- Email: [email protected]
On-line monitoring of a two-stage anaerobic digestion process using a BOD analyzer
Journal of Biotechnology 109 (2004) 263–275
On-line monitoring of a two-stage anaerobic digestion process using a BOD analyzer夽 Jing Liu a,∗ , Gustaf Olsson b , Bo Mattiasson a a
Department of Biotechnology, Center for Chemistry and Chemical Engineering, Lund University, P.O. Box 124, S-221 00 Lund, Sweden b Department of Industrial Electrical Engineering and Automation, Lund University, P.O. Box 118, S-221 00 Lund, Sweden Received 21 July 2003; accepted 18 November 2003
Abstract A computer-controlled biochemical oxygen demand (BOD) analyzer has been developed for fast estimation of biochemical oxygen demand (BODst ) automatically with the purpose of on-line monitoring of a process for conversion of biomass under field conditions. The instrument was tested by on-line monitoring of the connecting stream between two stages of a two-stage anaerobic process in laboratory scale. In the first stage, hydrolysis of sugar beet leaves and its conversion into volatile fatty acids and other low molecular weight substrates took place. The effluent from the first reactor was used as a feed stream to the second stage, i.e. an anaerobic contact reactor. The feed stream was sampled intermittently, diluted and analyzed by the BOD analyzer automatically in order to estimate the organic loading rate to the reactor. The results from this study demonstrated that the BOD analyzer could be a stand-alone and promising sensor device for rapid on-line monitoring of easily biodegradable organic substances in biological treatment processes. © 2004 Elsevier B.V. All rights reserved. Keywords: BOD analyzer; BODst ; Monitoring; On-line; Two-stage; Anaerobic digestion
1. Introduction In recent years, much of the research pertaining to anaerobic processes has been directed towards identifying parameters, which may be used to monitor the process and give an early warning of system instability (Cobb and Hill, 1991; Cord-Ruwisch et al., 1997). It has been identified that the activity of different 夽 Part of this work was presented at the Third International Symposium on Anaerobic Digestion of Solid Wastes, Munich/Garching, Germany, September 18–20, 2002. ∗ Corresponding author. Tel.: +46-228347; fax: +46-2224713. E-mail address: [email protected] (J. Liu).
groups of organisms involved in the anaerobic process can be measured indirectly by monitoring the metabolites. Some of the commonly suggested indicators include pH, alkalinity, concentration of volatile fatty acids (VFAs), dissolved organic monomers, dissolved hydrogen, gas production rate, as well as gas composition (Ahring et al., 1995; Ghosh, 1991; Hickey and Switzenbaum, 1991; Björnsson et al., 2001; Moletta et al., 1994). It is, however, essential to develop on-line methods for these parameters in order to indicate the status of the processes in real time. This paper presents an application of a computercontrolled analysis system which is based on a biosensor method for estimating the easily biodegradable
0168-1656/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jbiotec.2003.11.014
264
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
organic compounds that can give a fast indication of intermediate products from the initial step of an anaerobic process.
ated far below maximum reactor capacity to eliminate the risk of system overload.
1.1. Anaerobic biodegradation process
1.2. Approaches for improving the stability and efficiency of anaerobic process
The anaerobic decomposition of organic matter into methane and carbon dioxide is a multi-step process involving a well-organized community of various microbial populations. In the first step, organic polymers (e.g. polysaccharides, fats and proteins) are hydrolyzed into smaller units (e.g. sugars, long-chain fatty acid and amino acids) by extracellular hydrolases excreted by obligate or facultative anaerobes. The products resulting from the hydrolysis step are subsequently used as substrates in the next step which is often referred to as acidogenesis. This degradation pathway is often the fastest step and also gives a high-energy yield for the microorganisms. A wide variety of different microbes representing both obligate and facultative anaerobes are involved. The fermentation products resulting from the acidogenic steps include acetate and reduced intermediates (electron sinks) such as lactate, ethanol, propionate, butyrate and higher volatile fatty acids. Conversion of these electron sinks to acetate, carbon dioxide and hydrogen is then carried out in the acetogenic steps by obligate hydrogen-producing acetogenic bacteria. Because of a low-energy yield from the acetogenic degradation, acetogenic microorganisms are very slow-growing and sensitive to changes in organic load, flow rate and environmental changes. The substrates that can be used in the final methanogenic step are acetate, H2 /CO2 , methanol and formate. The most important methanogenic transformations are the acetoclastic reaction and the reduction of carbon dioxide. Very few known methanogens can perform the acetoclastic methane production, whereas nearly all known methanogenic species are able to produce methane from H2 /CO2 (Björnsson, 2000). Owing to the limiting energy available during anaerobic degradation, some of the microbial populations have a low growth rate, which makes them vulnerable and sensitive to changes in operational conditions. This can cause instability during both the start-up and steady-state running operations of a biomethanation process. Consequently, many biogas systems are oper-
One approach to overcome the problem is close monitoring the process with regard to selected parameters and using these signals for improving the operational performance by a better process control. Furthermore, the different microbial populations involved in anaerobic digestion do not have the same requirements on operational conditions. The growth rates and pH optima are different for acidogenic and methanogenic organisms. In a one-stage digester, the pH and the organic loading rate (OLR) are adjusted to suit the slow-growing methanogenic organisms at the expense of the relatively fast-growing acidogens and process efficiency as a whole. The operational conditions for the different microbial groups can be further optimized if the process is divided into two stages in separate reactors. The hydrolysis and acidogenesis take place in the first reactor. The effluent from this reactor is used as a feed stream to the second stage, a biomethanation reactor. The methanogens in the second stage can be effectively protected from overload and toxic shocks by close monitoring of the connecting stream. A monitoring parameter that can indicate the intermediate products (e.g. sugars, long-chain fatty acid and amino acids) from the first stage and consequently reflect a loading of easily biodegradable organic material to the biomethanation reactor could be selected with a high priority. The biochemical oxygen demand (BOD) determination has its widest application in measuring loading of biodegradable organic material to wastewater treatment plants and in evaluating the BOD removal efficiency of such a treatment process. It is an empirical test in which a standardized laboratory procedure is used to determine the relative oxygen requirements of wastewaters. The determination of conventional BOD includes the well-known BOD5 (APHA, 1992) and BOD7 (SIS, 1979) tests, which need 5 or 7 days of incubation at 20 ± 1 ◦ C in the dark, respectively. Since most organic pollutants can be biodegraded under both aerobic and anaerobic treatment, this type of assay also has value when it is applied to the anaerobic treatment. However, electron acceptors other than oxygen
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
are involved in anaerobic degradation. BOD5 can be a valid parameter for indicating the biodegradable and assimilable organic load to an anaerobic process and a possible subsequent treatment unit, such as an aerobic polishing unit, or a receiving surface body of water, and is crucial for evaluation of the process yield and stability (Speece, 1996). The conventional BOD test is time consuming and therefore not suitable for the purpose of on-line process monitoring. Fast determination of BOD can also be achieved using biosensor methods based on the respirometric principle. However, the use of BOD biosensors results in short-term BOD (BODst ) values that are not identical with the conventional BOD values in all cases. The microbial oxygen consumption measured by the conventional method is the sum of the oxygen used to oxidize both easily assimilable compounds and biodegradable polymers (e.g. starch and proteins), whereas BOD biosensors can normally only give responses to those fast and easily assimilable compounds in the samples (Liu and Mattiasson, 2002). Nevertheless, BODst seems to be a more valuable parameter for control purposes because it represents the assimilable organic matter supply within the time constraint, and is intimately related to the dynamic performance of the process. The BOD5 , on the other hand, merely characterizes the impact on the receiving water (Spanjers et al., 1993). 1.3. Outline of the work This paper presents a follow-up study from the earlier work (Liu et al., 2003). In order to overcome the limitations of a previously developed BOD biosensor (Liu et al., 2000), such as operation within a narrow linear concentration range and poor stability, a new sensor and an associated computer-based flow injection system have been designed and constructed. In this study, the in-house developed BOD analyzer was applied to on-line analysis of the connecting stream between the two stages of a two-stage anaerobic process in order to estimate OLR to the biogas reactor. The correlation between OLR and gas formation was studied for a discrete loading, a continuous loading with a relatively constant content of organic matter during each loading period, and a continuous loading with varied content of organic matter over time.
