Development of high-rate activated sludge processes for energy

Faculteit Bio-ingenieurswetenschappen
Academiejaar 2013-2014
Development of high-rate activated sludge processes for
energy-efficient wastewater treatment
Tim Van Winckel
Promotor: Prof. Dr. ir. Siegfried Vlaeminck
Tutor: Msc. Francis Meerburg
Masterproef voorgedragen tot het behalen van de graad van
Master in de bio-ingenieurswetenschappen: Milieutechnologie
De auteur en promotor geven de toelating deze scriptie voor consultatie beschikbaar te stellen en
delen ervan te kopiëren voor persoonlijk gebruik. Elk ander gebruik valt onder de beperkingen
van het auteursrecht, in het bijzonder met betrekking tot de verplichting uitdrukkelijk de bron
te vermelden bij het aanhalen van resultaten uit deze scriptie.
The author and promoter give the permission to use this thesis for consultation and to copy
parts of it for personal use. Every other use is subject to the copyright laws, more specifically
the source must be extensively specified when using from this thesis.
Ghent, June 2014
De promotor
Prof. dr. ir. Siegfried Vlaeminck
De tutor
Msc. Francis Meerburg
De auteur
Tim Van Winckel
iv
Acknowledgments
Het was een rollercoaster.
Toen ik een jaar geleden startte met mijn thesis vroeg ik mezelf af hoe ik in hemelsnaam een
thesis zou kunnen schrijven. Deze tekst bewijst echter dat het mij toch gelukt is.
Zonder de hulp van Francis, die mij dagdagelijks met raad en daad begeleide, altijd klaar stond
voor vragen en hulp en me leerde data op een kritische manier te bekijken en interpreteren,
had u wellicht dit tekstje nooit gelezen.
Alsook niet zonder de hulp van professor Vlaeminck, die eeuwig enthousiast altijd klaar stond
met nieuwe of vernieuwende ideeën en me mijn liefde voor wetenschap en waterzuivering heeft
doen ontdekken.
Ik zou ook graag Dries bedanken, die er elke woensdagavond opnieuw stond met vers A-slib
indien mijn reactor weeral was uitgespoeld.
BioMATH, voor het gebruik van hun respirometrische opstelling, dewelke mij vele nieuwe
inzichten in het HiCS-proces heeft bezorgd.
Professor Van Haandel voor het gebruik van zijn respirometrische opstelling en uiteraard heel
labMET voor de toffe, maar wetenschappelijk ernstige sfeer waar ik de beste wetenschapper in
mezelf naar boven kon halen.
Ik mag zeker ook niet mijn kotgenoten Wim en Kim, alsook mijn vrienden Kristof en Stijn
die allen ook hun thesis bij labMET doen vergeten. De vele ludieke, maar vaak toch zeker
wetenschappelijke gegronde gesprekken zal ik niet gauw vergeten!
Bedankt iedereen!
Tim
v
vi
List of Abbreviations
A/B-process
AD
anammox
AOB
Bx
BMP
BOD
CAS
COD
CS
DO
DWF
EPS
FID
GC
HiCS
HRAS
HRT
IS
kL a
MLSS
NOB
OLAND
OUR
PHA
adsorption-belebungsverfahren process
anaerobic digestion
anaerobic ammonium oxidation
ammonia oxidizing bacteria
specific loading rate
biochemical methane potential
Biological Oxygen demand
conventional activated sludge
chemical oxygen demand
contact stabilization
dissolved oxygen
dry weather flow
extracellular poly- meric substances
flame ionization detector
gas chromatograph
High-load contact stabilization
High-rate activated sludge
Hydraulic retention time
internal standard
volumetric gas transfer coefficient
mixed liquor suspended solids
nitrite oxidizing bacteria
oxygen-limited autotrophic nitrification/denitrification
oxygen uptake rate
polyhydroxyalkanoates
vii
viii
PSD
rbCOD
sbCOD
SBR
SCFA
SGP
SI
SOUR
SRT
SS
SVI
τ
TAN
TKN
TS
TSS
VER
VS
VSS
WWF
WWTP
particle size distribution
readily biodegradable COD
slowly biodegradable COD
sequencing batch reactor
short chain fatty acids
specific gas production
solubility index
specific oxygen uptake rate
sludge retention time
suspended solids
sludge volume index
contact time over stabilization time
total ammonia nitrogen
total kjeldal nitrogen
total solids
total suspended solids
volume exchange ratio
volatile solids
volatile suspended solids
wet weather flow
wastewater treatment plant
List of Symbols
Bv
Bx
fcv
fH
fx,aer
kd
OU
Qe
Qw
S
Si
SRT
τ
Vr
Vs
Vsa
Xe
Xr
Xs
Xw
Y
Yobs
Yresp,diss
volumetric loading rate (gCOD L-1 d-1 )
specific loading rate (gCOD gVSS-1 d-1 )
VSS/COD ratio (gVSS gCOD-1
endogenous residue fraction (-)
fraction of sludge aerated (-)
endogenous decay date (d-1 )
Oxygen uptake (mg O2 )
effluent flow rate (L d-1 )
waste flow rate (L d-1 )
Soluble COD removal (mg COD)
influent substrate concentration (gCOD L-1 )
sludge retention time (d)
contact time over stabilization time (-)
reactor volume (L)
Settler volume (L)
sample volume (L)
effluent biomass concentration (gVSS L-1 )
reactor biomass concentration (gVSS L-1 )
settler biomass concentration (gVSS L-1 )
waste biomass concentration (gVSS L-1 )
maximum yield (gVSS gCOD-1 )
observed yield (gVSS gCOD-1 )
respirometrical observed yield (COD COD-1 )
ix
x
Summary
This year, the conventional activated sludge (CAS) process celebrates its 100th birthday. Since
its conception, this process has not been fundamentally changed as CAS is able to purify
domestic wastewater with good removal efficiencies in terms of biodegradable carbonaceous
organics and nutrients. As we enter a new era where the scope of sustainability is much broader
than mere pollution mitigation, the overall sustainability of the CAS process is questioned.
