wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 9 7 3e5 9 8 0

Avai lab le a t www.sc iencedi rec t .com

journa l homepage : www.e lsev ie r . com/ loca te /wat res

State indicators for monitoring the anaerobic digestionprocess

Kanokwan Boe a, Damien John Batstone a,b, Jean-Phillippe Steyer a,c, Irini Angelidaki a,*aDepartment of Environmental Engineering, Technical University of Denmark, Building 113, DK-2800, Kgs. Lyngby, DenmarkbAdvanced Water Management Centre, The University of Queensland, St Lucia, QLD 4067, Australiac Laboratory of environmental biotechnology, French National Institute for Agronomic Research, Avenue des Etangs, 11100 Narbonne, France

a r t i c l e i n f o

Article history:

Received 24 February 2010

Received in revised form

6 June 2010

Accepted 14 July 2010

Available online 23 July 2010

Keywords:

Anaerobic digestion

Monitoring

Volatile fatty acids

Dissolved hydrogen

Biogas

* Corresponding author. Tel.: þ45 45251429;E-mail address: ria@env.dtu.dk (I. Angeli

0043-1354/$ e see front matter ª 2010 Elsevdoi:10.1016/j.watres.2010.07.043

a b s t r a c t

Anaerobic process state indicators were used to monitor a manure digester exposed to

different types of disturbances, in order to find the most proper indicator(s) for monitoring

the biogas process. Online indicators tested were biogas production, pH, volatile fatty acids

(VFA), and dissolved hydrogen. Offline indicators tested were methane and hydrogen

content in the biogas. A CSTR reactor with 7.2 L working volume was operated at a varying

hydraulic loading rate (HRT 10e20 days) for 200 days. During this period, the reactor was

overloaded with extra organic matter such as glucose, lipid, gelatine, and bio-fibers, in

order to create dynamic changes in the process state. Biogas production increased in

response to the increase in organic load with a slight decrease in methane content. pH was

relatively stable and did not show clear response to hydraulic load changes. However, pH

changes were observed in response to extra organic load. Individual VFA concentrations

were an effective indicator, with propionate persistent for the longest time after intro-

duction of the disturbance. Dissolved hydrogen was very sensitive to the addition of easily

degradable organics. However, it responded also to other disturbances such as slight air

exposure which had no impact on process performance. A combination of acetate,

propionate and biogas production is an effective combination to monitor this type of

digesters effectively.

ª 2010 Elsevier Ltd. All rights reserved.

1. Introduction of the biogas process are gas production, biogas composition,

Monitoring and control are important strategies for achieving

a better process stability and higher conversion efficiencies in

anaerobic digesters. Monitoring is a requirement for process

control. The lack of suitable process indicators results in the

limited control and optimization of anaerobic digestion. An

ideal indicator should reflect the current process status and

be straightforward to measure. Moreover, its response to the

process imbalances should be significant compared to back-

groundfluctuations.Thecommonindicators for themonitoring

fax: þ45 45932850.daki).ier Ltd. All rights reserved

pH, alkalinity and volatile fatty acids (VFA) (Hawkes et al., 1993).

Biogas production is the most commonly monitored indi-

cator, since it indicates the overall process performance and

can be measured by a number of robust online sensors.

However, it can poorly indicate an imbalanced state and often

decreaseswhen the process is already damaged (Moletta et al.,

1994). The low biogas production results not only fromprocess

inhibition but also from low reactor loading. pH is relatively

straightforward to measure and is often the only online liquid

stated measured parameter. A pH decrease can indicate an

accumulation of VFA. In a reactor with low buffering capacity,

.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 9 7 3e5 9 8 05974

pHcanbeauseful indicator.However, thepHresponsehas low

sensitivity in a well-buffered system (Bjornsson et al., 2000).

Biogas composition is a traditional parameter where low

methane percent (i.e. high carbon dioxide content) could

indicate inadequate process performance. However, the

carbondioxide content is dependent on pH and, consequently,

fluctuation of pH can affect the gas composition without

decreasing methane production (Hansson et al., 2002).

