The aim of this work was to analyse the applicability of electrical conductivity sensors for on-line monitoring the start-up period of an anaerobic fixed-bed reactor. The evolution of bicarbonate concentration and methane production rate was analysed. Strong linear relationships between electrical conductivity and both bicarbonate concentration and methane production rate were observed. On-line estimations of the studied parameters were carried out in a new start-up period by applying simple linear regression models, which resulted in a good concordance between both observed and predicted values. Electrical conductivity sensors were therefore identified as an interesting method for monitoring the start-up period of anaerobic fixed-bed reactors due to its reliability, robustness, easy operation, low cost, and minimum maintenance compared with the currently used sensors.

INTRODUCTION

Methane production represents the main challenge for anaerobic digestion (AD) of waste. Nowadays, high effort has been focused on the study of high-rate systems that use self-immobilised biomass (e.g. up-flow anaerobic sludge blanket, and expanded granular sludge blanket or fixed-bed bioreactors), where hydraulic retentions time (HRT) and solid retention time are uncoupled. However, anaerobic fixed-film reactors usually require of a long start-up time, which is critical to reach successful process performance (Escudié et al. 2011).

The start-up stage of a fixed-film process could be defined as ‘the time necessary for selection and proper spatial arrangement of the fittest bacterial strains of a consortium best tuned with each other on the substratum’ (Heppner et al. 1992). Optimising the start-up time is a key challenge in order to maximise the economic benefits of an anaerobic process (Weiland & Rozzi 1991). Hence, due to the high complexity and instability of AD processes, it is necessary to develop suitable monitoring systems enabling successfully starting-up, stabilising and optimising anaerobic fixed-film reactors (von Sachs et al. 2003; Escudié et al. 2011).

Biogas composition and flow rate are commonly used as indicators for monitoring the performance of AD processes. Moreover, the methane yield (YCH4), which is defined as the amount of methane produced per unit of removed organic matter, can also be used as an indirect parameter for evaluating the start-up period of an AD process (Michaud et al. 2002). Nevertheless, these indicators can be insufficient to evaluate the overall process performance since they could indicate a fault of the process when quick recovery is not possible anymore. On the other hand, the reliability and efficiency of the volatile fatty acid (VFA) concentration as state indicator for monitoring the performance of AD processes has been demonstrated (Boe et al. 2010). VFA concentration can be reliably monitored, for instance, by means of a titrimetric sensor or a Fourier transform infrared spectrometer (Steyer et al. 2002a, 2002b).However, the ideal on-line technique for bioprocess monitoring in full-scale applications should be robust, simple to maintain and to perform, and with a long lifespan and minimal investment cost (Aceves-Lara et al. 2012).

Electrical conductivity is defined as the ability of a solution to conduct electrical current and it is directly proportional to ion concentration. Moreover, it can be easily monitored: a cell formed by two electrodes is placed in the sample and the current between both electrodes is measured by means of the application of alternative potentials (Colombié et al. 2007).Therefore, conductivity measurements could be of great interest for monitoring and control AD processes, where ion concentration is mainly affected by both VFA and bicarbonate concentrations (Hawkes et al. 1994): two of the most reliable indicators of AD process performance (Olsson et al. 2005).

Several studies have been published where the feasibility of electrical conductivity sensors for bioprocess monitoring has been evaluated (see, for instance, Hoffmann et al. 2000; Varley et al. 2004; Aguado et al. 2006; Ellison et al. 2007). However, there is still a lack of knowledge regarding its applicability to AD processes. The main objective of this work was to analyse the applicability of electrical conductivity sensors for on-line monitoring the start-up period of an anaerobic fixed-bed reactor which was fed with winery wastewater. Bicarbonate concentration and methane production were analysed, and strong linear relationships between these parameters and electrical conductivity were observed. Theoretical calculations of both bicarbonate concentration and methane production from the on-line electrical conductivity measurements were conducted in a new start-up period, assessing the feasibility of electrical conductivity sensors for monitoring the start-up period of the evaluated AD process.