265
2. Methods 2.1. Two-stage anaerobic reactor Fresh sugar beet leaves were cut into small pieces in about 1 cm length and packed into a hydrolysis batch reactor with total volume of 2.0 l (the first stage). Water was sprayed over the sugar beet biomass from the top of the reactor and the effluent was collected in a storage tank. The temperature was kept at 35 ◦ C in the hydrolysis reactor, whereas the storage tank was left in room temperature. A small part of the effluent from the storage tank was continuously pumped to a 10-ml flask after passing through a nylon net (50 mesh), whereas the rest of effluent was pumped back to the hydrolysis reactor for recirculation (Fig. 1). The liquid in the flask was sampled intermittently for BODst estimation and fed to an anaerobic contact process with total volume of 1.5 l and operational temperature at 35 ◦ C (the second stage). The liquid in the flask was continuously renewed by pumping the fresh effluent in and sucking the remaining effluent back to the hydrolysis reactor. The flask therefore served as a sampling pool with a hydraulic retention time (HRT) of about 10 min. The whole flask was kept in a cooling water bath at 4 ◦ C to minimize the microbial growth and change of the sample composition. Suspended sludge and produced biogas flowed out of the reactor from an outlet tube on top of the reactor and were separated in a gas–liquid–solid separation unit. The biogas and effluent were released from the reactor, whereas the settled biomass was pumped back for recirculation. The biomethanation reactor was provided with a pH control unit (Inventron AB, Mölndal, Sweden) in order to avoid pH dropping below 7.0 when the system was overloaded. 2.2. Cultivation of microorganisms for the biosensor A sample of activated sludge was collected from the municipal wastewater treatment plant (Källby wastewater treatment plant, Lund, Sweden) and used as inoculum for an activated sludge process in laboratory scale. This process was continuously fed with a synthetic wastewater according to a recipe from the Organization for Economic Cooperation and Development (OECD) (OECD, 1981) at room temperature in order to obtain a suitable microbial population. After the
266
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
Fig. 1. Schematic presentation of the two-stage anaerobic process. The first stage is hydrolysis batch reactor, whereas a biomethanation reactor is referred to as the second stage.
process had been run for 2 weeks, it was considered to have stabilized. The microorganism population was exposed to selection pressure under the conditions applied. A consortium with stable acquired properties for degrading the organic wastes could be obtained after a period of time. The consortium was then used as the biocatalyst of the BOD sensor. 2.3. Instrumentation The analyzer used in this study consists of an in-house-developed biosensor for fast BODst estimation, a microcomputer-based flow injection analysis system for sequential operation, and a computer-based data acquisition system for data recording and processing. The instrument performed a sequence of operations automatically including dilution of a concentrated sample based on a preset ratio up to 200 times in a carrier solution (0.01 M potassium phosphate buffer, pH 7.1), sample analysis, data recording, data processing, and system self-cleaning. Fig. 2 shows a scheme of the BOD sensor system.
The sensor measurements were carried out according to the initial-rate method, i.e. the initial current change (dI/dt) after the sample injection was recorded as the sensor response. This parameter reflects increase of the bacterial respiration rate and, to a certain extent, is proportional to the substrate concentration in the sample. The sensor response was recorded for 60 s followed by a recovery time of less than 10 min. As an alternative, the peak height could be used as sensor signal as well. However, the data recording time should last at least for 90 s in order to obtain the maximum amplitude of current signal decrement. After optimization of the sample injection volume, carrier concentration and flow, cell density and thickness of immobilized biocatalyst layer on top of the dissolved oxygen electrode, as well as selection of a proper calibration solution, the new sensor has a broad linear detection range from 5 to approximately 700 mg BOD5 ·l−1 . Operational stability of the sensor has also been improved by pre-incubating the cells in a controlled condition for both limited nutrients and oxygen supply before immobilization. Furthermore,
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
267
Fig. 2. Scheme of the BOD biosensor system.
the immobilization and replacement procedure of biocatalyst have been very much simplified. This allows a convenient way to renew immobilized cells on a regular basis in order to compensate for the instability of a sensor when the microbial consortium is used as the biocatalyst. Further description of the new BOD sensor and its associated computer-based flow injection system will be described elsewhere (Liu et al., 2004a, 2004b).
ing used to calibrate the BOD sensor for analysis of samples with a similar composition. The analyzer was designed to provide on-line analysis algorithms for estimating a linear equation. By applying three-points calibration as shown in Fig. 3, i.e. analyzing the diluted standard solution with three different concentrations, the instrument will evaluate the linear equation which will be used for converting the sensor signal to the corresponding BOD value during the sample analysis.
2.4. Calibration of the sensor signal 2.5. Sample BODst estimation The BOD biosensor was calibrated against the standard solution which was a stabilized solution from the bioprocess being monitored. The standard solution was prepared by microbial hydrolysis of sugar beet biomass in the following procedure. A series of effluent samples from the hydrolysis reactor was collected during one batch operation over 1 week and stored at −20 ◦ C. The different effluent fractions were thawed, filtered through the nylon net (50 mesh). Equal volumes of the effluent fractions were mixed and diluted 10 times with deionized water. The BOD5 of this mixture was measured to be 682 ± 34 mg·l−1 (n = 12). The solution was then stored at −20 ◦ C and after thaw-
Once the sensor signal was calibrated, the instrument was ready to analyze the real samples. Based on the established linear equation, the computer converted the sensor responses to the corresponding BODst values. To evaluate operational performance of the instrument and the biomethanation reactor, a feeding solution that was prepared in the same manner as the standard solution was used as the feeding solution to the biogas reactor. For monitoring of a two-stage process, the connecting stream between the stages was filtered by the nylon net (50 mesh) before being measured by the
268
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
Fig. 3. An example of three-point calibration curve of the BOD biosensor. BOD5 values of three diluted standard solutions were 13.6, 34.1 and 54.6 mg·l−1 .
BOD-analyzer and used as a feed to the second stage. 2.6. Conventional BOD5 test The 5-day BOD test was carried out according to the American standard procedure (APHA, 1992). 2.7. Biogas analysis The volume of gas produced from the biomethanation reactor was monitored on-line by an in-housedeveloped gas flow meter (Liu et al., 2004c). This meter provides information of gas flow rate (ml·h−1 ) and total gas volume (ml) in real time and is capable to measure a gas flow from 1 ml·h−1 to 950 ml·h−1 . The biogas composition was measured off-line by a Varian 3350 gas chromatograph and was not considered as monitored variable in this study but only an indication of CH4 production.