Hydrogen content of biogas is a very sensitive indicator and is

connected to the imbalance between microbial groups in the

digestion process (Molina et al., 2009; Steyer et al., 2002). The

hydrogen content in biogas can easily be measured online

using a semiconductor sensor (Hornsten et al., 1991). However,

dissolved hydrogen may be more appropriate than gaseous

hydrogen, as it is not delayed by liquidegas transfer and could

better correlate to the VFA concentration (Pauss and Guiot,

1993). Dissolved hydrogen increases together with VFA accu-

mulation during sudden increase of organic load (Bjornsson

et al., 2001b). VFA is widely suggested as process indicator,

since it is the main pre-methanogenic intermediate (Jacobi

et al., 2009; Molina et al., 2009). VFA accumulation in anaer-

obic reactors indicates process imbalance (Ahring et al., 1995).

Moreover, individual VFA concentrations give specific infor-

mation for process diagnosis (Ahring et al., 1995; Cobb andHill,

1991). Total VFA concentration can be measured online by

titration (Feitkenhauer et al., 2002), or indirectly where light

spectroscopy is correlated to total VFA concentrations, by

using near infrared spectroscopy (NIR) (Holm-Nielsen et al.,

2008; Jacobi et al., 2009). However, to measure individual VFA,

online monitoring is more complex. The online monitoring of

individual VFAhas beenbased on sample filtration followed by

analysis in a gas chromatograph (Pind et al., 2003), or using

headspace extraction followed by analysis in a gas chromato-

graph (Boe et al., 2007).

Many of the studies cited above assessed only a limited

number of indicators, and often in processes operating under

unstressed state. Moreover, lack of an online sensor for indi-

vidual VFA limits the evaluation of this important indicator.

The aim of this study is to assess the suitability of different

anaerobic process indicators. A range of indicators, including

biogas production, pH, individual VFA, dissolved hydrogen,

and gas phasemethane and hydrogen contentwere compared

under different types of disturbances.

Table 1 e Summary of extra organic load added to the reactor

Day Amount added (g/day)

Lipid Glucose Gelatine

77 85 e e

112 157 e e

126 e 25 e

137 e 50 e

142 e e 25

161 e 50 e

168 e 100 e

185e187 4 e e

188e196 4 40 e

2. Material and methods

The experiment was carried out in a 9-L CSTR reactor with

a 7.2 L working volume. Cattle manure (3%TS, 2%VS) was used

as substrate for the reactor. The reactor was operated at 55 �Cat a varying hydraulic loading (10e20 days HRT) for 200 days

and was fed four times per day using a peristaltic pump

(Watson Marlow) controlled by a timer and relay.

To compare the indicators’ responses, hydraulic and

organic load disturbances were introduced. For hydraulic

disturbances, the feed volume was increased by increasing

the feed duration. For organic overload, different organic

compounds, besides the daily manure feed, were added into

the reactor as summarised in Table 1. Rapeseed oil and gela-

tine were used to represent lipid and protein, respectively.

Glucose was used to represent easily degradable carbohydrate

while bio-fiber containing arabinoxylans (Ispaghula Husk,

Vi-Siblin; Edwards et al., 2003) was used to represent slowly

degradable carbohydrate.

During operation, the responses of different process indi-

cators were measured. Online indicators were biogas

production, pH, volatile fatty acids (VFA), and dissolved

hydrogen. Offline indicators were percent methane and

hydrogen in the biogas. Biogas production was measured by

an automated displacement gas metering system with

a 100 mL reversible cycle and registration (Angelidaki et al.,

1992). The water used in gas meter was acidified to pH 3 by

HCl added NaCl to prevent CO2 dissolution. Gas production

data was recorded automatically every 6 h. pH was measured

online by a mini CHEM-pH Process Monitor (TPS Pty Ltd.,

Australia). The meter was calibrated against pH 4.00 and pH

6.88 buffers every second week. The pH was recorded auto-

matically every 10 min. Individual VFA concentrations were

measured by an online VFA monitoring system based on ex-

situ VFA extraction (Boe et al., 2007). The reactor had a liquid

circulation loop from which a 40 mL liquid sample was

pumped into an extraction chamber, acidified, added with

salt, and was heated in order to extract the VFA into gas phase

before injecting into a gas chromatograph (GC) for analysis.

The signal output from the GC was then sent to data pro-

cessing system for integration. The VFA concentrations were

analysed and recorded automatically every 6 h.

during experiment.