MATERIAL AND METHODS

Pilot plant description and operation

Figure 1 shows the flow diagram and the instrumentation of the continuous anaerobic fixed-bed reactor used in this study. This plant has a total volume of 0.358 m3. The support media (Cloisonyl: 180 m2·m−3 specific surface) fills 0.034 m3, leaving the resting 0.324 m3 of effective volume. In order to control the temperature when necessary, the anaerobic reactor is jacketed and connected to a water heating system. The reactor includes a pH-controller that feeds NaOH (30%) when necessary.
Figure 1

Flow diagram of the plant, including instrumentation (Nomenclature: FIT, flow-indicator-transmitter; PIT, pressure-indicator-transmitter; pH, pH-transmitter; CT, conductivity-transmitter; T, temperature sensor; PLC, programmable logic controller).

Figure 1

Flow diagram of the plant, including instrumentation (Nomenclature: FIT, flow-indicator-transmitter; PIT, pressure-indicator-transmitter; pH, pH-transmitter; CT, conductivity-transmitter; T, temperature sensor; PLC, programmable logic controller).

The plant was fed with raw winery wastewater from local cellars located in the area of Narbonne, France. Table 1 shows the main characteristics of the raw winery wastewater used in this study. The wastewater was stored in a feed tank (27 m3) that was connected to a dilution system of 0.2 m3. Hence, different organic loading rates (OLR) were tested. The HRT was maintained at 2.7 days, approximately. In the reactor, a portion of the liquid was recycled from the bottom to the top for both improving the mixing conditions and favouring the stripping of the produced gases from the liquid phase. The recycling flow rate during the experimental period was set to 550 L·h−1. The fresh substrate was mixed with the recycled liquid, and then introduced at the top of the reactor. The plant was operated at a controlled temperature of 35 °C. The pH set point was set to 7.2.

Table 1

Average characteristics of the raw winery wastewater

Parameter Unit Mean ± SD 
COD g COD·L−1 21.5 ± 0.8 
Acetate g COD·L−1 3.7 ± 0.4 
Propionate g COD·L−1 4.6 ± 0.8 
Butyrate g COD·L−1 2.8 ± 0.3 
Valerate g COD·L−1 1.5 ± 0.7 
Parameter Unit Mean ± SD 
COD g COD·L−1 21.5 ± 0.8 
Acetate g COD·L−1 3.7 ± 0.4 
Propionate g COD·L−1 4.6 ± 0.8 
Butyrate g COD·L−1 2.8 ± 0.3 
Valerate g COD·L−1 1.5 ± 0.7 

On-line monitoring

The on-line equipment used in this work consists of: a conductivity-temperature transmitter located in the recycling pipe; a flow-rate indicator transmitters for the winery wastewater fed-pump; an on-line titrimetric sensor (Anaerobic Control Analyser AnaSense®, AppliTek S.L.) for the measurements of total VFA and bicarbonate; and a gas flow-rate indicator transmitter (electromagnetic floater-based sensor KROHNE DK37E) and an on-line CH4 sensor (Bluesense BCP-CH4bio), both located in the biogas discharging pipeline. All gas measurements were normalised to standard conditions of temperature and pressure (STP).

Sampling and off-line measurements

Besides the on-line monitoring, biogas composition (CH4, CO2, O2, H2S and N2) was determined off-line using a gas chromatograph equipped with a thermic conductivity detector (GC-TCD, Perkin Elmer®, Clarus 480 GC). VFA composition (i.e. acetate (C2), propionate (C3), iso-butyrate and butyrate (iC4 and C4), and iso-valerate and valerate (iC5 and C5)) was determined by gas chromatography (Perkin Elmer®, Clarus 580 GC). Chemical oxygen demand (COD) was determined by the spectrophotometric micro-method (Tube Test MR, AQUALYTIC®), according to Standard Methods (2005).

RESULTS AND DISCUSSION

Process performance evaluation

Figure 2(a) shows the evolution of OLR during the experimental period. As this figure shows, OLR was progressively increased from 0 to 4 g COD L−1 d−1 in order to minimise possible disturbances on the biofilm formation. Moreover, the effluent VFA concentration was used as a state indicator of the process performance in order to avoid biomass-growth inhibitions due to possible overloads of the reactor. Thus, when effluent acetate concentrations higher than 1,600 mgCOD·L−1 were reached, a reduction in the OLR was conducted in order to avoid inhibition of propionic-reducing bacteria (see period from days 20 to 25).
Figure 2

Evolution throughout the operational period of (a) OLR and (b) reactor head-space pressure and methane production rate.