3. Results and discussion Before applying the BOD analyzer for process monitoring, the operational stability of the instrument was
tested by analyzing a diluted standard solution with BOD5 value in 20 mg·l−1 repeatedly over 4.5 days (45 assays). This standard solution was prepared with an additional filtration step (pore size: 0.45 m, Sartorius, Sweden) for removing tiny particles and microbial cells. As demonstrated in Fig. 4, the analyzer gave fairly constant BODst estimation over the whole testing period. The average value was 20 mg·l−1 with a standard deviation at ±1.0 mg·l−1 . The analyzer was then used to estimate the OLR to the biomethanation reactor. Three kinds of loading were studied. In the first two cases, the feeding medium was the stabilized solution from the bioprocess being monitored. In the last case, the feeding medium was the actual effluent from the first-stage reactor. It should be noticed that biogas production was directly related to the organic matter removal instead of the organic loading. However, when the organic loading is kept low and does not exceed the assimilation capacity of the microorganisms, bacteria will assimilate most of organic substrates that have been fed into the reactor. As a consequence, the amount of substrates that remains in the process effluent will be very low. The assimilated substrates can be considered to leave the process through two ports, i.e. mainly for biogas production and only a small
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
269
Fig. 4. BODst estimation of a diluted standard solution in 20 mg·l−1 over a time period of 4.5 days.
portion as excess biomass. Since the BOD-analyzer was not suitable for analyzing samples with different degrees of dilution simultaneously, automatic analysis of both the feeding stream and the effluent from the biomethanation reactor using one BOD analyzer was not possible in this study. On-line monitoring of the feeding stream was then selected, whereas the effluent quality was checked occasionally with the conventional BOD method after filtering the effluent through the same nylon net. The effluent quality test was only used for indicating the successful organic matter removal in this study. In the first two experiments, the −1 OLRs were less than 0.6 g BODst ·lreactor · d−1 , which is much smaller than a substrate conversion capacity −1 of 3–5 g COD·lreactor · d−1 reported for the anaerobic contact process (Armenante, 1993). The biogas production could therefore be considered relating to the organic loading based on an assumption that the substrates were completely assimilated by the bacteria. 3.1. Monitoring of the biomethanation reactor with a discrete organic loading The biomethanation reactor was fed intermittently with the feeding solution. Three kinds of OLR per re-
actor volume have been tested in turn with a time in−1 · terval of 12 h, i.e. 0.59, 0.46 and 0.41 g BODst ·lreactor d−1 . The HRT was kept at 1.2 days during the loading periods and zero loading was obtained by simply adjusting the flow rate to zero. The BOD analyzer was used to analyze samples of the feeding solution every 30 min. The gas production rate was measured every hour and the gas composition was analyzed twice during each loading period. The average methane content was 80.1% with a standard deviation of 4.9%. Organic matter in the effluent was checked at the end of each loading period. The BOD5 value was 54 ± 7 mg·l−1 . Fig. 5a and b illustrate a relation of the gas production rate and the OLR that was calculated based on the estimated BODst values and flow rate of the feeding stream. As shown in the figures, the gas formation was closely correlated to the estimated OLR. The biogas was produced increasingly after starting the organic load. A slight time delay of gas production was observed initially due to the time required for activating the bacteria that converts the biomass into CH4 and CO2 . The outlet tube was placed with its mouth directly below the surface of the fermentation broth. Due to the surface tension between the outlet tube wall and fermentation broth, the continuously produced biogas
270
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
Fig. 5. (a) Relation of the gas production rate and the organic loading rate. The loading was intermittent with a time interval of 12 h. (䊊) biogas production rate, () organic loading rate. (b) Relation of the average gas production rate and the average organic loading rate. The solid column represents the average organic loading rate, (䊊) average gas production rate.
was not evenly released from the reactor. In contrast, the gas was accumulated inside the reactor headspace and the fermentation broth was displaced until the mouth of the outlet tube was above the broth surface to allow passage of the gas. As a result, the biogas left the reactor intermittently. This was particularly obvious when the volume of biogas produced was small. Therefore the curve of gas production rate looks like vibrat-
ing up and down. Even then, it can be observed that the volume of produced gas (i.e. represented by the peak areas) during the loading period decreased correspondingly with the decreased organic load. Moreover, the gas production was gradually reduced for a period of a few hours after stopping the organic load until it eventually stopped completely. This phenomenon was due to the anaerobic biodegradation of the organic residues
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
271
Fig. 6. Correlation between gas formation and organic loading. (䊊) biogas production, () organic loading.
that were fed in the reactor previously. Once these organic residues have been degraded completely, the gas formation was stopped. Fig. 6 illustrates a relation between the total organic loading and the total biogas production. The vibration of biogas production curve is not obvious anymore owing to an enlarged scope of magnitude. A good correlation can be observed between the organic loading and gas production, and the above-mentioned phenomenon can also be viewed in the figure. The biogas yields during the loading period were 0.49, 0.47 and 0.44 lgas ·g BOD−1 st , respectively. 3.2. Monitoring of the biomethanation reactor with an increasing organic loading Operation of the biogas reactor was shifted to a continuous feeding with an increasing organic load. The OLR started at zero and increased by one step every 12 h. As shown in Fig. 7a and b, six different OLRs have been tested, i.e. 0, 0.180, 0.305, 0.384, 0.461 −1 and 0.566 g BODst ·lreactor · d−1 . The HRT was kept at 1.7 days during the loading period and zero loading was obtained by simply bringing the flow rate to zero. The BOD analyzer was used to measure samples of the feeding solution every 30 min. The gas production rate was analyzed once an hour and gas composition was analyzed twice during each loading step. The methane content was 90.4 ± 9.2%. A conventional
BOD test was carried out on the effluent at the end of each loading period and the BOD5 value was found to be 57 ± 8 mg·l−1 . As demonstrated in the figures, the gas formation rate increased correspondingly with the increased OLR. Once the OLR was shifted back from the highest value to zero, the gas production was stopped after the remaining organic substrates had been consumed completely. Fig. 8 shows the correlation between the total biogas production and the total organic loading. The biogas yield was progressively increased and followed the tendency of the increased OLR over the loading period with an average value of 0.52 lgas ·g BOD−1 st . However, the fermentation process deferred the gas production for short time, especially in the initial period of loading. This is due to the initial time delay for activating the bacteria that converts the biomass into CH4 and CO2 . 3.3. Monitoring of the biomethanation reactor with varied organic loading In this study, the hydrolysis batch reactor and biogas reactor were operated simultaneously. The biomass of sugar beet leaves was hydrolyzed and converted into volatile fatty acids and other low molecular weight compounds in the first stage. The effluent from this stage was analyzed by the BOD analyzer every 30 min and used as a feed stream to the second stage (the
272
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
Fig. 7. (a) Relation of the gas production rate and the increased organic loading rate. The loading rate started at 0 and increased in steps every 12 h. (䊊) biogas production rate, () organic loading rate. (b) Relation of the average gas production rate and the average organic loading rate. The solid column presents the average organic loading rate, (䊊) average gas production rate.
anaerobic contact digester) with a fixed flow rate in 0.3 ml·min−1 . The HRT was 3.5 days. The gas production rate was analyzed once an hour, and the gas composition was analyzed twice a day. The methane content over the period of loading was 81.0±3.2%. Based on the estimated BODst values, flow rate of the feed stream, and flow rate of biogas, a correlation of biogas production rate and OLR was figured out (Fig. 9). During the operational period of 120–264 h, good correction between OLR and the gas production rate can
be observed. However, there is a high biogas yield period between 70 and 120 h with an average value of 0.62 lgas ·g BOD−1 st . This is owing to the extra biogas formation from the organic substrates that had been accumulated in the reactor during a high organic feeding period between 25 and 70 h with an average value of −1 2.59 g BODst ·lreactor · d−1 . The microorganisms could not utilize all of the organic substrate that had been fed in during the high feeding period. Since there was a pH control unit for avoiding the pH dropping in the
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
273
Fig. 8. Correlation between gas formation and organic loading. (䊊) biogas production, () organic loading.
reactor, the volatile fatty acids and other low molecular weight compound not only accumulated in the reactor without inhibiting the subsequence fermentation steps, but also became the substrates for the extra gas formation in the following period. During an initial period between 0 and 25 h, the gas production
rate did not correlate well to the OLR. It was noticed that during this period the effluent samples were collected right after packing the biomass of sugar beet leaves into the reactor. The samples therefore contained tissue juice of fresh biomass, which was a complex mixture including high content of polymers such
Fig. 9. Relation of the gas production rate and the varied organic loading rate. (䊊) biogas production rate, () organic loading rate.
