1 international seminar on environmental odour management 2014 · santiago, 04.03.2014 1st...
TRANSCRIPT
Env. Odour ManagementSantiago, 04.03.2014
1st International Seminar1st International Seminaron Environmental Odour Management
20142014Santiago, Chile, 4‐5 March 2014g , ,
Considerations on Sampling and Measurement of Odours (VDI 3880 & VDI 3885/1) and enosesOdours (VDI 3880 & VDI 3885/1) and enoses
Prof. Dr.‐Ing. Franz‐Bernd Frechen,Prof. Dr. Ing. Franz Bernd Frechen,University of Kassel/Germany
Vice‐Chair, IWA Specialist Group “Odour and Volatile Emissions”(Chair from 1998 to 2011) und Bundessieger
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 1
(Chair from 1998 to 2011)Chair, DWA Committee “Emissions of waste water facilities”
u d u dess ege„Gesellschaft“
Env. Odour ManagementSantiago, 04.03.2014
VDI 3880“Olfactometry –Static sampling”
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 2
Env. Odour ManagementSantiago, 04.03.2014VDI 3880 “Olfactometry – Static sampling”
General Requirements on SamplingWorking conditions Sampling equipment Sampling bag Sampling bagPerformance of sampling for delayed olfactometryPredilution Sampling duration: 30 minutes Sampling duration: 30 minutesNumber of samples: min. 3 samples/source and operation condition Samples Storage: < 6h, or stability has to be proven: 6 measurements (3
i di t l ft li 3 ft d i d t ti diff fimmediately after sampling, 3 after desired storage time, difference of more than a factor of 1.5 is not acceptable)
Sample transport
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 3
Env. Odour ManagementSantiago, 04.03.2014VDI 3880 “Olfactometry – Static sampling”
Performance of sampling in relation of source typesDemarcation of active and passive sourcesActive sources (“diffuse sources” in EN 13725)Passive sources (“aerated diffuse sources” in EN 13725)Passive sources ( aerated diffuse sources in EN 13725)
Quality assurance
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 4
Env. Odour ManagementSantiago, 04.03.2014
air volume flow
source types according to VDI 3880
a o u e o
measurable5.2 Active sources
not measurable5.3 Passive sources
< 30 m/h>30 m/h
5 2 1 A ti i t5.2.1 Active point sourcesEN: Point Source
5.2.2 Active area sources 5.3.2 Passive area sources
EN: Fugitive Sources
EN: Aerated diffuse sources
5.3.3 Volume sources
EN: Diffuse Sources
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 5
Env. Odour ManagementSantiago, 04.03.2014types of sources
acti e pointpassive area source: e.g. landfill, wwtp, composting
active point source:e.g. stack
active area source:e.g. biofilter
volume source: e.g. building
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 6
Env. Odour ManagementSantiago, 04.03.2014active point: stack
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 7
Env. Odour ManagementSantiago, 04.03.2014active point: stack
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 8
source: DIN 4200 (Dec. 2000)
Env. Odour ManagementSantiago, 04.03.2014active area
Scheme of an active area sourceHood or total cover to be used
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 9
Env. Odour ManagementSantiago, 04.03.2014active area: hood or complete cover?
the advantage of a complete cover is that the entire waste gas stream is emitted via one defined orifice and the determination of the total source strength is possible by measuring here, taking only three samples The total odour flow rate can thus beonly three samples. The total odour flow rate can thus be reliably determined.
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 10
Env. Odour ManagementSantiago, 04.03.2014active area sources, e.g. biofilter: complete cover
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 11
Env. Odour ManagementSantiago, 04.03.2014active area: hood or complete cover?
the advantage of a complete cover is that the entire waste gas stream is emit‐ted via one defined orifice and the determination of the total source strength is possible by measuring here, taking only three samples The total odour flow rate can thus beonly three samples. The total odour flow rate can thus be reliably determined.
usage of the hood method is beneficial if, in addition to the usage o t e ood et od s be e c a , add t o to t eoverall odour flow rate, a meaningful statement is required on the condition of the source (e.g. biofilter function) and its
ti i ti O l ith i t li i hoptimization. Only with point sampling inhomogeneous areas (flow velocity, odour concentration) can be detected, delivering a better picture of the state of individual part areas and regions p p gof, e.g., a biofilter.
the decision depends on technical possiubility and question to
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 12
be answered!