Whereas conventional activated sludge scores low on energy efficiency and as a consequence,
cost effectiveness, high-rate activated sludge (HRAS) processes are proposed to tackle these
issues. As opposed to CAS, the low sludge retention times (SRT) used in HRAS imply that
much of the energy present in the wastewater in the form of COD can be recovered as sludge for
biogas production. Additionally, aeration energy imposes only a minor operational cost since
no nitrification occurs and next to oxidation, substrate removal through sorption and storage is
maximized. HRAS systems are expected to perform best as part of a two-stage activated sludge
system, in which biodegradable carbonaceous organics are removed first by a HRAS stage to
maximize COD recovery in the form of sludge while nitrogen is removed in a second stage,
operated with the very-energy efficient oxygen-limited nitrification/denitrification (OLAND)
process. In this thesis, two different HRAS designs were investigated: the A-stage or unmodified
HRAS process, and the high-load contact stabilization (HiCS) process, a newly developed highrate variant of the contact stabilization (CS) process. Little fundamental research has been
conducted on operational parameters of the A-stage, necessitating research on how to optimize
its energy efficiency. In addition, the HiCS process might have an increased sludge production
compared to the A-stage since its design entails a selection for micro-organisms that efficiently
employ sorption of particulates as a substrate removal technique. Two A-stage configurations
were tested at lab-scale at SRTnom of 0.5 and 1.5 days. Two HiCS reactors were operated at a
different ratio of contact time over the stabilization time (τ ); τ = 1 and τ = 71 . The reactors
were operated in sequencing batch reactor (SBR) mode and a complex synthetic wastewater was
used as influent. COD removal efficiency, yield and recoverable COD were used as performance
indicators, and lab-scale CAS and CS reactors were operated as reference controls. Biochemical
methane potential (BMP) and fermentation tests were conducted to evaluate two options to
valorize the produced sludge. Respirometry was used to characterize the synthetic wastewater
xi
xii
Summary
and the stabilization and contact phase of the HiCS reactors. The CAS reactor operated at an
actual sludge retention time SRTact = 10.15˘4.96 days and achieved the highest COD removal
efficiency (74˘12%). The A-stage operating at SRTact of 1.39˘0.83 days attained the highest
COD removal efficiency (51˘7%) of all high-rate reactors. Elevated biomass concentrations
in combination with a lower specific loading rate were most likely the dominant factors for
this high COD removal efficiency. The HiCS reactors attained poor COD removal efficiencies
(33˘15% and 30˘17% for τ =1 and τ = 71 respectively). The COD removal efficiency was
not influenced by τ indicating that other operational parameters are most likely influencing
the removal efficiency. Both A-stage reactors did not attain a higher maximum yield than the
CAS reactor, indicating that storage and adsorption was absent. The SRT does not influence
the maximum yield. The HiCS reactor operating at τ = 17 attained the higher maximum yield
of all reactors (0.79˘0.04% gVSS gCOD-1 ), suggesting that HiCS attains higher yields than an
A-stage reactor. It also achieved the highest fraction of COD that was recoverable.The HiCS
reactor operating at τ = 1 obtained a yield comparable to the CAS reactor and was lower than
the CS reactor, demonstrating that τ is a crucial parameter in the design of a HiCS reactor as
it profoundly influences the yield. Respirometric experiments during the HiCS operating cycle
indicated that the stabilization phase of the reactor operating at τ = 1 showed an increased
respirometric yield, suggesting that storage is present as a COD removal mechanism. The
yield over the complete cycle was evenly distributed over the contact and stabilization phase.