274
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
as polysaccharides, fats and proteins (Kaseng et al., 1992; Björnsson, 2000). These polymers might not be able to diffuse through the dialysis membrane or to be assimilated immediately by the immobilized cells in the BOD sensor. As a result, an underestimation of BOD value from the sensor occurred. Consequently, it resulted in underestimation of the OLR to the second stage. In contrast, the polymer of sugar beet biomass could be split into smaller units in the first-stage reactor over time. A better BOD estimation was therefore obtained after the initial operational period. This phenomenon was also observed in the early stage of this study (Liu et al., 2003).
paid to the samples with high content of organic polymers where underestimation of BOD values might occur. Of course, the use of this analyzer is not limited to this application example. It can also be used for on-line quality control of effluents in numbers of both municipal and industrial wastewater treatment plants that includes both aerobic and anaerobic operations. This will be particularly valuable to wastewater treatment plants since effluent standards are getting tighter. A rapid and frequent feedback of the effluent analysis will be obligated to tune the plant operation properly in order to meet tomorrow’s regulations.
Acknowledgements 4. Conclusions Monitoring of intermediate metabolites from anaerobic decomposition processes is important since it generates signals that can be used to monitor and control the process for stable start-up and steady-state running operations. Since it has been difficult to monitor such a process on-line, many biogas systems are operated far below their maximal reactor capacity due to a desire to eliminate the risk of system overload. If that happens, the methanogens may get killed and one then has to reinoculate the reactor and start from the very beginning. The strategy to overcome the problem is to combine a suitable reactor design with a closely monitored and controlled process in order to exploit the cost benefit of a smaller anaerobic system and allow waste treatment at a higher specific rate. Among the commonly suggested monitoring parameters, one that indicates the intermediate products (e.g. easily assimilable organic compounds) and consequently reflects a loading rate of biodegradable organic material should preferably be included. This paper reports an example using the stand-alone BOD analyzer for on-line monitoring of the intermediate products in an anaerobic digestion process. The experimental results demonstrate successful utilization of this sensor device and the BOD biosensor method can be a useful and valuable on-line method for fast indication of easily biodegradable organic substances from the initial step of a two-stage anaerobic process. The on-line BODst estimation can then be used for keeping the OLR under control in order to avoid acidification of the methanogenic step. However, attention should be
The Swedish Agency for Research Co-operation with Developing Countries (SAREC), the Swedish National Energy Administration (STEM), and the Energy Supply Committee of Southern Sweden (DESS) are gratefully acknowledged for the financial support.
References Ahring, B., Sandberg, M., Angelidaki, I., 1995. Volatile fatty acids as indicators of process imbalance in anaerobic digestors. Appl. Microbiol. Biotechnol. 43 (3), 559–565. APHA, 1992. Standard Methods for the Examination of Water and Wastewater, Section 5210, 18th ed. American Public Health Association Water Works Association, American Water Environment Federation, Washington, DC, pp. 5.1–5.6. Armenante, P.M., 1993. Bioreactors. In: Levin, M.A., Gealt, M.A. (Eds.), Biotreatment of Industrial and Hazardous Waste. McGraw-Hill, New York, NY, pp. 65–112. Björnsson, L., 2000. Intensification of the biogas process by improved process monitoring and biomass retention. Doctoral Dissertation, Department of Biotechnology, Lund University, Sweden. Björnsson, L., Hörnsten, E.G., Mattiasson, B., 2001. Utilization of a palladium-metal oxide semiconductor (Pd-MOS) sensor for on-line monitoring of dissolved hydrogen in anaerobic digestion. Biotechnol. Bioeng. 73 (1), 35–43. Cobb, S.A., Hill, D.T., 1991. Volatile fatty acid relationships in attached growth anaerobic fermenters. Trans. ASAE 34 (6), 2564–2572. Cord-Ruwisch, R., Mercz, T.I., Hoh, C.-Y., Strong, G.E., 1997. Dissolved hydrogen concentration as an on-line control parameter for the automated operation and optimization of anaerobic digesters. Biotechnol. Bioeng. 56 (6), 626–634. Ghosh, S., 1991. Pilot-scale demonstration of two-phase anaerobic digestion of activated sludge. Water Sci. Technol. 23 (7–9), 1179–1188.
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275 Hickey, R.F., Switzenbaum, M.S., 1991. Thermodynamics of volatile fatty acid accumulation in anaerobic digesters subject to increases in hydraulic and organic loading. Res. J. Water Pollut. Control Fed. 63 (2), 141–144. Kaseng, K., Ibrahim, K., Paneerselvam, S.V., Hassan, R.S., 1992. Extracellular enzymes and acidogen profiles of a laboratory-scale two-phase anaerobic digestion system. Process Biochem. 27 (1), 43–47. Liu, J., Björnsson, L., Mattiasson, B., 2000. Immobilised activated sludge based biosensor for biochemical oxygen demand measurement. Biosens. Bioelectron. 14 (12), 883–893. Liu, J., Mattiasson, B., 2002. Microbial BOD sensors for wastewater analysis. Water Res. 36 (15), 3786–3802. Liu, J., Olsson, G., Mattiasson, B., 2003. Monitoring of two-stage anaerobic biodegradation using a BOD biosensor. J. Biotechnol. 100 (3), 261–265. Liu, J., Olsson, G., Mattiasson, B., 2004a. Short-term BOD (BODst ) as a parameter for on-line monitoring of biological treatment process, Part I: A novel design of BOD biosensor for easy renewal of bio-layer. Biosens. Bioelectron. (in press). Liu, J., Olsson, G., Mattiasson, B., 2004b. Short-term BOD (BODst ) as a parameter for on-line monitoring of biological treatment process, Part II: Instrumentation of integrated flow
275
injection analysis (FIA) system for BODst estimation. Biosens. Bioelectron. (in press). Liu, J., Olsson, G., Mattiasson, B., 2004c. A volumetric meter for monitoring of low gas flow rate from laboratory-scale biogas reactor. Sens. Actuators B 97, 369–372. Moletta, R., Escoffier, Y., Ehlinger, F., Coudert, J.-P., Leyris, J.-P., 1994. On-line automatic control system for monitoring an anaerobic fluidized-bed reactor: response to organic overload. Water Sci. Technol. 30 (12), 11–20. Organisation for Economic Cooperation and Development (OECD), 1981. Activated sludge, respiration inhibition test. OECD Guidelines Testing Chem. 209, 1–10. SIS-Standardiseringskommissionen i Sverige, 1979. Water analysis—Determination of Biochemical Oxygen Demand, BOD, of Water—Dilution Method (Svensk Standard SS 02 81 43 E), first ed. The Swedish Standards Institution, Stockholm, pp. 1–9. Spanjers, H., Olsson, G., Klapwijk, A., 1993. Determining influent short-term biochemical oxygen demand by combined respirometry and estimation. Water Sci. Technol. 28 (11/12), 401–414. Speece, R.E., 1996. Operational considerations. In: Speece, R.E. (Ed.), Anaerobic Biotechnology for Industrial Wastewaters. Archae Press, pp. 69–112.
