HIT-Norge kursus Laboratorie Ingeniører
Future Climate ?
Det CO2-neutrale samfund !
Hvordan skaber vi en bæredygtig udvikling?
Bioenergi er en alsidig og ikke helt uvæsentligt spiller i
det internationale energi mix i fremtiden!
Jens Bo Holm-Nielsen
Ph.D., Center for Bioenergi og Green Engineering
Institute for Energy Technology
Aalborg University, Esbjerg Institute of Technology
Niels Bohrs vej 8, 6700 Esbjerg
Cell; +45 2166 2511
E-mail: jhn@et.aau.dk
www.acabs.dk; www.iet.aau.dk;
1
Process Analytical Technologies for
Anaerobic Digestion Systems
- Robust Biomass Characterisation, Process Analytical
Chemometrics, and Process Optimisation
Jens Bo Holm-Nielsen, Ph.D.
Head of Center of Bioenergy and
Green Engineering
ACABS Research Group: Applied Chemometrics, Analytical Chemistry, Applied Biotechnology &
Bioenergy, and Sampling Research Group (ACABS) , Esbjerg Institute of Technology, Aalborg University
2
Biogas for a sustainable clean environment
and renewable energy production
LIGHT
PHOTOSYNTHESIS
O2
ANIMAL
MANURE
ORGANIC WASTE
Source: JBHN/TAS
CO2
BIOFERTILISER
BIOGAS
PLANT
H2O
CHP-GENERATION
BIOGAS AS VEHICLE FUEL
3
Estimated amounts of animal manure in EU-27 (based on Faostat, 2003)
Country
Cattle
Pigs
[1000Heads]
Cattle
[1000Heads]
Cattle
manure
Pigs
1000livestock units
1000livestock units
Pig
manure
Total
manure
[106 tons]
[106 tons]
[106 tons]
Austria
2051
3125
1310
261
29
6
35
Belgium
2695
6332
1721
529
38
12
49
2
11
Bulgaria
672
931
429
78
9
Cyprus
57
498
36
42
1
1
2
Czech R.
1397
2877
892
240
20
5
25
Denmark
1544
13466
986
1124
22
25
46
Estonia
250
340
160
28
4
1
4
Finland
950
1365
607
114
13
3
16
France
19383
15020
12379
1254
272
28
300
Germany
13035
26858
8324
2242
183
49
232
Greece
600
1000
383
83
8
2
10
Hungary
723
4059
462
339
10
7
18
Ireland
7000
1758
4470
147
98
3
102
Italy
6314
9272
4032
774
89
17
106
Latvia
371
436
237
36
5
1
6
Lithuania
792
1073
506
90
11
2
13
Luxembourg
184
85
118
7
3
0
3
18
73
11
6
0
0
0
Netherlands
3862
11153
2466
931
54
20
75
Poland
5483
18112
3502
1512
77
33
110
Portugal
1443
2348
922
196
20
4
25
Romania
2812
6589
1796
550
40
12
52
Slovakia
580
1300
370
109
8
2
11
Slovenia
451
534
288
45
6
1
7
Spain
6700
25250
4279
2107
94
46
140
Sweden
1619
1823
1034
152
23
3
U.K.
10378
4851
6628
405
146
9
EU-27
91364
160530
58348
13399
1284
295
Malta
4
26
155
1578
Energy potential of pig and cattle manure in EU-27
Total
manure
[106
tons]
1,578
Biogas
Methane
Potential
Potential
[106
m3]
[106 m3]
[PJ]
[Mtoe]
31,568
20,519
827
Methane heat of combustion: 40.3
18.5
MJ/m3;
1 Mtoe = 44.8
PJ
Assumed methane content in biogas: 65%
Biogas
Production
&
Forecast:
Actual 2008 production of biogas in EU 27:
7 Mtoe
2012-2015 EU forecast
15 Mtoe
Manure potentials
18.5-20 Mtoe
Organic waste and byproducts
15-20 Mtoe
Crops and crop residuals
20-30 Mtoe
Total long term forcast Biogas
60 Mtoe
Biogas can cover 1/3 of EU’s total RES 20% demands year 2020
5
AD Co-digestion heterogeneous
feedstock’s
- Manure
- Food waste
- Organic by-products
- Crops
6
Fig. 4.3: The four principal process steps in the general anaerobic digestion – biogas
production process. The dashed boxes indicate important intermediate compounds
(Holm-Nielsen et al 2008, in. prep).