Method of addition

Bio-fiber

e Added once, directly into the reactor

e Mixed with feed and fed 4 times

e Added once, directly into the reactor

e Added once, directly into the reactor

e Added once, directly into the reactor

e Added once, directly into the reactor

e Added once, directly into the reactor

8 Mixed with feed and fed 4 times

8 Mixed with feed and fed 4 times

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 9 7 3e5 9 8 0 5975

During day 122e128, the dissolved hydrogen was measured

by an online hydrogen micro-sensor (Unisense A/S, Aarhus,

Denmark). The sensor principle is based on hydrogen diffusion

from the liquid through a sensor tip silicone membrane, to the

platinum anode which is polarized against an internal refer-

ence. The flow of electrons from the oxidizing anode to the

internal referencereflects linearly thehydrogenpartialpressure

around the sensor tip and is in the pico-amp range. A picoam-

meter converted the resulting oxidation current to a signal. The

signal output from the picoammeter was then recorded in the

data processing system (Unisense A/S, Aarhus, Denmark).

During day 157e200, the dissolved hydrogen was measured

by the online hydrogen measuring system developed by

Bjornsson et al. (2001a). The system applied a liquid-to-gas

membraneextractionforextractingdissolvedhydrogenfromthe

liquid content. Thedissolvedhydrogendiffused throughaTeflon

membrane immersed in the reactor. The diffused hydrogenwas

then oxidized at the surface of a Palladium-Metal Oxide semi-

conductor (Pd-MOS) sensor. The picoammeter converted the

resultingoxidationcurrent toasignal.Thesignaloutput fromthe

picoammeter was then recorded in the data processing system.

The results from hydrogen sensors were presented here as

relative numbers of signal outputs compared to the initial signal,

since it was found unreliable to calibrate absolute concentration

of dissolved hydrogen inmanure against water.

Gas phase methane and carbon dioxide were measured

offlinebyagaschromatograph(Mikrolab, Arhus)equippedwith

thermal conductivity detector and a glass column 20m� 3mm

ID packed with Poropack Q (10/80). The temperature of the

injector, the detector and the oven was isothermal at 55 �C.Heliumwas used as a carrier gas with the flow rate 40mL/min.

Gas phase hydrogen was measured by a gas chromatograph

(Mikrolab, Arhus) equippedwith thermal conductivity detector

and a packed column 4.5 m � 3 mm ID Molsieve 5A 10/80. The

injector and detector temperature was 90 �C. The temperature

programwas isothermal at 80 �C.Nitrogenwasusedas a carrier

gas with the flow rate 20 mL/min.

Online data processing was done by a programmable logic

control (PLC) system (Versamax PLC, GE Fanuc Automation

Europe S.A, Luxembourg), with a PC interface. All calculations,

including peak area calculation of the GC were managed

within the PLC. The interface and data logging on the PLCwere

using GE Cimplicity HMI 6.1 (HMI, GE Fanuc Automation

Europe S.A, Luxembourg).

3. Results

All the measured indicators showed response to the changes

in hydraulic and organic load. During the start-up period (day

0e20), very high VFA concentrations, up to 70 mM, were

observed. Biogas production and VFA levels increased while

pH changed by 0.5e1 unit. Acetate and butyrate were themost

dominant VFA. After day 20, acetate and butyrate decreased

relatively quickly while propionate was the most persistent.

3.1. Response to lipid addition

Two lipid additions were introduced by adding 85 g and 157 g

of rapeseed oil directly into the reactor at day 77 and day 110,

respectively (Fig. 1). No increase in biogas production and only

a small increase of VFA were observed after the first addition.

While after the second one a drop of both biogas production

and methane percent, but no clear response in both pH and

VFA were observed. After the second addition, most of the oil

came out undigested with the effluent from the top of reactor

and biogas production returned slowly to normal levels.

3.2. Response to glucose addition

Four glucose additions were introduced by adding 25, 50, 50

and 100 g of glucose directly into the reactor at day 126, 137,

161 and 168, respectively. The results from the first two

additions are shown in Fig. 2, and the results from the last two

additions are shown in Fig. 3.

