 g/l

%-- 28.80 g/l

* 38.10 g/l

703 YAN, LO, AND PINDER: INSTABILITY CAUSED BY HIGH STRENGTH OF CHEESE WHEY

COD Concentration (g/l) 1ooc

1 + 9.93 g/l 1 \

0.1 ' I I I 1 I 1

4.6 12.5 25.0 37.5 50. 62.5 87.5 112.

Reactor Height (cm) Figure 5. Profile of COD.

propionic acid concentrations were accumulated at the bottom. These high concentrations also extended to a height of 37.5 cm above the reactor bottom. For example, the pH value remained around 3 at a height of 37.5 cm. In particular, the propionic acid concentration remained high throughout the reactor (Fig. 4). Consequently, the overall reactor performance was affected. The process became unstable 14 days after the reactor was initially fed at this

concentration. This was indicated by a decrease in gas production from 67 to 61 L/day, and an increase in effluent COD from 55 to 643 mg COD/L, and also an increase in effluent acetic and propionic acid concentrations to 80 and 64 mg/L, respectively.

The upward extension of the acidogenic as well as the methanogenic zones causes the intrusion of the acidogenic phase into the region previously occupied by methanogens.

COD Reduction (YO)

I I

I I I I I I

4 .O 12.5 25. 37.5 50. 62.5 87.5 112

Reactor Height (cm) Figure 6. Profile of COD reduction.


The methanogens, previously highly active, in this region could be rendered inactive under the acidic environment. Moreover, the methanogens could not be easily replenished in a newly established methanogenic phase to counteract the accumulation of VFA concentrations, due to the very slow growth rate of the methanogens. It can be predicted that the acidogenic region will extend into the upper portion of the reactor as the substrate loading is increased until the whole region is occupied by the acidogenic reaction and fermentation fails. This is the bottleneck of a suspended growth system. In general, the maximum influent concentration accepted in the UASB processes has been 30 g COD/L, to the best knowledge of the authors, based on a literature search, even for those substrates which were not quickly acidified.

7 ,

6 -

5 -

4 -

Control of Instability

With an increase in influent concentration, VFA and COD in the reactor increased. In other words, more of the VFAs produced in the first step accumulated. This could indicate that the rate of acidogenesis increased with the increase in influent COD. Let the acetic acid concentration at sample port 1 represent the accumulation of VFA in the first phase. The difference in acetic acid between ports 1 and 2 represents the degradation capacity of VFA in the second phase. The requirement for maintaining the anaerobic system in a dynamic balance is that the degradation capacity of VFA in the second phase be greater than the accumulation of VFA in the first phase. Based on this idea, the best influent concentration can be found with regard to the system stability.

- Accu mu lat io n - ~ ~ - - - D eg rada t ion

Using linear regression analysis, a set of empirical mod- els for the accumulation of acetic acid and propionic acid with an increase in influent concentration has been developed which is the best fit to the experimental data (Ref. 12, thesis Appendix A).

AA = -0.27 + 0.183C0 - 0.0102C0~ + 0.000197C0~

PA = 0.112 - 0.00893C0 + 0.00108C0~

where AA = acetic acid, PA = propionic acid, and Co = feed strength in g COD/L.

The accumulation and degradation of acetic acid as calculated by these equations are shown as functions of concentration in Figure 7. The degradation rate first increased until the influent concentration reached 20 g COD/L, then declined. Between 15 and 28 g COD/L, the degradation curve is above the accumulation curve, which means that in this region the degradation capacity exceeds the accumulation capacity. Therefore, this would be the optimal influent concentration for a cheese whey anaerobic fermentation system using a UASB reactor. This conclusion agrees with the experimental results.

1 -

Other Options of Stability Control

It is now clear that instability can be caused by the accumulation of VFAs resulting from higher organic strength of whey. The excess VFAs, in turn, have suppressed the activity of the methanogens. Control strategies then, for minimization of the impact of VFAs in whey anaerobic digestion, should be those that result in a reduced level of VFAs. This would include operation at an influent concentration of whey between 15 and 28 g COD/L at the

0 .4 10 20 30 40 50

Influent Concentration (911) Figure 7. Accumulation and degradation of acetic acid.

YAN, LO, AND PINDER: INSTABILITY CAUSED BY HIGH STRENGTH OF CHEESE WHEY 705

given flow rate or HRT as discussed in the previous section. Within this range of applied concentration, the degradation capacity for VFAs is greater than total accumulation of VFAs (Fig. 7).