On-line monitoring of a two-stage anaerobic digestion process using a BOD analyzer夽 Jing Liu a,∗ , Gustaf Olsson b , Bo Mattiasson a a
Department of Biotechnology, Center for Chemistry and Chemical Engineering, Lund University, P.O. Box 124, S-221 00 Lund, Sweden b Department of Industrial Electrical Engineering and Automation, Lund University, P.O. Box 118, S-221 00 Lund, Sweden Received 21 July 2003; accepted 18 November 2003
Abstract A computer-controlled biochemical oxygen demand (BOD) analyzer has been developed for fast estimation of biochemical oxygen demand (BODst ) automatically with the purpose of on-line monitoring of a process for conversion of biomass under field conditions. The instrument was tested by on-line monitoring of the connecting stream between two stages of a two-stage anaerobic process in laboratory scale. In the first stage, hydrolysis of sugar beet leaves and its conversion into volatile fatty acids and other low molecular weight substrates took place. The effluent from the first reactor was used as a feed stream to the second stage, i.e. an anaerobic contact reactor. The feed stream was sampled intermittently, diluted and analyzed by the BOD analyzer automatically in order to estimate the organic loading rate to the reactor. The results from this study demonstrated that the BOD analyzer could be a stand-alone and promising sensor device for rapid on-line monitoring of easily biodegradable organic substances in biological treatment processes. © 2004 Elsevier B.V. All rights reserved. Keywords: BOD analyzer; BODst ; Monitoring; On-line; Two-stage; Anaerobic digestion
1. Introduction In recent years, much of the research pertaining to anaerobic processes has been directed towards identifying parameters, which may be used to monitor the process and give an early warning of system instability (Cobb and Hill, 1991; Cord-Ruwisch et al., 1997). It has been identified that the activity of different 夽 Part of this work was presented at the Third International Symposium on Anaerobic Digestion of Solid Wastes, Munich/Garching, Germany, September 18–20, 2002. ∗ Corresponding author. Tel.: +46-228347; fax: +46-2224713. E-mail address: [email protected] (J. Liu).
groups of organisms involved in the anaerobic process can be measured indirectly by monitoring the metabolites. Some of the commonly suggested indicators include pH, alkalinity, concentration of volatile fatty acids (VFAs), dissolved organic monomers, dissolved hydrogen, gas production rate, as well as gas composition (Ahring et al., 1995; Ghosh, 1991; Hickey and Switzenbaum, 1991; Björnsson et al., 2001; Moletta et al., 1994). It is, however, essential to develop on-line methods for these parameters in order to indicate the status of the processes in real time. This paper presents an application of a computercontrolled analysis system which is based on a biosensor method for estimating the easily biodegradable
0168-1656/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jbiotec.2003.11.014
264
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
organic compounds that can give a fast indication of intermediate products from the initial step of an anaerobic process.
ated far below maximum reactor capacity to eliminate the risk of system overload.
1.1. Anaerobic biodegradation process
1.2. Approaches for improving the stability and efficiency of anaerobic process
The anaerobic decomposition of organic matter into methane and carbon dioxide is a multi-step process involving a well-organized community of various microbial populations. In the first step, organic polymers (e.g. polysaccharides, fats and proteins) are hydrolyzed into smaller units (e.g. sugars, long-chain fatty acid and amino acids) by extracellular hydrolases excreted by obligate or facultative anaerobes. The products resulting from the hydrolysis step are subsequently used as substrates in the next step which is often referred to as acidogenesis. This degradation pathway is often the fastest step and also gives a high-energy yield for the microorganisms. A wide variety of different microbes representing both obligate and facultative anaerobes are involved. The fermentation products resulting from the acidogenic steps include acetate and reduced intermediates (electron sinks) such as lactate, ethanol, propionate, butyrate and higher volatile fatty acids. Conversion of these electron sinks to acetate, carbon dioxide and hydrogen is then carried out in the acetogenic steps by obligate hydrogen-producing acetogenic bacteria. Because of a low-energy yield from the acetogenic degradation, acetogenic microorganisms are very slow-growing and sensitive to changes in organic load, flow rate and environmental changes. The substrates that can be used in the final methanogenic step are acetate, H2 /CO2 , methanol and formate. The most important methanogenic transformations are the acetoclastic reaction and the reduction of carbon dioxide. Very few known methanogens can perform the acetoclastic methane production, whereas nearly all known methanogenic species are able to produce methane from H2 /CO2 (Björnsson, 2000). Owing to the limiting energy available during anaerobic degradation, some of the microbial populations have a low growth rate, which makes them vulnerable and sensitive to changes in operational conditions. This can cause instability during both the start-up and steady-state running operations of a biomethanation process. Consequently, many biogas systems are oper-
One approach to overcome the problem is close monitoring the process with regard to selected parameters and using these signals for improving the operational performance by a better process control. Furthermore, the different microbial populations involved in anaerobic digestion do not have the same requirements on operational conditions. The growth rates and pH optima are different for acidogenic and methanogenic organisms. In a one-stage digester, the pH and the organic loading rate (OLR) are adjusted to suit the slow-growing methanogenic organisms at the expense of the relatively fast-growing acidogens and process efficiency as a whole. The operational conditions for the different microbial groups can be further optimized if the process is divided into two stages in separate reactors. The hydrolysis and acidogenesis take place in the first reactor. The effluent from this reactor is used as a feed stream to the second stage, a biomethanation reactor. The methanogens in the second stage can be effectively protected from overload and toxic shocks by close monitoring of the connecting stream. A monitoring parameter that can indicate the intermediate products (e.g. sugars, long-chain fatty acid and amino acids) from the first stage and consequently reflect a loading of easily biodegradable organic material to the biomethanation reactor could be selected with a high priority. The biochemical oxygen demand (BOD) determination has its widest application in measuring loading of biodegradable organic material to wastewater treatment plants and in evaluating the BOD removal efficiency of such a treatment process. It is an empirical test in which a standardized laboratory procedure is used to determine the relative oxygen requirements of wastewaters. The determination of conventional BOD includes the well-known BOD5 (APHA, 1992) and BOD7 (SIS, 1979) tests, which need 5 or 7 days of incubation at 20 ± 1 ◦ C in the dark, respectively. Since most organic pollutants can be biodegraded under both aerobic and anaerobic treatment, this type of assay also has value when it is applied to the anaerobic treatment. However, electron acceptors other than oxygen
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
are involved in anaerobic degradation. BOD5 can be a valid parameter for indicating the biodegradable and assimilable organic load to an anaerobic process and a possible subsequent treatment unit, such as an aerobic polishing unit, or a receiving surface body of water, and is crucial for evaluation of the process yield and stability (Speece, 1996). The conventional BOD test is time consuming and therefore not suitable for the purpose of on-line process monitoring. Fast determination of BOD can also be achieved using biosensor methods based on the respirometric principle. However, the use of BOD biosensors results in short-term BOD (BODst ) values that are not identical with the conventional BOD values in all cases. The microbial oxygen consumption measured by the conventional method is the sum of the oxygen used to oxidize both easily assimilable compounds and biodegradable polymers (e.g. starch and proteins), whereas BOD biosensors can normally only give responses to those fast and easily assimilable compounds in the samples (Liu and Mattiasson, 2002). Nevertheless, BODst seems to be a more valuable parameter for control purposes because it represents the assimilable organic matter supply within the time constraint, and is intimately related to the dynamic performance of the process. The BOD5 , on the other hand, merely characterizes the impact on the receiving water (Spanjers et al., 1993). 1.3. Outline of the work This paper presents a follow-up study from the earlier work (Liu et al., 2003). In order to overcome the limitations of a previously developed BOD biosensor (Liu et al., 2000), such as operation within a narrow linear concentration range and poor stability, a new sensor and an associated computer-based flow injection system have been designed and constructed. In this study, the in-house developed BOD analyzer was applied to on-line analysis of the connecting stream between the two stages of a two-stage anaerobic process in order to estimate OLR to the biogas reactor. The correlation between OLR and gas formation was studied for a discrete loading, a continuous loading with a relatively constant content of organic matter during each loading period, and a continuous loading with varied content of organic matter over time.