The respirometric yield of the reactor operating at τ = 71 showed a much larger discrepancy
between the stabilization and contact phase, possibly due to storage which might explain the
elevated reactor maximum yield. Anaerobic digestion of the different sludges showed that the
COD content of the A-sludge was 100% convertible into CH4 whereas the HiCS-sludge was
only 42% convertible. However, the total specific methane production was nearly the same for
every type of sludge, as the the COD/VS ratio of the HiCS-sludge was higher. It is not clear
why the HiCS-sludge did not covert well; Differences in the anaerobic sludge inoculum used
may be the cause of this discrepancy. Fermentation of the HiCS sludge produced 0.64˘0.13
gCODSCFA gVSsubstrate -1 , which is higher than typical values for specific production reported
in literature. The very young age of the sludge might be the cause of this elevated production,
and the same trend is observed for anaerobic digestion and fermentation of young sludge. To
conclude, there is no significant difference in sludge yield between the two A-stages operated
at different SRTs, but the A-stage operated at a higher SRT achieved higher COD removal
efficiencies. An optimized HiCS reactor is a viable if not better alternative to the A-stage in
terms of sludge production and COD recovery. The parameter τ was shown to be critical in
determining the performance of HiCS and its optimization should be a priority when designing
a HiCS reactor. Furthermore, other parameters such as HRT and SRT need to be further
optimized to achieve improvements in COD recovery efficiency.
Samenvatting
Het conventioneel actief slib (CAS) systeem viert dit jaar zijn 100ste verjaardag. Sinds de uitvinding is het proces niet fundamenteel veranderd omdat CAS in staat is huishoudelijk afvalwater
te zuiveren met goede efficiëntie in termen van biologisch afbreekbare organische koolstof en
nutrinten verwijdering. De term duurzaamheid wordt momenteel echter veel breder gezien dan
enkel het beperken van vervuiling en hierdoor kunnen we de volledige duurzaamheid van het
CAS proces in vraag stellen. Conventioneel actief slib scoort laag inzake energie efficintie en zijn
als gevolg daarvan ook niet kosteneffectief. Hoogbelaste actief slib (HRAS) systemen worden
voorgesteld om dit probleem aan te pakken. In tegenstelling tot CAS, werken HRAS systemen
met een zeer lage slib retentietijd (SRT), wat betekent dat veel van de energie aanwezig in het
afvalwater in de vorm van chemische zuurstof vraag (CZV) kan worden teruggewonnen in de
vorm van slib dat kan gebruikt worden voor biogas productie. Bovendien is de kost gelinkt met
hier gering omdat geen nitrificatie plaatsvind en naast oxidatie ook substraat wordt verwijderd
door sorptie en opslag. HRAS systemen opereren het beste in een twee-staps actief slib systeem
waar eerst de CZV wordt verwijderd door een HRAS stap om CZV recuperatie te maximaliseren
en vervolgens stikstof wordt verwijderd in een tweede stap met het oxygen-limited autotrophic
nitrification/detrification (OLAND) proces. In deze thesis worden twee verschillende HRAS
configuraties op de proef gesteld: De A-stap of een niet-gemodificeerd HRAS systeem en een
nieuw ontwikkeld hoog belast contact stabilisatie (HiCS) systeem, een hoog belaste variant
van het contact stabilisatie (CS) proces. Weinig fundamenteel onderzoek is reeds verricht op
de operationele parameters van de A-stap, wat noodzakelijk is om deze te optimaliseren. Het
HiCS proces kan een hogere slibopbrengst vertonen in vergelijking met de A-stap omdat er
een selectie van micro-organismen kan optreden dewelke efficint sorptie en opslag exploiteren.
Twee A-stap reactoren werden getest op laboratoriumschaal met een SRTnom van 0.5 en 1.5
dagen. Twee HiCS reactoren met een verschillende contact tijd over stabilisatie tijd (τ ); τ = 1
en τ = 71 . De reactoren werden geopereerd in een sequencing batch reactor (SBR) en complex
synthetisch afvalwater werd gebruikt als influent. CZV-verwijderingsefficiëntie, slibopbrengst
en recupereerbare CVZ werden gebruikt als prestatie indicatoren en laboratoriumschaal CAS
en CS reactoren werden gebruikt als controle. Een biochemische methaan potentieel (BMP) en
fermentatietest werden uitgevoerd om de twee manier van slibvalorisatie onderling te evalueren.