Application of a high influent concentration is desirable because of the nature of cheese whey. Therefore, other control strategies have advantages and are worthy of evaluation and development. Control of pH is an existing and very common way to stabilize a stressed anaerobic system which suffers from poor buffer capacity when highloading is applied. However, the link between pH and system stability has not been clearly shown. In addition, it must be realized that a high concentration of cations added to neutralize acids has an inhibitory effect on methane bacteria.

Studies2 have shown that molecular hydrogen and in- terspecies hydrogen transport play an important role in anaerobic digestion. According to thermodynamics, the accumulations of VFAs are regulated by hydrogen pressure, which is moderated by hydrogenotrophic association (including methane-producing bacteria, nitrogen-reducing bacteria, and sulfate-reducing bacteria). On the other hand, the oxidation of VFAs can proceed also only when the hydrogen pressure is kept extremely low. The rate of degradation of VFAs can be enhanced through the collective effects of hydrogenotrophic association. Therefore, it is rea- sonable to believe that system stability could be effectively controlled through a syntrophic association. Further studies are needed on the effect of sulfate or nitrate additions. The effect of sulfate addition will be discussed in a separate study.

SUMMARY

Two reaction stages, namely acidogenesis and methanogenesis, were distinguished in the UASB reactor by the profiles of the substrate concentrations, which indicated that cheese whey is easily converted to short chain fatty acids and that the rate of the first stage is much faster than the second stage. The appearance of two stages in the same reactor was associated with either the nature of the

substrate or process stress and could be attributed to the fact that the rate of VFA production exceeded the rate of their utilization. To control system stability, either the reactor should be fed with influent concentration in the optimal range of 15 to 28 g COD/L at an HRT of 5 days, or a pH-control system should be used. A new control strategy with hydrogenotrophic association is possible but demands further research.

References

1. Boening, P.H., Larsen, V.F. 1982. Anaerobic fluidized bed whey treatment, Biotechnol. Bioeng. 24: 2539-2556.

2. Harper, S. R., Pohland, F. G. 1986. Biotechnology report: Recent developments in hydrogen management during anaerobic biological wastewater treatment. Biotechnol. Bioeng. 28: 585 -602.

3. Kelly, C. R., Switzenbaum, M. S. 1984. Anaerobic treatment: Tem- perature and nutrient effects. Agric. Waste 1 0 135-154.

4. Lo, K. V., Liao, P. H. 1986. Digestion of cheese whey with anaerobic rotating biological contact reactor. Biomass 10: 243 -252.

5. Marshall, D., Timbers, G.E. 1982. Development and testing of a prototype fixed-film anaerobic digester. Paper No. 82-65 19. Winter meeting of ASAE, St. Joseph, MI.

6. Nodstedt, R. A,, Thomas, M. V. 1984. Inoculum requirements for start-up of anaerobic fixed bed reactors. Paper No. 84-4090. Summer meeting of ASAE, University of Tennessee, Knoxville, TN.

7. Samons, R., Van den Berg, B., Pepter, R., Hade, C. 1984. Dairy waste treatment using industrial scale fixed-film and up-flow sludge bed anaerobic digesters: Design and start-up experience. Proceedings of the 39th Industrial Waste Conferences, Purdue University, W. Lafayette, IN, pp. 235-241.

8. Switzenbaum, M. A,, Danskin, S. C. 1982. Anaerobic expanded bed treatment of whey. Agric. Waste 4: 411-426.

9. Wildenauer, F.X., Winter, J . 1985. Anaerobic digestion of high strength acidic whey in a pH-controlled up-flow fixed-film loop reactor. Appl. Microbiol. Biotechnol. 22: 367-372.

10. Williams, D. W. 1984. Fixed-film anaerobic digestion of cheese whey. Paper No. 84-4095. Summer meeting of ASAE, University of Tennessee, Knoxville, TN.

11. Yan, J . Q., Lo, K. V., Liao, P. H. 1989. Anaerobic digestion of cheese whey using up-flow anaerobic sludge blanket reactor. Biol. Waste 27: 289 - 305.

12. Yan, J. Q. 1991. Anaerobic digestion of cheese whey using an up-flow anaerobic sludge blanket reactor, Ph.D. thesis, University of British Columbia, Canada.


instability caused by high strength of cheese whey in a uasb reactor

Documents