Figure 2

Evolution throughout the operational period of (a) OLR and (b) reactor head-space pressure and methane production rate.

Figure 2(b) shows the evolution of the head-space pressure in the anaerobic reactor and the methane production rate throughout the experimental period. It is important to note that outflow of biogas from the reactor was not detected until the head-space pressure reached the minimum value for biogas discharge. As Figure 2(b) shows, although the head-space pressure started increasing around day 5 (mainly due to carbon dioxide formation), it was observed that no methane production was monitored until day 20. Important to note is that the methane composition in the biogas started being noticeable around day 5, and reached a steady-state value around day 20 (up to 90% v/v). This increase in the biogas methane composition resulted in a consequent decrease in the biogas carbon dioxide composition, thus modifying the alkalinity of the liquid phase (pH was controlled at 7.2). Hence, non-stationary bicarbonate concentrations were observed throughout this initial stage (i.e. until reaching steady-state conditions around day 20 of the operational period).

Figure 3 shows the evolution throughout the operational period of the COD removal efficiency (Figure 3(a)) and methane yield (Figure 3(b)). From day 0 to 20, the COD removed in the system was attributed to the anabolism of anaerobic biomass (i.e. initial growth, fixation, and adaptation of the biomass to the new environmental conditions) (Michaud et al. 2002). During this initial period, high COD removal efficiencies were observed mainly due to the low OLR levels applied. From days 25 to 30, a decrease in the COD removal efficiency was observed. This behaviour was related to the previously-commented VFA accumulation in the reactor. Nonetheless, an increasing COD removal efficiency associated with the catabolism of methanogenic archaea was observed after day 30. On the other hand, a decrease in the methane yield from around 0.32 to 0.10 LCH4 STP·g−1CODREMOVED was observed from days 25 to 30 (see Figure 3(b)). In accordance with Michaud et al. (2002), this behaviour was attributed to possible process disturbances during the initial contact between the microorganism and the fixed support media, mainly related to attachment constraints during the biofilm formation and consolidation. Therefore, a minimum time of 20 days was required to reach a functional anaerobic biomass consortium, whilst a minimum time of 35 days was required for achieving a functional anaerobic biofilm. After day 35, a quite stable COD removal efficiency (up to 85%) was achieved, which resulted in a continuous increase in the methane yield throughout time. This behaviour highlighted the development and maturing of a stable biofilm.
Figure 3

Evolution throughout the operational period of (a) COD removal efficiency (total COD removed and COD removed for methane production) and (b) methane yield.

Figure 3

Evolution throughout the operational period of (a) COD removal efficiency (total COD removed and COD removed for methane production) and (b) methane yield.

Monitoring system development

Similarities between the evolution of electrical conductivity and both bicarbonate concentration and methane production rate were observed during the previously-analysed operating period. Hence, the evolution of electrical conductivity, bicarbonate concentration and methane production rate was analysed, resulting in linear relationships between these parameters. Since ion conductivity is sensitive to temperature, electrical conductivity was corrected to 25 °C (G25) as proposed by Colombié et al. (2007). Moreover, in order to account for possible variations in methane production due to variations in temperature and pressure, methane flow rate was expressed at STP. As commented before, increasing OLR were applied during the start-up period of the anaerobic fixed-bed reactor, and the evolution of G25, bicarbonate concentration and methane production rate was evaluated.

For an established G25 operating range (0.5 mS·cm−1 < G25 < 4 mS·cm−1), a linear regression (R2 coefficient of 0.97) between both bicarbonate concentration and G25 was observed (see Figure 4(a) and Equation (1)). Deviations between calibration and experimental data (see Figure 4(a), 1.1 mS·cm−1 < G25 < 2.3 mS·cm−1) were mainly related to the previously-commented accumulation of VFAs observed during the star-up period of the plant. 
formula
1
where (ΔBic)est is the estimated change in the bicarbonate concentration (mmol·L−1); (ΔG)25 is the measured change in G25 (mS·cm−1); and αBic/G is the regression coefficient including the confidence interval at 95% (37.01 ± 0.04 mmol·cm·L−1·mS−1).
Figure 4

Linear regressions obtained between (a) bicarbonate concentration and electrical conductivity, (b) methane production rate and bicarbonate concentration, and (c) methane production rate and electrical conductivity.