265
2. Methods 2.1. Two-stage anaerobic reactor Fresh sugar beet leaves were cut into small pieces in about 1 cm length and packed into a hydrolysis batch reactor with total volume of 2.0 l (the first stage). Water was sprayed over the sugar beet biomass from the top of the reactor and the effluent was collected in a storage tank. The temperature was kept at 35 ◦ C in the hydrolysis reactor, whereas the storage tank was left in room temperature. A small part of the effluent from the storage tank was continuously pumped to a 10-ml flask after passing through a nylon net (50 mesh), whereas the rest of effluent was pumped back to the hydrolysis reactor for recirculation (Fig. 1). The liquid in the flask was sampled intermittently for BODst estimation and fed to an anaerobic contact process with total volume of 1.5 l and operational temperature at 35 ◦ C (the second stage). The liquid in the flask was continuously renewed by pumping the fresh effluent in and sucking the remaining effluent back to the hydrolysis reactor. The flask therefore served as a sampling pool with a hydraulic retention time (HRT) of about 10 min. The whole flask was kept in a cooling water bath at 4 ◦ C to minimize the microbial growth and change of the sample composition. Suspended sludge and produced biogas flowed out of the reactor from an outlet tube on top of the reactor and were separated in a gas–liquid–solid separation unit. The biogas and effluent were released from the reactor, whereas the settled biomass was pumped back for recirculation. The biomethanation reactor was provided with a pH control unit (Inventron AB, Mölndal, Sweden) in order to avoid pH dropping below 7.0 when the system was overloaded. 2.2. Cultivation of microorganisms for the biosensor A sample of activated sludge was collected from the municipal wastewater treatment plant (Källby wastewater treatment plant, Lund, Sweden) and used as inoculum for an activated sludge process in laboratory scale. This process was continuously fed with a synthetic wastewater according to a recipe from the Organization for Economic Cooperation and Development (OECD) (OECD, 1981) at room temperature in order to obtain a suitable microbial population. After the
266
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
Fig. 1. Schematic presentation of the two-stage anaerobic process. The first stage is hydrolysis batch reactor, whereas a biomethanation reactor is referred to as the second stage.
process had been run for 2 weeks, it was considered to have stabilized. The microorganism population was exposed to selection pressure under the conditions applied. A consortium with stable acquired properties for degrading the organic wastes could be obtained after a period of time. The consortium was then used as the biocatalyst of the BOD sensor. 2.3. Instrumentation The analyzer used in this study consists of an in-house-developed biosensor for fast BODst estimation, a microcomputer-based flow injection analysis system for sequential operation, and a computer-based data acquisition system for data recording and processing. The instrument performed a sequence of operations automatically including dilution of a concentrated sample based on a preset ratio up to 200 times in a carrier solution (0.01 M potassium phosphate buffer, pH 7.1), sample analysis, data recording, data processing, and system self-cleaning. Fig. 2 shows a scheme of the BOD sensor system.
The sensor measurements were carried out according to the initial-rate method, i.e. the initial current change (dI/dt) after the sample injection was recorded as the sensor response. This parameter reflects increase of the bacterial respiration rate and, to a certain extent, is proportional to the substrate concentration in the sample. The sensor response was recorded for 60 s followed by a recovery time of less than 10 min. As an alternative, the peak height could be used as sensor signal as well. However, the data recording time should last at least for 90 s in order to obtain the maximum amplitude of current signal decrement. After optimization of the sample injection volume, carrier concentration and flow, cell density and thickness of immobilized biocatalyst layer on top of the dissolved oxygen electrode, as well as selection of a proper calibration solution, the new sensor has a broad linear detection range from 5 to approximately 700 mg BOD5 ·l−1 . Operational stability of the sensor has also been improved by pre-incubating the cells in a controlled condition for both limited nutrients and oxygen supply before immobilization. Furthermore,
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
267
Fig. 2. Scheme of the BOD biosensor system.
the immobilization and replacement procedure of biocatalyst have been very much simplified. This allows a convenient way to renew immobilized cells on a regular basis in order to compensate for the instability of a sensor when the microbial consortium is used as the biocatalyst. Further description of the new BOD sensor and its associated computer-based flow injection system will be described elsewhere (Liu et al., 2004a, 2004b).
ing used to calibrate the BOD sensor for analysis of samples with a similar composition. The analyzer was designed to provide on-line analysis algorithms for estimating a linear equation. By applying three-points calibration as shown in Fig. 3, i.e. analyzing the diluted standard solution with three different concentrations, the instrument will evaluate the linear equation which will be used for converting the sensor signal to the corresponding BOD value during the sample analysis.
2.4. Calibration of the sensor signal 2.5. Sample BODst estimation The BOD biosensor was calibrated against the standard solution which was a stabilized solution from the bioprocess being monitored. The standard solution was prepared by microbial hydrolysis of sugar beet biomass in the following procedure. A series of effluent samples from the hydrolysis reactor was collected during one batch operation over 1 week and stored at −20 ◦ C. The different effluent fractions were thawed, filtered through the nylon net (50 mesh). Equal volumes of the effluent fractions were mixed and diluted 10 times with deionized water. The BOD5 of this mixture was measured to be 682 ± 34 mg·l−1 (n = 12). The solution was then stored at −20 ◦ C and after thaw-
Once the sensor signal was calibrated, the instrument was ready to analyze the real samples. Based on the established linear equation, the computer converted the sensor responses to the corresponding BODst values. To evaluate operational performance of the instrument and the biomethanation reactor, a feeding solution that was prepared in the same manner as the standard solution was used as the feeding solution to the biogas reactor. For monitoring of a two-stage process, the connecting stream between the stages was filtered by the nylon net (50 mesh) before being measured by the
268
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
Fig. 3. An example of three-point calibration curve of the BOD biosensor. BOD5 values of three diluted standard solutions were 13.6, 34.1 and 54.6 mg·l−1 .
BOD-analyzer and used as a feed to the second stage. 2.6. Conventional BOD5 test The 5-day BOD test was carried out according to the American standard procedure (APHA, 1992). 2.7. Biogas analysis The volume of gas produced from the biomethanation reactor was monitored on-line by an in-housedeveloped gas flow meter (Liu et al., 2004c). This meter provides information of gas flow rate (ml·h−1 ) and total gas volume (ml) in real time and is capable to measure a gas flow from 1 ml·h−1 to 950 ml·h−1 . The biogas composition was measured off-line by a Varian 3350 gas chromatograph and was not considered as monitored variable in this study but only an indication of CH4 production.