Env. Odour ManagementSantiago, 04.03.2014active area sources, e.g. biofilter: hood
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 13
Env. Odour ManagementSantiago, 04.03.2014active area sources, e.g. biofilter: hood
how many points have to be sampled?Rectangular sources are subdivided into a grid of part areas that are as
square as possible. In the case of ground areas up to 100 m2, part areas are to be created of at f g p , p f
least approximately 10 m2 each. At least four part areas are to be formed on ground areas < 40 m2.
In the case of ground areas upwards of 100 m2 to 2 000 m2, at least one In the case of ground areas upwards of 100 m to 2 000 m , at least one further part area is to be additionally created per roughly 100 m2.
For ground areas > 2 000 m2 30 part areas are usually sufficient « For ground areas > 2 000 m , 30 part areas are usually sufficient.«
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 14
Env. Odour ManagementSantiago, 04.03.2014passive sources: how to sample ‐ no airflow present?
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 15
Env. Odour ManagementSantiago, 04.03.2014passive sources: how to sample ‐ no airflow present?
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 16
Env. Odour ManagementSantiago, 04.03.2014passive sources: how to sample ‐ no airflow present?
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 17
Env. Odour ManagementSantiago, 04.03.2014passive sources: how to sample ‐ no airflow present?
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 18
Env. Odour ManagementSantiago, 04.03.2014passive sources: how to sample ‐ no airflow present?
passive area source?or
active area source??
passive area source !!
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 19
Env. Odour ManagementSantiago, 04.03.2014VDI 3880 “Olfactometry – Static sampling”
Citerion set by VDI 3880 for use of sampling system on area sources: 30 m3/(m2 * h)
passive (diffuse) source: use flow‐through hoodith d l i
active (aerated diffuse) source:use sampling hood or complete cover
1
biofilter: 30 to more than 150 m3/(m2*h)
with odorless air
aeration tank: 2 to 7 m3/(m2*h)
0
0 10 20 30 40 50 60 70 80 90 100itt d i fl t i 3/( 2*h)
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 20
emitted airflow rate in m3/(m2*h)
Env. Odour ManagementSantiago, 04.03.2014passive area sources
Scheme of a passive area sourceflow-through sampling hood to be used
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 21
Env. Odour ManagementSantiago, 04.03.2014flow‐through hood on a passive area source
odorless air area related Odor Fl R t OFRtemperature &
cod = 0Qin = Qout = fixed
Flow Rate OFRqspez = cod * Qoutin ouE/(m²*h)
temperature & pressure:
inside = outside
odour emitting surface, e.g. wastewater, waste, sludge …
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 22
Env. Odour ManagementSantiago, 04.03.2014sampling with the flow‐through sampling hood
surface covered by box in m² flush air volume flow in m³/h
flushing rate in m³/(m² ∙ h)
odour concentration after sampling box in ouE/m³
area related odour flow rate qspez in ouE/(m²∙h)
total emitting area in m²odour flow rate OFR in ouE/h
only the knowledeg of the area related odour flow rate is reasonable and makes comparision possible!
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 23
p p
Env. Odour ManagementSantiago, 04.03.2014Lemberger box
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 24
Env. Odour ManagementSantiago, 04.03.2014test of hoods during elaboration of VDI 3880
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 25
Env. Odour ManagementSantiago, 04.03.2014test of hoods during elaboration of VDI 3880
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 26
Env. Odour ManagementSantiago, 04.03.2014test of hoods during elaboration of VDI 3880
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 27
Env. Odour ManagementSantiago, 04.03.2014VDI 3880 – hood 1 (no labyrinth)
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 28
Env. Odour ManagementSantiago, 04.03.2014VDI 3880
final agreement on the type of hood:No labyrinth Shaped like can be seen in sampleDimensions see next slideDimensions see next slide
1 inlet fan 4 outlet fan1 inlet fan 4 outlet fan 2 activated carbon filter 5 sampling port3 manometer 6 diffuser plate
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 29
(VDI 3880, 2011)
Env. Odour ManagementSantiago, 04.03.2014VDI 3880
final agreement on the requirements of the hood
inner dimensions of the flow ductL x W x H, in mm
1000 x 500 x 130
surface exposed to flow, in m² 0,5
mean flow velocity, in cm/s 6,4
contact time of air above surface,in s
15,6in s
area specific ventilation rate,in m³/(m²*h)
30
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 30
Env. Odour ManagementSantiago, 04.03.2014
VDI 3885/1VDI 3885/1“OlfactometryOlfactometry –
Measurement of the odour emission capacity
of liquids”of liquids
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 31
Env. Odour ManagementSantiago, 04.03.2014Where does the odour come from?