At approximately 1 day after the 25 g glucose was added,

biogas production increased shortly (Fig. 2a), pH droppedwhile

dissolved hydrogen increased (Fig. 2b), and VFA concentration

increased slightly while methane percent did not show signifi-

cant response (Fig. 2c and d). Hydrogen content in biogas

increased slightly during the same period that dissolved

hydrogen increased. However, the values were very low and

fluctuated. At day 123, dissolved hydrogen showed some

response fewminutesafter the reactorwasopened to repair the

effluent tube.Additionof 50g glucoseat day 137 showedsimilar

response to theaddition of 25 g glucose, however,with stronger

response of VFA, where butyrate, iso-valerate and valerate

increased slightly at both day 137 and 161. At day 161, the dis-

solved hydrogen increased sharply, along with a pH drop,

a slight increase in hydrogen content and a slight decrease in

biogas methane content (Fig. 3b and d).

After the addition of 100 g glucose, biogas production

increased while methane content decreased (Fig. 3a and d).

Acetate and butyrate increased significantly and followed by

an increase of iso-butyrate, iso-valerate, valerate and propio-

nate concentrations (Fig. 3c). pH values dropped and dissolved

hydrogen increased sharply (Fig. 3b). Dissolved hydrogen

dropped back very quickly, while pH slowly increased over

severalhours.VFAtook longer time todecreaseback tonormal.

Also, the gas phase hydrogen content increased slightly.

3.3. Response to protein addition

Proteinwasaddedatday 142byadding25g gelatinedirectly into

the reactor (Fig. 2). Only biogas production and acetate concen-

tration slightly increased while the rest of VFA and pH did not

show significant response. There was no data of dissolved

hydrogen and biogas composition available during this period.

3.4. Response to continuous addition of extraorganic load

The reactor was daily fed with extra organic load, mixed with

the manure in the influent flask, during day 185e196. The

results are shown in Fig. 3. From day 185, the feed (manure)

was supplemented with 8 g bio-fiber and 4 g rapeseed oil per

day. Acetate started to increase and pH began to drop while

small response of dissolved hydrogen was noticed. From day

188, the feedwas also supplementedwith 40 g glucose per day.

At this point, the rest of VFA started to increase. A strong

Fig. 1 e Reactor results during day 1e120.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 9 7 3e5 9 8 05976

response of dissolved hydrogen was observed, as well as a pH

decrease. Biogas production increased significantly while the

methane content decreased slightly. As the process was

continued to be fed with extra organic load, the dissolved

hydrogen and acetate concentration were the first to decrease

followed by butyrate and iso-butyrate. The rest of VFA started

to decrease shortly after extra organic load was ceased except

for propionate which continued increasing. The pH increased

slightly after removal of extra organic load, but did not reach

the previous level. Dissolved hydrogen decreased back to the

normal level.

4. Discussion

4.1. Reactor response to extra organic load

The reactor responded quickly to the addition of glucose

because glucose is easily degradable. Dissolved hydrogen, VFA

and biogas production levels were all good indicators of

response to this stimulus. After the first lipid addition, no

increase in biogas production and only small increase in VFA

were observed, which could be explained by the slow degra-

dation of lipid. A similar observation was previously reported

by Bjornsson et al. (2001b), where the lipid addition gave low

production of VFA and they suggested that this was due to

hydrolysis being the rate-limiting step for lipid digestion. After

the second addition with doubling the amount of lipid, the

drop in biogas production with small increase of VFA

suggested that the process was probably inhibited by long-

chain fatty acids (LCFA) from the oil. LCFA is known to be

inhibitory to all groups of microorganisms (Angelidaki and

Ahring, 1992; Rinzema et al., 1994). In this case, acidogens

were also inhibited and concentration of VFA alone was not

a suitable indicator for this disturbance. Moreover, the fact

that the reactor slowly recovered after the undigested oil

washed out with the effluent suggested that the reactor was

recovered due to dilution of LCFA by the new feed, rather than

adaptation of microorganisms, agreeing with observations by

Pereira et al. (2003) and Rinzema et al. (1994). The addition of

protein at day 142 did not disturb the process as seen that only

acetate increased slightly and the biogas increased.

4.2. Analysis of indicators

The criteria used for comparing the process indicators in this

paper were their responses during different disturbances, in

relating to their baselines under normal operation. From the

experiment, it was noticed that acetate exhibited faster

dynamics and fluctuated more than propionate. Acetate

increased very fast after the increase of hydraulic or organic

load. However, it decreased again few days later while organic

load was still high. This could be due to fast growth rates of

aceticlastic methanogens compared to propionate degraders,

or a larger population of aceticlasts. Propionate degraders are

having growth rates around 0.49 day�1, while aceticlastic

methanogens around 0.6 day�1 (Angelidaki et al., 1999). Buty-

rate responded also very quickly to an overload. However,

Fig. 2 e Reactor results during day 120e147.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 9 7 3e5 9 8 0 5977

butyrate accumulation seemed to depend on the substrate

composition and not all disturbances could increase butyrate.