7
Comparison of analytical strategies for process monitoring
(Mortensen 2006)
Time consuming versus on-line real time measurements!
Different PAT/PAC
laboratory approaches
and strategies
- Off-line;
- At-line;
- On-line;
(McLennan 1995)
8
Fundamental disciplines of Process Analytical Chemistry (PAC)
(Mortensen, 2006)
Numerous technologies can be applied in a PAT measuring programs for process
understanding and controlling. The technologies can be categorized in four major areas:
1. Technologies that imply use of Process Analytical Technology or Process
Analytical Chemistry
2. Technologies for monitoring and control of the process and end products
3. Technologies for continuous improvement of gained process knowledge
4. Technologies for acquisition and analysis of multivariate data
( FDA - PAT Guidance, 2005)
9
Ribe Biogas Plant; a full scale test facility for several R&D
projects. Sampling points for feedstock’s and inoculums
10
Paper 1; Introduction of TOS correct sampling
& on-line PAT measurements in full scale applications
Composite sampling !!!
Schematic illustration of primary sampling at the full-scale biogas plant in which a two-step
composite sampling approach was used. Eight 10L primary increments were individually
mass-reduced to 1L, before being compounded. Mechanical agitation is essential to keep the
bio-slurries in a state of maximum homogenization while being sub-sampled. This compound
sampling scheme is in full accordance with the principles of TOS.
11
Sampling unit operations:
TOS-correct sampling
1. Structurally correct sampling is the only safeguard
against sampling bias
2. Heterogeneity characterization of 0-D lots
3. Homogenization, mixing, blending
4. Composite sampling
5. Representative mass reduction
6. Particle size reduction (comminution or crushing)
7. Lot dimensionality transformation (3D or 2D → 1-D or 0-D)
(Petersen et al., 2005)
12
Transflexive embedded NIR recurrent loop
measurement system; developed by CAU, Kiel, D
Figure 4. TENIRS stand-alone prototype, view from the front and back, from Andree et al.
(2005).8 The TENIRS system consists of: (1) ZEISS CORONA 45 NIR spectrophotometer;
(2) measuring cell; (3) pump; (4) multi-way valve; (5) sample holder with 1 L container; (6)
frequency converter; (7) Control PC.
13
Primary sampling point
developed for on-line laboratory and
meso-scale R&D projects (KAU-AAUE)
Fig. 7.8. Prototype sampling device. 10 increments each of 10 ml were sampled during a period
of 10 minutes – an effective composite sample for chemical reference analysis.
(Holm-Nielsen et al. 2007)
14
Flow-through measuring cell, TENIRS
Remarks: Tested path length 3 mm and 6 mm. Learning process during these trials to change from transflection
towards pure reflection measurements in the heterogenious bio-slurries and other brown-black liquid medias.
15
Figure 5: Raw NIR spectra of all original 63 samples, spectra were used as log (1/R).
Because of the low resolution nature of the TENIRS spectra, expressed as very broad,
continuous peaks and valleys, there was found no need for more specific pre-treatments,
16
e.g. derivatives or MSC, see text. Note one gross - and several minor outliers.
17
Where:
E 
hc
E is the energy,
h Planck’s constant,
c the speed of the light, and
λ the wavelength.

18
Figure 7.9 Principles behind creation of a multivariate calibration models
19
Figure 7.10. Steps in PSL-regression. During step 1 - a
multivariate calibration model is generated. In step 2 the calibration
model from step 1 is applied on new X-data in order to predict the
corresponding unknown y-data, (Petersen, 2005).
Important: A 3. step for real full scale implementation and process controlling: Y
predicted tested versus Y test-set validation – needed routine validation in on-line Food, 20
Fuels, Pharma industries etc
Paper 2; Off-line & simulated at-line PAT studies
of important intermediates of the AD process. Codigestion of manure and energycrops/maize silage.
