Electronic Nose for Reactor Stability Monitoring of an Agricultural Co-
digestion Biogas Plant
Environmental Sciences and Management Department
University of Liège
G. Adam S. Lemaigre P. Delfosse A-C. Romain
1 Progress in biogas III, 10-11 September, 2014, Stuttgart
2 Pearce et al., 2003
What is an electronic nose (e-nose) ?
It’s an array of complementary low-specificity gas sensors increased specificity using sensor array pattern (like a signature)
Introduction
Why e-nose technology?
Introduction
3
• Anaerobic digestion process monitoring:
• Online monitoring: [CH4], [CO2], biogas production, pH
• Offline analysis: alkalinity, Volatile Fatty Acids (individuals/total), etc.
• No online tool for early warning of anaerobic digestion process disorders
• E-nose advantages:
• Online monitoring
• Gas phase sampling (easier than liquid-phase sampling in anaerobic reactors)
• Rapid turn-over of gas phase of the reactor (hours)
Sample Transport
(days)
Register sample (hours)
Measure (hours-days)
Results transmission
Decision
Sample Measure (minutes)
Decision
Adapted from Holm-Nielsen, 2008
Actual situation
Ideal situation
Phase I: 100 L pilot-scale CSTR monitoring
4
e-nose = array of 6 low-specificity gas sensors and a dilution system (25x)
Liquid phase Total solids [%], volatile solids [%TS] pH Alkalinity [ml CO2] NH4
+ [g L-3]
Gas phase CH4 [%], CO2[%], H2S (ppm), H2 (ppm) E-nose
Material and methods
Phase II: Full-scale reactor monitoring
5
Material and methods
6
Material and methods
E-nose: array of 7 low-specificity commercial gas sensors (Figaro Engineering inc.)
During e-nose monitoring (650 days): weekly: VFA, VS, TS, VFA/TIC (FOS/TAC) Every 2 weeks: total ammonia nitrogen (TAN)
7
Faascht farm (BE)
Co-digestion biogas plant of 750 kW 3 CSTR + two storage tanks + Digestate drying unit Substrates (18 000 T): •Food industry waste (54 %) •Cattle manure/slurry (33 %) •Maize silage (8 %)
Limited process monitoring capabilities: On-line: CH4, CO2, H2S and O2 (prior to CHP) When low gas quality/production: VFA, N-NH4
+ in the sludge
Material and methods
8
Results – Pilot-scale monitoring
0
0.5
1
1.5
2
2.5
3
0
1
2
3
4
5
6
7
8
TIC
(m
l CO
2 g
-1, N
-NH
4 (
g kg
-1)
pH
pH
TIC
N-NH4
0
2
4
6
8
10
OLR
(gV
S L-1
d-1
), g
as
pro
du
ctio
n r
ate
(m³
m-3
d
ay--1
)
OLR
biogas production
0
2
4
6
8
10
12
0 10 20 30 40 50 60 70 80 90 100 110 120
T2/l
im9
9 ,
SPE/
SPEl
im9
9
Time (days)
ratioT2/lim99
ratioQ/lim99
limit
Alkalinity decrease
pH decrease
E-nose indicators
9
0
20
40
60
80
100
CH
4 a
nd
CO
2 c
on
ten
t (%
)
CH4 CO2
0
1
10
100
1000
10000
0
2000
4000
6000
8000
10000
H2
(p
pm
)
H2
S (p
pm
)
H2S
H2
0
2
4
6
8
10
12
0 10 20 30 40 50 60 70 80 90 100 110 120
T2/l
im9
9 ,
SPE/
SPEl
im9
9
Time (days)
ratioT2/lim99
ratioQ/lim99
limit
Alkalinity decrease
pH decrease
Results – Pilot-scale monitoring
CH4-CO2
H2S-H2
E-nose indicators
10
Results – Pilot-scale monitoring
Scores PC1 (44.5%)
Sco
res P
C2
(2
3.8
%)
-70 -60 -50 -40 -30 -20 -10 0 10-350
-300
-250
-200
-150
-100
-50
0
50Stable pH and TAC (days 1 to 78)
Stable pH, decreasing TAC (days 78 to 98)
Decreasing TAC and pH (days 98 to 113)
day 102
days 105-106
days 109-110
days 113
Stable operation cluster
PCA monitoring Different changes in the gas phase and liquid phase are observed by the e-nose
TIC decrease
Phase II: Full-scale reactor monitoring
11
Results
12
0
1000
2000
3000
4000
5000
6000
VFA
an
d T
AN
[m
g L-
1]
Acetate
Propionate
TVFA
TAN
6.5
7
7.5
8
8.5
9
0.00
0.10
0.20
0.30
0.40
0.50
0 50 100 150 200 250 300 350 400 450 500 550 600 650
pH
VFA
/TIC
[-]
Time (days)
VFA/TIC
pH
VFA
Unknown cause CH4 CHP interruption
VFA
TAN
Results – real-scale monitoring
1 2 3 4
Signal drift decreased model perfomance
13
Results – real-scale monitoring
1 2 3 4
Adaptive model: detection of variation in process state
14
ok ok ??
Results – real-scale monitoring
1 2 3 4
1. Low gas quality and production. One engine turned off
2. VFA > 3500mg/L
3. Emptying and refilling reactor
4. VFA > 4000mg/L T-NH3> 3500 mg/L
5
15
Conclusions and Perspectives
Highlights
• Gas phase monitoring should be considered to assess anaerobic digestion reactor state
• The e-nose could detect process AD disorders by monitoring the gas phase at the pilot-scale level
• A simple indicator, derived from the complex e-nose data, summarizes reactor state
• At the real-scale level, the e-nose failed for robust monitoring of the reactor state
16
Thanks for your attention
LABORELECLABORELECLABORELEC
Biogas Rohlingerhof
L’Europe investit dans votre avenir
Projet cofinancé par l’Union Européenne via le FEDER dans le cadre du programme INTERREG IV-A Dieses Projekt wird von der EU über den EFRE-Fonds im Rahmen des Programms INTERREG IV-A kofinanziert
European Project Interreg IVa
ECOBIOGAZ (2012-2015) www.ecobiogaz.eu
OPTIBIOGAZ (2009-2012)
www.optibiogaz.eu
Acknowledgements