3. Results and discussion Before applying the BOD analyzer for process monitoring, the operational stability of the instrument was
tested by analyzing a diluted standard solution with BOD5 value in 20 mg·l−1 repeatedly over 4.5 days (45 assays). This standard solution was prepared with an additional filtration step (pore size: 0.45 m, Sartorius, Sweden) for removing tiny particles and microbial cells. As demonstrated in Fig. 4, the analyzer gave fairly constant BODst estimation over the whole testing period. The average value was 20 mg·l−1 with a standard deviation at ±1.0 mg·l−1 . The analyzer was then used to estimate the OLR to the biomethanation reactor. Three kinds of loading were studied. In the first two cases, the feeding medium was the stabilized solution from the bioprocess being monitored. In the last case, the feeding medium was the actual effluent from the first-stage reactor. It should be noticed that biogas production was directly related to the organic matter removal instead of the organic loading. However, when the organic loading is kept low and does not exceed the assimilation capacity of the microorganisms, bacteria will assimilate most of organic substrates that have been fed into the reactor. As a consequence, the amount of substrates that remains in the process effluent will be very low. The assimilated substrates can be considered to leave the process through two ports, i.e. mainly for biogas production and only a small
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
269
Fig. 4. BODst estimation of a diluted standard solution in 20 mg·l−1 over a time period of 4.5 days.
portion as excess biomass. Since the BOD-analyzer was not suitable for analyzing samples with different degrees of dilution simultaneously, automatic analysis of both the feeding stream and the effluent from the biomethanation reactor using one BOD analyzer was not possible in this study. On-line monitoring of the feeding stream was then selected, whereas the effluent quality was checked occasionally with the conventional BOD method after filtering the effluent through the same nylon net. The effluent quality test was only used for indicating the successful organic matter removal in this study. In the first two experiments, the −1 OLRs were less than 0.6 g BODst ·lreactor · d−1 , which is much smaller than a substrate conversion capacity −1 of 3–5 g COD·lreactor · d−1 reported for the anaerobic contact process (Armenante, 1993). The biogas production could therefore be considered relating to the organic loading based on an assumption that the substrates were completely assimilated by the bacteria. 3.1. Monitoring of the biomethanation reactor with a discrete organic loading The biomethanation reactor was fed intermittently with the feeding solution. Three kinds of OLR per re-
actor volume have been tested in turn with a time in−1 · terval of 12 h, i.e. 0.59, 0.46 and 0.41 g BODst ·lreactor d−1 . The HRT was kept at 1.2 days during the loading periods and zero loading was obtained by simply adjusting the flow rate to zero. The BOD analyzer was used to analyze samples of the feeding solution every 30 min. The gas production rate was measured every hour and the gas composition was analyzed twice during each loading period. The average methane content was 80.1% with a standard deviation of 4.9%. Organic matter in the effluent was checked at the end of each loading period. The BOD5 value was 54 ± 7 mg·l−1 . Fig. 5a and b illustrate a relation of the gas production rate and the OLR that was calculated based on the estimated BODst values and flow rate of the feeding stream. As shown in the figures, the gas formation was closely correlated to the estimated OLR. The biogas was produced increasingly after starting the organic load. A slight time delay of gas production was observed initially due to the time required for activating the bacteria that converts the biomass into CH4 and CO2 . The outlet tube was placed with its mouth directly below the surface of the fermentation broth. Due to the surface tension between the outlet tube wall and fermentation broth, the continuously produced biogas
270
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
Fig. 5. (a) Relation of the gas production rate and the organic loading rate. The loading was intermittent with a time interval of 12 h. (䊊) biogas production rate, () organic loading rate. (b) Relation of the average gas production rate and the average organic loading rate. The solid column represents the average organic loading rate, (䊊) average gas production rate.
was not evenly released from the reactor. In contrast, the gas was accumulated inside the reactor headspace and the fermentation broth was displaced until the mouth of the outlet tube was above the broth surface to allow passage of the gas. As a result, the biogas left the reactor intermittently. This was particularly obvious when the volume of biogas produced was small. Therefore the curve of gas production rate looks like vibrat-
ing up and down. Even then, it can be observed that the volume of produced gas (i.e. represented by the peak areas) during the loading period decreased correspondingly with the decreased organic load. Moreover, the gas production was gradually reduced for a period of a few hours after stopping the organic load until it eventually stopped completely. This phenomenon was due to the anaerobic biodegradation of the organic residues
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
271
Fig. 6. Correlation between gas formation and organic loading. (䊊) biogas production, () organic loading.
that were fed in the reactor previously. Once these organic residues have been degraded completely, the gas formation was stopped. Fig. 6 illustrates a relation between the total organic loading and the total biogas production. The vibration of biogas production curve is not obvious anymore owing to an enlarged scope of magnitude. A good correlation can be observed between the organic loading and gas production, and the above-mentioned phenomenon can also be viewed in the figure. The biogas yields during the loading period were 0.49, 0.47 and 0.44 lgas ·g BOD−1 st , respectively. 3.2. Monitoring of the biomethanation reactor with an increasing organic loading Operation of the biogas reactor was shifted to a continuous feeding with an increasing organic load. The OLR started at zero and increased by one step every 12 h. As shown in Fig. 7a and b, six different OLRs have been tested, i.e. 0, 0.180, 0.305, 0.384, 0.461 −1 and 0.566 g BODst ·lreactor · d−1 . The HRT was kept at 1.7 days during the loading period and zero loading was obtained by simply bringing the flow rate to zero. The BOD analyzer was used to measure samples of the feeding solution every 30 min. The gas production rate was analyzed once an hour and gas composition was analyzed twice during each loading step. The methane content was 90.4 ± 9.2%. A conventional
BOD test was carried out on the effluent at the end of each loading period and the BOD5 value was found to be 57 ± 8 mg·l−1 . As demonstrated in the figures, the gas formation rate increased correspondingly with the increased OLR. Once the OLR was shifted back from the highest value to zero, the gas production was stopped after the remaining organic substrates had been consumed completely. Fig. 8 shows the correlation between the total biogas production and the total organic loading. The biogas yield was progressively increased and followed the tendency of the increased OLR over the loading period with an average value of 0.52 lgas ·g BOD−1 st . However, the fermentation process deferred the gas production for short time, especially in the initial period of loading. This is due to the initial time delay for activating the bacteria that converts the biomass into CH4 and CO2 . 3.3. Monitoring of the biomethanation reactor with varied organic loading In this study, the hydrolysis batch reactor and biogas reactor were operated simultaneously. The biomass of sugar beet leaves was hydrolyzed and converted into volatile fatty acids and other low molecular weight compounds in the first stage. The effluent from this stage was analyzed by the BOD analyzer every 30 min and used as a feed stream to the second stage (the
272
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
Fig. 7. (a) Relation of the gas production rate and the increased organic loading rate. The loading rate started at 0 and increased in steps every 12 h. (䊊) biogas production rate, () organic loading rate. (b) Relation of the average gas production rate and the average organic loading rate. The solid column presents the average organic loading rate, (䊊) average gas production rate.
anaerobic contact digester) with a fixed flow rate in 0.3 ml·min−1 . The HRT was 3.5 days. The gas production rate was analyzed once an hour, and the gas composition was analyzed twice a day. The methane content over the period of loading was 81.0±3.2%. Based on the estimated BODst values, flow rate of the feed stream, and flow rate of biogas, a correlation of biogas production rate and OLR was figured out (Fig. 9). During the operational period of 120–264 h, good correction between OLR and the gas production rate can
be observed. However, there is a high biogas yield period between 70 and 120 h with an average value of 0.62 lgas ·g BOD−1 st . This is owing to the extra biogas formation from the organic substrates that had been accumulated in the reactor during a high organic feeding period between 25 and 70 h with an average value of −1 2.59 g BODst ·lreactor · d−1 . The microorganisms could not utilize all of the organic substrate that had been fed in during the high feeding period. Since there was a pH control unit for avoiding the pH dropping in the
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
273
Fig. 8. Correlation between gas formation and organic loading. (䊊) biogas production, () organic loading.
reactor, the volatile fatty acids and other low molecular weight compound not only accumulated in the reactor without inhibiting the subsequence fermentation steps, but also became the substrates for the extra gas formation in the following period. During an initial period between 0 and 25 h, the gas production
rate did not correlate well to the OLR. It was noticed that during this period the effluent samples were collected right after packing the biomass of sugar beet leaves into the reactor. The samples therefore contained tissue juice of fresh biomass, which was a complex mixture including high content of polymers such
Fig. 9. Relation of the gas production rate and the varied organic loading rate. (䊊) biogas production rate, () organic loading rate.