Figure 4

Linear regressions obtained between (a) bicarbonate concentration and electrical conductivity, (b) methane production rate and bicarbonate concentration, and (c) methane production rate and electrical conductivity.

Equation (1) highlights the possibility of using electrical conductivity sensors for on-line monitoring the bicarbonate concentration during the start-up period of an anaerobic fixed-bed reactor. Nevertheless, the use of these sensors could be enhanced by accounting for possible variations in other parameters such as VFA concentration since electrical conductivity is proportional to the total ion concentration (Aceves-Lara et al. 2012).

For an established Bic operating range (20 mmol·L−1 < Bic < 60 mmol·L−1), a linear regression (R2 coefficient of 0.89) between both bicarbonate concentration and methane production rate was also observed (see Figure 4(b) and Equation (2)). 
formula
2
where (ΔQCH4)est is the estimated change in the specific methane production from the measured bicarbonate concentration (LCH4 STP·d−1·L−1); (ΔBic) is the change in the bicarbonate concentration (mmol·L−1); and αQCH4/Bic is the regression coefficient including the confidence interval at 95% (0.0152 ± 0.0002 LCH4 STP·mmol−1·d−1).

The linear dependence shown in Equation (2) states the importance of bicarbonate concentration for starting-up an anaerobic fixed-bed reactor. On the other hand, an increasing dispersion of the monitored methane production rate was observed for bicarbonate concentrations higher than 60 mmol·L−1. Above this value, it was assumed that the process was completely started-up, thus methane production depended on different operating conditions.

Equation (3) is obtained by combining Equations (1) and (2), which illustrates that for an established G25 operating range (1.2 mS·cm−1 < G25 < 2.2 mS·cm−1), a nearly linear regression (R2 coefficient of 0.82) between both G25 and methane production is observed (see Figure 4(c)). Deviations between experimental and predicted values shown in Figure 4(c) (R2 coefficient was 0.82) could be related to variations in other operating variables (e.g. VFA concentration). 
formula
3
where αQCH4/G is the regression coefficient including the confidence interval at 95% (0.446 ± 0.005 LCH4 STP·cm·d−1·L−1·mS−1).

In this study, suitable anaerobic biofilm stability for methane production was observed when the system conductivity achieved values around 1.2 mS·cm−1. This value established the minimum starting-up time (20 days in our work, see Figures 2(b) and 3) required to develop a functional and stable anaerobic biomass consortium. In this respect, a minimum G25 value of around 1 mS·cm−1 was identified as the minimum value that predicted a suitable bicarbonate concentration to achieve a proper performance of anaerobic biomass consortia.

Finally, in order to validate the applicability of electrical conductivity sensors for monitoring the start-up period of this AD process, the obtained simple regression models were used to estimate both bicarbonate concentration and methane production rate in a new start-up period. Figure 5(a) shows the evolution of G25, the experimentally measured bicarbonate concentration, and the estimated bicarbonate concentration. Figure 5(b) shows the evolution of G25, the experimentally measured methane production rate, and the estimated methane production rate. These figures show the data evolution during the calibration period and the new start-up period. Around day 16 in Figure 5, a failure in the pH-control system caused an increase of pH in the system up to values around 10. Subsequently, the reactor was partially cleaned with water and the process was re-started up (a considerable amount of biomass was lost due to the disturbance). As can be observed in Figure 5, linear regression models were enough to successfully predict the bicarbonate concentration (R2 coefficient of 0.91) and the methane production rate (R2 coefficient of 0.81) from the measured electrical conductivity.
Figure 5

Validation results of the linear regression models to estimate both (a) bicarbonate concentration and (b) methane production rate.

Figure 5

Validation results of the linear regression models to estimate both (a) bicarbonate concentration and (b) methane production rate.

The results shown in this work point to electrical conductivity as a possible state indicator for the start-up period of an anaerobic fixed-bed reactor. It is important to highlight that, apart from their low acquisition cost, on-line electrical conductivity sensors are reliable, simple and require low maintenance for day-to-day monitoring and control (Vanrolleghem & Lee 2003). Although methane production rate is one of the most adequate state indicators for monitoring an AD process, measuring methane composition is commonly expensive (i.e. on-line and off-line methane analysers are much more costly and difficult to maintain than electrical conductivity sensors). Moreover, measuring bicarbonate concentration is not easy. Hence, as commented above, this work provides insights into the use of electrical conductivity as a possible state indicator for AD processes (e.g. electrical conductivity could help rapidly identifying possible process disturbances). Hence, the use of these low-cost industrially feasible on-line sensors would help engineers to handle day-to-day tasks in full-scale AD systems.