Butyrate accumulatedmostly during the period of adding high

concentrationofglucose, forexample,by100gofglucoseatday

168. However, only a slight increase in butyrate was observed

during additions of 50 g glucose at day 137 and 161, and no

butyrate accumulation was observed during the addition of

25 g glucose at day 125. Valerate showed similar responses

as butyrate, however, with much lower concentration level.

Iso-butyrate and iso-valerate showed similar responses as

butyrate and valerate, respectively. However, the iso-formwas

more persistent in the reactor than the normal-form. During

continuousoverload,propionate remained in thereactormuch

longer than other VFA. Moreover, while acetate and butyrate

started to decrease after the reactor had been exposed to

organicoverload for sometime,propionatekeptaccumulating.

This could be due to the fact that propionate degradation is the

most thermodynamic unfavourable among other VFA degra-

dation,whichmade propionate degraders the slowest growing

and most sensitive compared to acetate and butyrate

degraders which could faster increase their degradation rate

(Ozturk, 1991). In this case, propionate would be a better

parameter to indicate process stress. This observation was

similar to the study of Nielsen et al. (2007) where they sug-

gested that propionate was the best indicator to describe the

normalizing of the process.

Dissolved hydrogen had strong response specially when

adding glucose to the feed. This is consistent with hydrogen

beingamajorproduct fromglucosedegradation (Batstoneetal.,

2002). However, it responded also to other disturbances such as

slight air exposure which had no impact on process perfor-

mance. This ismore likely a semiconductor response to change

in redox. In principle, the increase of both VFA and dissolved

hydrogen in the reactor couldbe linkeddirectly to the increased

Fig. 3 e Reactor results during day 155e200.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 9 7 3e5 9 8 05978

activities of acidogens. Thedecrease of acetate could reflect the

increased activity of aceticlastic methanogens. Similarly, the

rapid decrease inhydrogen could be due to increased activity of

hydrogenotrophic methanogens. Fig. 4 shows the dynamics of

dissolved hydrogen in the reactor during continuous overload.

The fast response of hydrogen was clearly observed as oscilla-

tions following the feed interval of four times per day. It was

noticed that the peak of dissolved hydrogen was reached

around 30min after each feed, corresponding to small pH drop.

Moreover, thedissolvedhydrogendecreasedagainwithin a few

hours, while pHwas still increasing.

The decrease of pH corresponded to the VFA accumulation

during overload. However, the level of pH change was not

significant enough to indicate the state of the process in this

case due to the high buffer capacity in manure digester. The

response of pH also corresponded to the dissolved hydrogen

during sudden overload but not during gradual overload. This

could be explained by the pH response as being the result of

overall ion interactions in the solution, while dissolved

hydrogen measurements are not compensated by these

interactions. Thus, during sudden overload of particular

substrate such as glucose, where hydrogen production was

high, the dissolved hydrogen response could be correlated to

the pH. Moreover, it was noticed that after removal of

continuous organic addition, the pH was not back to the

previous level due to VFA accumulation. Dissolved hydrogen

dropped rapidly to normal level.

Biogas production responded very quickly to the change in

organic load. However, it could not indicate the imbalanced

state of the reactor. During continuous overload the biogas

production was increasing along with the VFA concentration.

The overload was then removed due to high VFA concentra-

tion. If only the biogas production was used as indicator of the

process status, the imbalance would not had been discovered

Fig. 4 e Dynamics of dissolved hydrogen and pH during organic overload.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 9 7 3e5 9 8 0 5979

and one would had continued with the same loading which

may have led to process failure.

