Programme Filires Biologiques
Trying to see through the black-box observation and data collection for modeling a biofiltration process
Pascal BoissonModel-based optimization of biofilm systems 29/05/09
2Context
Attached biomass processes are : Diverse (MBBR, BAF, MABR, etc.) Complex
High level of heterogeneities Quantitatively important interactions of qualitatively different
phenomenon
And they combine several advantages : Compact and intensive (BAF) Ease of wwtp upgrade options (IFAS) Interesting features (e.g. solids residence time)
Central treatment process in present and future wwtpdesign
3Context
Nevertheless :Much of the knowledge is either
Theoretical and as such, essentially orientedtoward fundamental problems
Operational and essentially empirical Biofilm / Biofilm reactors
This not completely satisfying
Observation and data collection for modeling a biofiltration process
General outlineExperimental based model development
5Experimental approach overview
Two different approaches: Treatment efficiency figures
i) What is the maximum treatment efficiency that I canreach?
ii) What are the required operational conditions for thismaximum?
iii) Is it an economic optimum? That the engineering way It leads to operational decision but not a full
description of the process itself thats a black box
6Experimental approach overview
Then, the question we are adressing: How to understand a process with very limited
internal observation possibilities? How to access a precise estimation of key
internal fluxes (growth, hydrolysis, ammonification, etc.)?
How to reduce uncertainty Then basically two approaches:
Steady state mass balancing and routing Dynamic temporal and spatial approach
7Context
Integrating knowledge about already known fundamental processes acquire robust predictive capacities identify limiting processes to focus innovation efforts on these
aspects
Biostyr (upflow biological aeratedfilters) systems: have been chosen as a model
process should serve as a first step for
methods development (to beapplied to other biofilm process)
Observation and data collection for modeling a biofiltration process
Mass Balance over a Biostyr
Back to basics
9Mass balance - Concepts
Based on simple and robust hypothesis Steady state Conserved quantities (water flows, CHONP, charges,
COD) Composition data (biomass, substrate, )
mass balancing leads to different and important possibilities: Data reconciliation Internal flux estimation
See e.g. Meijer et al 2002,Puig et al 2008 for more discussion on data reconciliationSee Ekama 2009, Iwa GMP 2008, and Lavoisier 1789 for mass balance
10
Mass balance - Estimation of the internal fluxes
Compartiments Flux Bilan
In and Out Conversions
Influent Effluent Backwash In Gas Out Gas H Aer Gr H Ano Gr A Aer Gr
Q g/j Vin - Vout - Vlav 0,0 P gP/j TPin - TPout - TPlav 0,0
TN gN/j TNin - TNout - TNlav (non mesur) (non mesur) - N2 produit 0,0
NOx- gN/j NOx- in - NOx- out - NOx- lav - NOx- consomm + NOx- produit 0,0 dO2 gO/j - O2 out O2 in - O2 cons - O2 cons - O2 cons 0,0
DCO gDCO/j DCOin - DCOout - DCOlav - DCOcons - DCOcons + DCOprod 0,0 TC/CO2 gC/j TCin - TCout - TClav CO2 in - CO2 out - CO2 produit - CO2 produit + CO2 consomm 0,0
Estimated from stochiometry
Measured values,Deduced from mass balance,
11
Mass balance - Concepts
Can we go deeper? Based on elemental cycles (COD, N, TC) Based on basic fractionation Based on ASM conceptual models
Other transformations process are accessible Hydrolysis (and decay) Ammonification
12
Mass Balance Undertsanding the cycles
-1
40-9-15
-32
140-37
Oxydation Nitrification
Dnitrification
-40
-12
TN-130
-50
200
Hydrolysis
Nitrogen Flux (gN/cycle)
NH4+
NOx-4
20
40 1
37
-4
4
-12
-80-15
-8
Ammonification
44
-44
-39
39
OK
OK
OK
OK
Norgsoluble
Norgparticulaire
-28
13
Mass Balance - Conclusion
Why is this approach that important on biofiltration process: Global process intensities over a filtration cycle
Interest for a modelling a biofilm reactor
Trial and error is not an option!
So many unknowns that one should at least calculate all the quantities that he is able to evaluate from a simple mass balance:
There are still many things to play with afterward: Initial state Kinetics and laws Biofilm specific process (thickness, boundary layers, filtration, etc.)
14
Mass Balance - Conclusion
interesting (I hope) and light approach on pilot scale calibrated flowmeters common lab analysis (possible with minikits only ) gaz analyser for a few days
is only global (in time and space) assumes homogeneity over the process aggregates the filtration cycle as a steady state process
and forget about variations in time Then what should be done :
Evaluate other functionning points Loop with the filter profiles approach Integrate this results in a biofilter model
Fate of suspended solids
Profiles and kineticsA bit deeper
16
Profils methods
Liquid sampling Media sampling
17
Bulk concentration profiles
CODt
0
100
200
300
400
500
ED P1 P2 P3 P4 P5 P6 EF
Sampling point
C
O
D
t
(
m
g
/
L
)
DCOt -12hDCOt -15h
MES
020406080
100120140
ED P1 P2 P3 P4 P5 P6 EF
Sampling point
M
E
S
(
m
g
/
L
)
MES -12hMES -15h
CODs
0
50
100
150
200
ED P1 P2 P3 P4 P5 P6 EF
Sampling point
C
O
D
s
(
m
g
/
L
)
DCOs -12hDCOs -15h
CODp
0
50
100
150
200
250
ED P1 P2 P3 P4 P5 P6 EF
Sampling point
C
O
D
p
(
m
g
/
L
)
DCOp -12hDCOp -15h
d_DCOt
-15,00
-10,00
-5,00
0,00
5,00
P1->P2
P2->P3
P3->P4
P4->P5
P5->P6
Sampling point
C
O
D
t
(
k
g
/
m
^
3
/
j
)
d_DCOt - 12h
d_DCOt - 15h
Bilan demasse
d_DCOs
-8,00
-6,00
-4,00
-2,00
0,00
2,00
P1->P2
P2->P3
P3->P4
P4->P5
P5->P6
Sampling point
r
(
k
g
/
m
^
3
/
j
)
d_DCOs -12hd_DCOs -15hBilan demasse
d_DCOp
-12,00
-8,00
-4,00
0,00
4,00
P1->P2
P2->P3
P3->P4
P4->P5
P5->P6
Sampling point
r
(
k
g
/
m
^
3
/
j
)
d_DCOp -12hd_DCOp -15hBilan demasse
d_MES
-6,00
-4,00
-2,00
0,00
2,00
P1->P2
P2->P3
P3->P4
P4->P5
P5->P6
Sampling point
r
(
k
g
/
m
^
3
/
j
)
d_MES - 12h
d_MES - 15h
Bilan demasse
Concentrations profiles Reaction rate profiles
18
Discussion
This is nice and obvisously needed: Estimation of the reaction rates Estimation of the filtration behavior
but This can lead to very long lab days This need special equipement
Pilot plant is the correct scale This leads to very very long lab days
19
Limits
After some tests with a model, it appearsthat: Air transfer is one of the most important
operational parameter as it influence everythingand what is published is clearly not enought
Results are also very sensitive to the physicalfiltration model that we choose
And that is where we should focus next
Any questions ?
Thanks for you attentionand have a nice day