CONCLUSIONS

Linear relationships between electric conductivity and both bicarbonate concentration and methane production rate were observed. The obtained linear relationships successfully predicted both bicarbonate concentration and methane production rate from on-line electrical conductivity measurements. This highlighted the possibility of using electrical conductivity sensors as a cheap and simple method for on-line monitoring the start-up period of an anaerobic fixed-bed reactor.

ACKNOWLEDGEMENTS

This research work has been supported by the EU (FP7-SME-2008-1 called project ADD CONTROL 232302) and the Spanish Research Foundation (MICINN FPI grant BES-2009-023712), which is gratefully acknowledged.

REFERENCES

REFERENCES
American Public Health Association/American Water Works Association/Water Environment Federation
2005
Standard Methods for the Examination of Water and Wastewater
, 21st edn.
American Public Health Association/American Water Works Association/Water Environmental Federation
,
Washington, DC
,
USA
.
Boe
K.
Batstone
D. J.
Steyer
J. P.
Angelidaki
I.
2010
State indicators for monitoring the anaerobic digestion process
.
Water Research
44
(
20
),
5973
5980
.
Ellison
W. J.
Pedarros-Caubet
F.
Caubet
R.
2007
Automatic and rapid measurement of microbial suspension growth parameters: application to the evaluation of effect or agents
.
Journal of Rapid Methods & Automation in Microbiology
15
(
4
),
369
410
.
Hawkes
F. R.
Guwy
A. J.
Hawkes
D. L.
Rozzi
A. G.
1994
On-line monitoring of anaerobic digestion: application of a device for continuous measurement of bicarbonate alkalinity
.
Water Science and Technology
30
(
12
),
1
10
.
Heppner
B.
Zellner
G.
Diekmann
H.
1992
Start-up and operation of a propionate-degrading fluidized-bed reactor
.
Applied Microbiology and Biotechnology
36
(
6
)
810
816
.
Hoffmann
F.
Schmidt
M.
Rinas
U.
2000
Simple technique for simultaneous on-line estimation of biomass and acetate from base consumption and conductivity measurements in high-cell density cultures of Escherichia coli
.
Journal of Microbiology and Biotechnology
70
(
3
),
358
361
.
Michaud
S.
Bernet
N.
Buffière
P.
Roustan
M.
Moletta
R.
2002
Methane yield as a monitoring parameter for the start-up of anaerobic fixed film reactors
.
Water Research
36
(
5
),
1385
1391
.
Olsson
G.
Nielsen
M. K.
Yuan
Z.
Lynggaard-Jensen
A.
Steyer
J. P.
2005
Instrumentation, Control and Automation in Wastewater Systems
.
IWA Publishing
, London, UK.
Steyer
J.-P.
Bouvier
J. C.
Conte
T.
Gras
P.
Harmand
J.
Delgenes
J. P.
2002a
On-line measurement of COD, TOC, VFA, total and partial alkalinity in anaerobic processes using infra-red spectrometry
.
Water Science and Technology
45
(
10
),
133
138
.
Steyer
J.-P.
Bouvier
J. C.
Conte
T.
Gras
P.
Sousbie
P.
2002b
Evaluation of a four year experience with a fully instrumented anaerobic digestion process
.
Water Science and Technology
45
(
4–5
),
495
502
.
Vanrolleghem
P. A.
Lee
D. S.
2003
On-line monitoring equipment for wastewater treatment processes: state of the art
.
Water Science and Technology
47
(
2
),
1
34
.
Varley
J.
Brown
A. K.
Boyd
J. W. R.
Dod
P. W.
Gallagher
S.
2004
Dynamic multi-point measurement of foam behaviour for a continuous fermentation over a range of key process variables
.
Biochemical Engineering Journal
20
(
1
),
61
72
.
Weiland
P.
Rozzi
A.
1991
Start-up, operation, monitoring and control of high-rate anaerobic treatment systems: discusser's report
.
Water Science and Technology
24
(
8
),
257
277
.