A
At-line
testing of 4 anaerobic digestion fermenters – key analytes
The Reidling biogas plant. Fermentor 1 is the flat roofed fermentor to the left. This
fermentor have been in full operation a year. The fermentor in the center part of the photo
is fermentor number 2, which was started in February 2005, photo JBHN, 03/05 21
”Unfortunately, we had to ….”
Fermentor no. 2 at the Strem biogas plant. Insulation was not completed due to the fact that
the mesofilic fermentation process generated surplus of heat for keeping the process
temperature and even slightly increase in temperature was registrated. Inoculum at this biogas
plant was cow manure, but after start up the feedstock was almost maize silage. Photo JBHN,
22
03/05.
Examples of
incorrect sampling
points in fermentation
systems in biotech.
- Food-feed-fuels-pharma
industries
from: Esbensen & Mortensen: ”TOS – the missing
link in PAT”
in: Bakeev (Ed.) ”PAT” (Blackwell, 2009)
23
ACABS’ recurrent loop
Generalised on-line PAT measurement and simultaneous TOS-sampling concept
- necessary requirement for optimal multivariate calibration – i.e. lowest RMSEP
24
Figure 6: PCA score plot of fermentor data sets from both the Reidling (R) and Strem (S) locations
based on TENIRS spectra.
One outlier described in the text has been excluded (Strem58). [S1, S2] and [R1, R2] signifies the25
number 1 and 2 reactors of either locations respectively (see text).
Figure 7: PLS1 prediction model, Y = volatile solids content (VS); Two segment
cross-validation. Two PLS-components. TENIRS data. One outlier removed (Strem 58).
26
Figure 9. PLS1 prediction model. Y = Acetic acid; 2-segment crossvalidation. 6 PLS-components. 14 minor outliers removed.
Figure 8. PLS1 pred. model. Y = Ammonium-N; 2-segment crossvalidation. 4 PLS-components. One outlier removed (Strem 58).
Data from Austrian biogas plants 2005
•
•
Co-digestion AD plants
Very heterogeneous
•
•
Many analytes
influencing the signals
•
•
•
•
There will always
be an uncertanty
in the rage of
10-20 %
2xRMSEP: 20%
Reference: Holm-Nielsen, J.B., Andree, H., Lindorfer, H., Esbensen, K.H. (2007). Transflexive embedded near infrared monitoring
for key process intermediates in anaerobic digestion/biogas production, Journal of Near Infrared Spectroscopy, vol. 15, pp. 123-135,
27
DOI: 10.1255/jnirs.719
Paper 3: Glycerol spiking trials in 3 parallel
fermentors, full on-line PAT lab-trials 2006.
Figure 4. Glycerol development during the AD process trials, shown for each individual fermenter.
Concentration measured in the fermenter effluent.
28
Paper 3: Glycerol spiking trials in 3 parallel
fermentors, full on-line PAT lab-trials 2006.
Figure 5. Volatile fatty acids contents. Development for each fermenter.
29
Figure 7. Glycerol PLS-1 model; number of required PLS components = 2; One outlier was removed;
Test set validation was made from data obtained in fermenter no.2 and tested against data from
fermenter no.1 and no.3; Measures of precisions r2 = 0.96 and slope 1.04
30
Figure 10. Total VFA PLS-1 model; number of required PLS-components = 3; Two outliers were removed;
Test set validation where data from the fermenter no.2 were tested against calibration data from fermenter
31
no.1 and no.3; measures of precision: r2 = 0.98, and accuracy; slope = 1.03
Paper 4: On-line PAT monitoring
– meso-scale anaerobic digestion trials.
ACABS/AAUE research group and Research Center Bygholm(DJF)/AU
Trial plan: 1. Increased loading rate of pig manure and 2. Sudden changes – increase in
temperature; for developing critical data spanning; total trail period 37 days.
32
Sampling facility. Primary sampling – 10 increments.
Secondary sampling for wet chemistry 5 increments for each vial
33
34
Spanning the analytes; min. and max. values measured in the lab.
Concentration level of acetate and propanoic acid and corrected
biogas production during the trial period.