4.3. Suitability of indicators

Several indicators showed interesting responses to the

increase of organic load. The parameters that had the fastest

response were dissolved hydrogen, pH and acetate, followed

by butyrate (in case of glucose), propionate (in case of high

overload), and biogas, respectively. The response time of

biogas composition was quite slow probably due to large

headspace volume of the reactor. However, none of these

indicators showed response to all perturbations. Thus, the

combination of different indicators might be necessary to

cover all imbalance situations. As propionate was the most

persistent in the reactor, it could be an important indicator to

determine the degree of process imbalance. On the other

hand, although biogas production could not indicate process

imbalance, it is main product of interest reflecting overall

process performance. Thus, the combination of acetate/

propionate and biogas production is an effective group of

indicators of both performance, and process balance. This is

in contrast to traditional indicators, which aremainly pH, and

either acetate only, or a combination of all organic acids.

However, it should be remarked that the indicators suggested

here were tested in a manure digester which had very high

buffering capacity. For the processwith low buffering capacity

such as sludge digester or high-rate anaerobic digester, the pH

could still be a useful indicator.

Other important factors to be considered when choosing

the state indicator for the full-scale application are reliability

and robustness of the online meters. Gas production and pH

are easy to measure and most of the anaerobic wastewater

treatment plants have gas and pH meters as standard

instruments (Spanjers and van Lier, 2006). However, the liquid

phase parameters such as individual VFA are still measured

through manual analysis. The individual VFA online moni-

toring used in this experiment (Boe et al., 2007) is under

further development of the industrial prototype to improve

the robustness for operation in full-scale plants.

5. Conclusions

Dissolved hydrogen was sensitive to organic overload, espe-

cially when glucose was present in the feed. However, it also

responded to oxygen exposure which did not show any effect

on the process performance. The pH responded as fast as

dissolved hydrogen but with very small change whichmade it

difficult to indicate the status of the process based on pH

value. Acetate and propionate were very sensitive to organic

overload. Propionate was the most persistent parameter

which was effective indicator of stress status of the process

while acetate decreased faster and was more fluctuated. The

sensitivity of gas phase composition in this study was quite

low, probably due to large headspace volume of the reactor,

and slow gaseliquid dynamics. Biogas productionwas still the

important parameter for indicating overall reactor perfor-

mance although it could not indicate the stress status of the

reactor. A monitoring of both individual VFA such as acetate,

propionate and biogas productionwould allow combination of

process state and performance.

Acknowledgements

This work was supported by the Ph.D. scholarship from the

Institute of Environment and Resources, Technical University

of Denmark. The work of J.P. Steyer was supported by the

European Community’s Human Potential Programme under

contract MEIF-CT-2005-009500 (CONTROL-AD4H2) and both

are greatly acknowledged.

r e f e r e n c e s

Ahring, B.K., Sandberg, M., Angelidaki, I., 1995. Volatile fatty acidsas indicators of process imbalance in anaerobic digestors.Applied Microbiology and Biotechnology 43, 559e565.

Angelidaki, I., Ahring, B.K., 1992. Effects of free long-chain fattyacids on thermophilic anaerobic digestion. AppliedMicrobiology and Biotechnology 37, 808e812.

Angelidaki, I., Ellegaard, L., Ahring, B.K., 1992. Compactautomated displacement gas metering system formeasurement of low gas rates from laboratory fermentors.Biotechnology and Bioengineering 39, 351e353.

Angelidaki, I., Ellegaard, L., Ahring, B.K., 1999. A comprehensivemodel of anaerobic bioconversion of complex substrates tobiogas. Biotechnology and Bioengineering 63, 363e372.

Batstone, D.J., Keller, J., Angelidaki, I., Kalyuzhny, S.V.,Pavlostathis, S.G., Rozzi, A., Sanders, W.T.M., Siegrist, H.,Vavilin, V.A., 2002. The IWA anaerobic digestion model no 1.(ADM1). Water Science and Technology 45, 65e73.

wat e r r e s e a r c h 4 4 ( 2 0 1 0 ) 5 9 7 3e5 9 8 05980

Bjornsson, L., Murto, M., Mattiasson, B., 2000. Evaluation ofparameters for monitoring an anaerobic co-digestion process.Applied Microbiology and Biotechnology 54, 844e849.

Bjornsson, L., Hornsten, E.G., Mattiasson, B., 2001a. Utilisation ofa Pd-MOS sensor for on-linemonitoring of dissolved hydrogen inanaerobicdigestion.BiotechnologyandBioengineering73,35e43.

Bjornsson, L., Murto, M., Jantsch, T.G., Mattiasson, B., 2001b.Evaluation of new methods for the monitoring of alkalinity,dissolved hydrogen and the microbial community inanaerobic digestion. Water Research 35 (12), 2833e2840.