274
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275
as polysaccharides, fats and proteins (Kaseng et al., 1992; Björnsson, 2000). These polymers might not be able to diffuse through the dialysis membrane or to be assimilated immediately by the immobilized cells in the BOD sensor. As a result, an underestimation of BOD value from the sensor occurred. Consequently, it resulted in underestimation of the OLR to the second stage. In contrast, the polymer of sugar beet biomass could be split into smaller units in the first-stage reactor over time. A better BOD estimation was therefore obtained after the initial operational period. This phenomenon was also observed in the early stage of this study (Liu et al., 2003).
paid to the samples with high content of organic polymers where underestimation of BOD values might occur. Of course, the use of this analyzer is not limited to this application example. It can also be used for on-line quality control of effluents in numbers of both municipal and industrial wastewater treatment plants that includes both aerobic and anaerobic operations. This will be particularly valuable to wastewater treatment plants since effluent standards are getting tighter. A rapid and frequent feedback of the effluent analysis will be obligated to tune the plant operation properly in order to meet tomorrow’s regulations.
Acknowledgements 4. Conclusions Monitoring of intermediate metabolites from anaerobic decomposition processes is important since it generates signals that can be used to monitor and control the process for stable start-up and steady-state running operations. Since it has been difficult to monitor such a process on-line, many biogas systems are operated far below their maximal reactor capacity due to a desire to eliminate the risk of system overload. If that happens, the methanogens may get killed and one then has to reinoculate the reactor and start from the very beginning. The strategy to overcome the problem is to combine a suitable reactor design with a closely monitored and controlled process in order to exploit the cost benefit of a smaller anaerobic system and allow waste treatment at a higher specific rate. Among the commonly suggested monitoring parameters, one that indicates the intermediate products (e.g. easily assimilable organic compounds) and consequently reflects a loading rate of biodegradable organic material should preferably be included. This paper reports an example using the stand-alone BOD analyzer for on-line monitoring of the intermediate products in an anaerobic digestion process. The experimental results demonstrate successful utilization of this sensor device and the BOD biosensor method can be a useful and valuable on-line method for fast indication of easily biodegradable organic substances from the initial step of a two-stage anaerobic process. The on-line BODst estimation can then be used for keeping the OLR under control in order to avoid acidification of the methanogenic step. However, attention should be
The Swedish Agency for Research Co-operation with Developing Countries (SAREC), the Swedish National Energy Administration (STEM), and the Energy Supply Committee of Southern Sweden (DESS) are gratefully acknowledged for the financial support.
References Ahring, B., Sandberg, M., Angelidaki, I., 1995. Volatile fatty acids as indicators of process imbalance in anaerobic digestors. Appl. Microbiol. Biotechnol. 43 (3), 559–565. APHA, 1992. Standard Methods for the Examination of Water and Wastewater, Section 5210, 18th ed. American Public Health Association Water Works Association, American Water Environment Federation, Washington, DC, pp. 5.1–5.6. Armenante, P.M., 1993. Bioreactors. In: Levin, M.A., Gealt, M.A. (Eds.), Biotreatment of Industrial and Hazardous Waste. McGraw-Hill, New York, NY, pp. 65–112. Björnsson, L., 2000. Intensification of the biogas process by improved process monitoring and biomass retention. Doctoral Dissertation, Department of Biotechnology, Lund University, Sweden. Björnsson, L., Hörnsten, E.G., Mattiasson, B., 2001. Utilization of a palladium-metal oxide semiconductor (Pd-MOS) sensor for on-line monitoring of dissolved hydrogen in anaerobic digestion. Biotechnol. Bioeng. 73 (1), 35–43. Cobb, S.A., Hill, D.T., 1991. Volatile fatty acid relationships in attached growth anaerobic fermenters. Trans. ASAE 34 (6), 2564–2572. Cord-Ruwisch, R., Mercz, T.I., Hoh, C.-Y., Strong, G.E., 1997. Dissolved hydrogen concentration as an on-line control parameter for the automated operation and optimization of anaerobic digesters. Biotechnol. Bioeng. 56 (6), 626–634. Ghosh, S., 1991. Pilot-scale demonstration of two-phase anaerobic digestion of activated sludge. Water Sci. Technol. 23 (7–9), 1179–1188.
J. Liu et al. / Journal of Biotechnology 109 (2004) 263–275 Hickey, R.F., Switzenbaum, M.S., 1991. Thermodynamics of volatile fatty acid accumulation in anaerobic digesters subject to increases in hydraulic and organic loading. Res. J. Water Pollut. Control Fed. 63 (2), 141–144. Kaseng, K., Ibrahim, K., Paneerselvam, S.V., Hassan, R.S., 1992. Extracellular enzymes and acidogen profiles of a laboratory-scale two-phase anaerobic digestion system. Process Biochem. 27 (1), 43–47. Liu, J., Björnsson, L., Mattiasson, B., 2000. Immobilised activated sludge based biosensor for biochemical oxygen demand measurement. Biosens. Bioelectron. 14 (12), 883–893. Liu, J., Mattiasson, B., 2002. Microbial BOD sensors for wastewater analysis. Water Res. 36 (15), 3786–3802. Liu, J., Olsson, G., Mattiasson, B., 2003. Monitoring of two-stage anaerobic biodegradation using a BOD biosensor. J. Biotechnol. 100 (3), 261–265. Liu, J., Olsson, G., Mattiasson, B., 2004a. Short-term BOD (BODst ) as a parameter for on-line monitoring of biological treatment process, Part I: A novel design of BOD biosensor for easy renewal of bio-layer. Biosens. Bioelectron. (in press). Liu, J., Olsson, G., Mattiasson, B., 2004b. Short-term BOD (BODst ) as a parameter for on-line monitoring of biological treatment process, Part II: Instrumentation of integrated flow
275
injection analysis (FIA) system for BODst estimation. Biosens. Bioelectron. (in press). Liu, J., Olsson, G., Mattiasson, B., 2004c. A volumetric meter for monitoring of low gas flow rate from laboratory-scale biogas reactor. Sens. Actuators B 97, 369–372. Moletta, R., Escoffier, Y., Ehlinger, F., Coudert, J.-P., Leyris, J.-P., 1994. On-line automatic control system for monitoring an anaerobic fluidized-bed reactor: response to organic overload. Water Sci. Technol. 30 (12), 11–20. Organisation for Economic Cooperation and Development (OECD), 1981. Activated sludge, respiration inhibition test. OECD Guidelines Testing Chem. 209, 1–10. SIS-Standardiseringskommissionen i Sverige, 1979. Water analysis—Determination of Biochemical Oxygen Demand, BOD, of Water—Dilution Method (Svensk Standard SS 02 81 43 E), first ed. The Swedish Standards Institution, Stockholm, pp. 1–9. Spanjers, H., Olsson, G., Klapwijk, A., 1993. Determining influent short-term biochemical oxygen demand by combined respirometry and estimation. Water Sci. Technol. 28 (11/12), 401–414. Speece, R.E., 1996. Operational considerations. In: Speece, R.E. (Ed.), Anaerobic Biotechnology for Industrial Wastewaters. Archae Press, pp. 69–112.