35
Model for probionic acid; MSC corrected, test set validation
36
Statistics from best calibration models; MSC corrected, test set validation
37
Linkogas full scale trials, 2007 - 2008
- on-line PAT monitoring fermenter 3, 2400 m3
38
Linkogas trials: Experimental planning
Dates
Gly [%]
Gly [t/w]
Gly [flow]
Expected VFA's
10-10-2007
0,00 %
0,00 t/w
0,00 kg/h
None
15-10-2007
0,80 %
7,5 t/w
60,0 kg/h
Low
23-10-2007
1,50 %
13,6 t/w
113,1 kg/h
Medium
01-11-2007
2,00 %
81,1 t/w
150,8 kg/h
High
22-11-2007
2,50 %
22,6 t/w
188,5 kg/h
Very High
Illustration of the sampling procedure, incremental and fractional shovelling sampling
39
Linkogas full scale trials, 2007
- on-line PAT monitoring fermenter 3, 2400 m3
Figure 7.12.Volatile fatty acids concentration v.s. biogas production
(Conc. mg/L, Prod. m³)
Biogas production measures to the left hand side and VFA acid
concentration levels to the right.
40
Linkogas full scale trials, 2007
- on-line PAT monitoring fermenter 3, 2400 m3
VFA - PLS-1 model
Figure 7.13. Measured vs. Predicted plot for the total VFA model. The black line (the diagonal) indicates the target line, while the blue line is the
regression line. The VFA PLS-1 model; number of required PLS-components = 10; Two-segment cross validation was used due
41to low
concentration dataset, and low spanning. Measures of precision: r2 = 0.83, and accuracy; slope = 0.92 (Holm-Nielsen et al. 2008).
Thesis conclusions
and some reflections
• It is possible to make satisfactory models of important AD-intermediates
like VFA’s and ammonium, documented by several Bioenergy/ACABSgroup studies
• Biogas – AD production is a complex biological process, which can be
monitored and controlled much more advanced and robust/simple in the
future. It is one of the most suitable processes for regulation of supply
integration in the energy sector of the RES sources.
• The AD process is a key technology in biorefinering. It shows a huge
potential in the future energy supply chain: real sustainable energy
production
• These studies have demonstrated the direct potential for a fast track to
full process monitoring, regulation and control
• Next step will be to implement PAT technologies also in the biorefinery
sector, including developing process test platforms – there is an urgent
need for more test facilities implemented in full scale operations
• Contribution to on-going R&D&D co-operations between universities
and bioenergy industries - regionally, nationally and internationally
• Bioenergy potentials in sustainable combination with other RES systems
(wind-power, solar-power, hydro-and wave-power) will be able to supply
~50-100% of the entire world-wide energy supply in the medium to long
term!
42
Full-scale R & D Test facilities
PAT - platform in biorefinery projects
Biorefinery test facilities (complete process control loops and facilities).
Process platform means here main process steps;( process steps 1,...n.),
(Njoku & Holm-Nielsen, 2008)
43
Biorefineries for the future
1-2-3 generations! Challange fast replacements of oil refineries
Figure.8.1 Biorefinery concept. Biorefineries are complex production clusters where the sources (raw materials) are e.g. biomass from agriculture
forestry and society, resulting in a broad range of processed products, for example foods / feeds / fuels / fibres and fertilizers. The figure 44
illustrates
typical stages /treatments in integrated biomass utilization systems (Nielsen C. et al. 2005). Pre-treatment of ligno-cellulosic materials is crucial for future
biorefining.
Thank you for fruitful cooperation to all partners:
Danish and International Companies,
ILV-University of Kiel, D, - IFA-Tulln, BOKU, A,
DJF – Aarhus University, - Risoe/DTU,
Danish Biogas Plant Owners – Linkogas, Ribe Biogas, Thorsoe
Students and staff at the Bioenergy and Chemistry
laboratories; Esbjerg Institute of Technology,
Aalborg University
Special thank you to Kim H. Esbensen (the thesis supervisor), & Torben Rosenørn and
Lars Døvling Andersen at the INS-Faculty at AAU, to make this R&D&D work possible.
Thank you for your attention!
45