Boe, K., Batstone, D.J., Angelidaki, I., 2007. An innovative onlineVFA monitoring system for the anaerobic process, based onheadspace gas chromatography. Biotechnology andBioengineering 96 (4), 712e721.

Cobb, S.A., Hill, D.T., 1991. Volatile fatty acid relationships inattached growth anaerobic fermenters. Transactions of theASAE 34 (6), 2564e2572.

Edwards, S., Chaplin, M.F., Blackwood, A.D., Dettmar, P.W., 2003.Primary structure of arabinoxylans of Ispaghula husk andwheat bran. Proceedings of the Nutrition Society 62, 217e222.

Feitkenhauer, H., Sachs, J.V., Meyer, U., 2002. On-line titration ofvolatile fatty acids for the process control of anaerobicdigestion plants. Water Research 36, 212e218.

Hansson, M., Nordberg, A., Sundh, I., Mathisen, B., 2002. Earlywarning of disturbances in a laboratory-scale MSW biogasprocess. Water Science and Technology 45 (10), 255e260.

Hawkes, F.R., Guwy, A.J., Rozzi, A.G., Hawkes, D.L., 1993. Newinstrument for on-line measurement of bicarbonate alkalinity.Water Research 27 (1), 167e170.

Holm-Nielsen, J.B., Lomborg, C.J., Oleskowicz-Popiel, P.,Esbensen, K.H., 2008. On-line near infrared monitoring ofglycerol-boosted anaerobic digestion processes: evaluation ofprocess analytical technologies. Biotechnology andBioengineering 99 (2), 302e313.

Hornsten, E.G., Lundstrom, I., Nordberg, A., Mathisen, B., 1991.The use of palladium metal oxide semiconductor structures inquantitative studies of H2 and H2S in processes related tobiogas production. Bioprocess Engineering 6, 235e240.

Jacobi, H.F., Moschner, C.R., Hartung, E., 2009. Use of nearinfrared spectroscopy in monitoring of volatile fatty acids in

anaerobic digestion. Water Science and Technology 60 (2),339e346.

Moletta, R., Escoffier, Y., Ehlinger, F., Coudert, J.P., Leyris, J.P.,1994. On-line automatic control system for monitoring ananaerobic fluidized-bed reactor: response to organic overload.Water Science and Technology 30 (12), 11e20.

Molina, F., Garcıa, C., Roca, E., Lema, J.M., Molina, F.,Castellano, M., 2009. Selection of variables for on-linemonitoring, diagnosis, and control of anaerobicdigestion processes. Water Science and Technology 60(3), 615e622.

Nielsen, H.B., Uellendahla, H., Ahring, B.K., 2007. Regulation andoptimization of the biogas process: propionate as a keyparameter. Biomass and Bioenergy 31 (11e12), 820e830.

Ozturk, M., 1991. Conversion of acetate, propionate and butyrateto methane under thermophilic conditions in batch reactors.Water Research 12, 1509e1513.

Pauss, A., Guiot, S.R., 1993. Hydrogen monitoring in anaerobicsludge bed reactors at various hydraulic regimes andloading rates. Water Environment Research 65 (3),276e280.

Pereira, M.A., Cavaleiro, A.J., Mota, M., Alves, M.M., 2003.Accumulation of long chain fatty acids onto anaerobic sludgeunder steady state and shock loading conditions: effect onacetogenic and methanogenic activity. Water Science andTechnology 48 (6), 33e40.

Pind, P.F., Angelidaki, I., Ahring, B.K., 2003. A new VFA sensortechnique for anaerobic reactor systems. Biotechnology andBioengineering 82 (1), 54e61.

Rinzema, A., Boone, M., van Knippenberg, K., Lettinga, G.,1994. Bactericidal effect of long chain fatty acids inanaerobic digestion. Water Environment Research 66 (1),40e49.

Spanjers, H., van Lier, J.B., 2006. Instrumentation in anaerobictreatment e research and practice. Water Science andTechnology 53 (4e5), 63e76.

Steyer, J.-P., Bouvier, J.C., Conte, T., Gras, P., Sousbie, P., 2002.Evaluation of a four year experience with a fully instrumentedanaerobic digestion process. Water Science and Technology 45(4e5), 495e502.