Industry Sewer System, esp. pressure mains Wastewater and Sludge Treatment and Disposal I li id h i i l d
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 32
In may cases, liquids are the original odour sources
Env. Odour ManagementSantiago, 04.03.2014Odour measurement in the liquid: The OEC
The Odour Emission Capacity OEC of a liquid is the totalThe Odour Emission Capacity OEC of a liquid is the total amount of odorants present in that liquid that can be
stripped from this liquid under standardized conditions. [Frechen and Köster, Wat.Sci.Tech.,1998]
It is given in European odour units per cubic meter of liquid. Its unit is ou /m3Its unit is ouE/m3
liquid
The OEC method is covered by the new VDI guideline 3885/1. It will be y g /published 2014
As we strip off the odorants into air samples, quantification can be done b th t d d lf t t i th d di t CEN 13725 2003by the standard olfactometric method according to CEN 13725:2003
Also, different analytical procedures can be applied to the stripped air, lti i E i i C iti f th b t l d
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 33
resulting in Emission Capacities of the substances analysed
Env. Odour ManagementSantiago, 04.03.2014OEC test reactor
ff l & li ioff gas outlet & sampling point
headspace
liquid
< 15 Liter
qsample30 Liter
© DESEE
fine bubbleaerator
odorless airca 0 30 m
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 34
ca. 0.30 m© DESEE
Env. Odour ManagementSantiago, 04.03.2014evaluation – type a
2 000t
1 500
2 000olfaktometrische Meßwerte
GE E
/m3 Lu
ftE/m
3 air olfactometric results
1 500
gemessen
tion
in G
c odin ou E
measured
A 1
1 000
onze
ntra
entration
test duration
500
100hsst
offk
oou
r con
ce
A 2
A 3
00 500 1 000 1 500 2 000
Luftmenge in Liter
100
Ger
uch
odo
cumulated airflow in Liter
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 35
Luftmenge in Litercumulated airflow in Liter
Env. Odour ManagementSantiago, 04.03.2014evaluation – type a
the measurement curve does 2 000olfaktometrische Meßwerte
m3 Lu
ftair olfactometric results
fall below the „lower integrationlimit LIL“ of cod = 100 ouE/m3
air
h OEC l h1 000
1 500
gemessen
ntra
tion
in G
E E/m
ionc o
din ou E/m
3
measured
A 1
the OECtotal equals theOECmeasured (blue area) andis the area (integral) between
500
100chss
toffk
onze
ndo
ur co
ncen
trati
A 2
A 3
test duration
s t e a ea ( teg a ) bet eethe LIL (horizontal line atcod = 100 ouE/m3
air) and the measurement curve
00 500 1 000 1 500 2 000
Luftmenge in Liter
Ger
ucod
cumulated airflow in Liter
the OECtotal is calculated as the sum of areas A1 to A3
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 36
Env. Odour ManagementSantiago, 04.03.20142nd + 3rd SOU in operation at the Bottrop Sewer
© Frechen© DESEE
Purpose: avoid overdosing avoid overdosing avoid underdosing
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 37
© DESEE
Env. Odour ManagementSantiago, 04.03.2014Research‐SOU in operation in Berlin at BWB+KWB
© Frechen© DESEE
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 38
© DESEE
Env. Odour ManagementSantiago, 04.03.2014
electronic noses
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 39
Env. Odour ManagementSantiago, 04.03.2014can eNoses measure the olfactometric odour conc.?
The perfect eNose would be able to measure the REAL odourconcentration. This means that a mathematical model exists that converts 4 to 10 sensors readings into one number (the REAL odour concentration)odour concentration)
To establish such a model, it is necessary to do parallel measurements witheasu e e ts t eNose olfactometer
Th ll l di (th bi th th ti l l The more parallel readings (the bigger the mathematical sample size), the better the mathematical model can be
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 40
Env. Odour ManagementSantiago, 04.03.2014DESEE’s data pool
DESEE’ datapool: 5 eNose brands 3 olfactometry labs Different origins of odours Different origins of odours
project eNose period olfactomery origin of odour numberVienna A 2005 Lab 1 wastewater 382Vienna B 2005 Lab 1 wastewater 156Vienna B 2005 Lab 1 wastewater 156Hilter A 4/2008 ‐ 5/2008 Lab 2 wastewater 40emscher sewer A 1/2007 ‐ 1/2008 Lab 2 wastewater 150Biofilter C 2009 Lab 2 wastewater 12OEC with wastewater A,A 2010 Lab 2 wastewater 88safety research part 1 A,A 2007 drugs/explosives (8)safety research part 2 A,A 2010 drugs/explosives (40)l b t t C 2009 h d l hid (20)lab test C 2009 hydrogen sulphide (20)Verden A,A 2010/2011 Lab 3 wastewater 110Aachen A,A 2011 Lab 3 wastewater 156Berlin A,B,D,E 2011/2012 Lab 3 wastewater 1429
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 41
Berlin A,B,D,E 2011/2012 Lab 3 wastewater 1429total with olfactometry: 2.523
Env. Odour ManagementSantiago, 04.03.2014Possible sources of problems and errors
Sampling Uncertainty of olfactometric measurement The eNose itself The odour’s nature The mathematical evaluation
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 42
Env. Odour ManagementSantiago, 04.03.2014Possible sources of problems and errors
Uncertainty of olfactometric measurement It has to be differentiated between THE REAL ODOUR and THE ODOUR
MEASURED BY OLFACTOMETRY Olfactometry measurement has an error itself. According to Boeker (2005) y g ( )
this can be estimated to be 50%. This means: if an eNose perfectly explains the odour as measured by
olfactometry, (claimed by some vendors of eNoses: R2 = 1 !!), the modelolfactometry, (claimed by some vendors of eNoses: R 1 !!), the model has an error of 50%!
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 43
Env. Odour ManagementSantiago, 04.03.2014Possible sources of problems and errors
The eNose itself t90‐time
no information provided by the manufacturers t90‐time differs from sensor to sensor90
t90‐time differs from odour to odour Some sensor signals do not converge to a final value at all – then there is no
t90‐time at all …
60 000
70.000
80.000
90.000
g
11.500
12.000
12.500
g
30.000
40.000
50.000
60.000
sensor re
ading
10.000
10.500
11.000
sensor re
ading
0
10.000
20.000
1 2 3 4 5 6 7 8 9 10111213141516171819202122232425262728293031323334353637383940414243
time [s]
9.000
9.500
1 2 3 4 5 6 7 8 9 10111213141516171819202122232425262728293031323334353637383940414243
time [s]
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 44
Env. Odour ManagementSantiago, 04.03.2014Possible sources of problems and errors
The eNose itself t90‐time oscillating readings – mean used
10 380
10.400
10.420
10.340
10.360
10.380
or re
ading
10.300
10.320
senso
10.260
10.280
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 82 85 88 91 94 97 100
103
106
109
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 45
time [s]
Env. Odour ManagementSantiago, 04.03.2014Possible sources of problems and errors
The eNose itself t90‐time oscillating readings – mean used do 2 eNoses of the same type and age deliver the same readings? do 2 eNoses of the same type and age deliver the same readings?
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 46
Env. Odour ManagementSantiago, 04.03.2014Possible sources of problems and errors
The eNose itself t90‐time oscillating readings – mean used do 2 eNoses of the same type and age deliver the same readings? do 2 eNoses of the same type and age deliver the same readings? aging of sensors
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 47
Env. Odour ManagementSantiago, 04.03.2014Possible sources of problems and errors
The odour’s nature Although already well established in several (specific) industrial processes,
eNoses still have problems with the very varying composition of odours –e.g. wastewater odours, but also othersg ,
0,8
11
210 cocaine_solo
0,2
0,4
0,6
39
cocaine_beer
0
48
5
6
7
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 48
6
Env. Odour ManagementSantiago, 04.03.2014Possible sources of problems and errors
Mathematical evaluation of the raw data is the crucial step in eNose usage. Methods may aim forQualification (“which kind of smell is that”)
100%S1 (aromatic) AK Bottrop
40%
60%
80%
100%
S2 (broadr.)
S3 (aromatic)S9 (sulphur, chlor)
S10 (methane-aliph)
Schwarzbach
0%
20%( )
S4 (hydrogen)S8 (broad, alcohol)
( p , )Druckrohrleitung
Biogasanlage: Mais-Silo (feucht)
S5 (arom-aliph)
S6 (broad, th )
S7 (sulphur, organic)
Silo (feucht)
Bauernhof: Gülletank
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 49
methane)Legende: S1…S10 sind die Sensoren des Multisensormessgerätes
Env. Odour ManagementSantiago, 04.03.2014Mathematical evaluation
Mathematical evaluation of the raw data is the crucial step in eNose usage. Methods may aim forQualification (“which kind of smell is that”) Pattern recognition with
100%S1 (aromatic) AK Bottrop
Pattern recognition with● Configuration frequency analysis● Shape analysis● Neural networks
40%
60%
80%
100%
S2 (broadr.)
S3 (aromatic)S9 (sulphur, chlor)
S10 (methane-aliph)
Schwarzbach
0%
20%( )
S4 (hydrogen)S8 (broad, alcohol)
( p , )Druckrohrleitung
Biogasanlage: Mais-Silo (feucht)
S5 (arom-aliph)
S6 (broad, th )
S7 (sulphur, organic)
Silo (feucht)
Bauernhof: Gülletank
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 50
methane)Legende: S1…S10 sind die Sensoren des Multisensormessgerätes
Env. Odour ManagementSantiago, 04.03.2014Mathematical evaluation
Mathematical evaluation of the raw data is the crucial step in eNose usage. Methods may aim forQualification (“which kind of smell is that”) Pattern recognition with Pattern recognition with
● Configuration frequency analysis● Shape analysis● Neural networks
Quantification (“ouE/m3”) Partial Least Square PLS Linear Regression Nonlinear Regression Logistic Regression (below or above a given limit value) Neural networks (able to achieve nearly perfect explanation, but ability for
prediction is crucial, see below!)
in this presentation, we deal with quantification
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 51
Env. Odour ManagementSantiago, 04.03.2014Simple mathematical errors (example)
selection of data used for evaluation: The smaller the mathematical data sample size, the bigger the possibility
to get a deterministic model. In the case of ten sensors and eleven unknown coefficients in a model one always gets around 100% correct y gclassification, if using eleven measurements or less.
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 52
Env. Odour ManagementSantiago, 04.03.2014eNoses: caution with vendor’s claims ...
Most important with mathematical models is to distinct between explanation (done in nearly 99% of cases) andprediction capabilities determination by data splitting!
If vendors claim that their eNose is capable of measuring odour If vendors claim that their eNose is capable of measuring odourwith a correlation coefficient of near 1, they usually talk about explanation of historical data (ouE/m3 measured using olfactometry vs. ouE/m3 calculated from the eNose raw data)
There are no vendors known that really check for prediction abilities If so they should testify that theyabilities. If so, they should testify that they used the bootstrap‐procedure: derive a model for explanation basing on
part (e.g. 50%) of the data (training data) (this is the explanation part, ibl ith hi h l ti ffi i t ) dpossibly with high correlation coefficients), and
applied this model to the other part of the data for prediction.
Correlation coefficients near 1 for prediction are not achievable
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 53
as set out before (REAL odour vs OLFACTOMETRIC odour)
Env. Odour ManagementSantiago, 04.03.2014
Correlation using model derived from training data (dataset 1)(explanation)
m3
eNos
eou
E/m
/ 3
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 54
ouE/m3 measured
Env. Odour ManagementSantiago, 04.03.2014
Application of the model to the test data set (dataset 2)(prediction) ...
m3
eNos
eou
E/m
/ 3
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 55
ouE/m3 measured
Env. Odour ManagementSantiago, 04.03.2014
Thank you for your patient attention !Thank you for your patient attention !
www.uni‐kassel.de\fb14\siwawi
Department of Sanitary and Environmental Engineering (DESEE) Head: Prof. Dr.‐Ing. F.‐B. Frechen www.uni-kassel.de/fb14/siwawiSl
ide 56