xiii
xiv
Samenvatting
Respirometrie werd gebruikt om het synthetisch influent en de contact- en stabilisatiefase van
het HiCS proces te karakteriseren. De CAS reactor dewelke opereerde bij een actuele SRTact
van 10.15˘4.96 dagen behaalde de hoogste CZV verwijderingsefficiëntie (74˘12%). De A-stap
met een SRT van 1.39˘0.83 dagen behaalde de hoogste CZV verwijderingsefficiëntie (51˘7%)
van alle HRAS reactoren. De HICS reactoren behaalden relatief slechte CZV verwijderingsefficiënties (33˘15% en 30˘17% voor respectievelijk τ =1 en τ = 17 ). De CZV verwijderingsefficiëntie werd niet beı̈nvloed door τ , wat er op wijst dat andere parameters verantwoordelijk
zijn voor de verwijderingsefficiëntie. Beide A-stap reactoren hadden geen hogere maximale slibopbrengst dan de CAS controle reactor, wat er op wijst dat sorptie of opslag afwezig was. De
SRT lijkt de maximale slibopbrengst ook niet te beı̈nvloeden. De HiCS reactor met τ = 17 behaalde de hoogste slibopbrengst van alle reactoren (0.79˘0.04% gDS gCVZ-1 ), wat suggereert
dat HiCS systemen een hogere slibopbrengst kunnen behalen dan A-stap reactoren. Deze reactor had eveneens de grootste CZV recuperatiepotentieel. De HiCS reactor dewelke opereerde
op τ = 1 behaalde een vergelijkbare maximale slibopbrengst als de CAS reactor en zelfs lager
dan de CS reactor. Dit toont aan dat τ een zeer grote invloed heeft op de slibopbrengst en
dus zeer belangrijk is in het ontwikkelen van dergelijk systeem. Respirometrisch experimenten
van de HiCS cyclus toonde aan dat de respirometrische slibopbrengst van de reactor die werkte
op τ = 1 een verhoogde respirometrische slibopbrengst vertoonde in de stabilisatieperiode,
wat suggereert dat opslag aanwezig was als CZV verwijderingstechniek. De opbrengst berekend over de totale cycles verschilde echter niet veel van deze over de stabilisatieperiode. De
respirometrische slibopbrengst tussen de stabilisatie- en contactfase vertoonde en veel groter
verschil, wat de verhoogde slibopbrengst in de reactor kan verklaren. Anaerobe vergisting van
de verschillende types slib toonde aan dat het A-slib 100% vergistbaar tot CH4 . Het HiCS
slib was echter maar 42% vergistbaar. De reden hiervoor is niet duidelijk; verschillen in het
gebruikt anaeroob slib zou misschien dit groot verschil kunnen verklaren. Fermentatie van het
HiCS slib produceerde 0.64˘0.13 gCZVVVZ gVSsubstraat -1 , welke hoger is dan tot nu toe in de
literatuur gerapporteerd staat. De zeer jonge aard van het slib kan een potentiële reden zijn
omdat de zelfde trend wordt geobserveerd bij anaerobe digestie van jong slib. Er kan besloten
worden dat er geen significant verschil is in slibopbrengst voor de twee A-stap reactoren bij
verschillende SRT, maar A-stap die werkte op een hogere SRT behaalde wel een hogere CVZ
verwijderingsefficiëntie. Een geoptimaliseerde HiCS reactor is een vatbaar, al dan niet beter alternatief voor de A-stap in termen van slibopbrengst en CVZ herwinning. τ bleek van cruciaal
belang te zijn bij het ontwerpen van een HiCS reactor. Bovendien moeten andere parameters, zoals de HRT en SRT verder geoptimaliseerd worden om vooruitgang te boeken in CZV
verwijderingsefficiëntie.
Contents
Summary
xi
Samenvatting
xiii
List of Figures
xix
List of Tables
I
xxiii
literature review
1
1 Water - an abundant yet scarce resource
3
2 The activated sludge process
5
2.1
The basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5
2.2
State-of-the-art . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5
2.3
Sustainability issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6
2.4
The future of wastewater treatment
7
. . . . . . . . . . . . . . . . . . . . . . . .
3 High-rate activated sludge
11
3.1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11
3.2
Biological Sorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11
3.3
A/B-process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14
3.4
Contact Stabilization Process . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15
3.5
Energy & Operational costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17
4 Operational Parameters for activated sludge processes
19
4.1
Sludge retention time (SRT) . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19
4.2
Loading rate (Bx ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
22
xv
xvi
Contents
4.3
Dissolved Oxygen level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
22
4.4
Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
23
5 Research Question
25
II
27
Materials, Methods and Reactors
6 Reactors
6.1
6.2
6.3
6.4
29
General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
29
6.1.1
Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
30
6.1.2
Influent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
30
6.1.3
Inoculation & Start-up . . . . . . . . . . . . . . . . . . . . . . . . . . . .
32
6.1.4
SRT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
32
6.1.5
Sampling, follow-up and analysis . . . . . . . . . . . . . . . . . . . . . .
32
A-stage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
32
6.2.1
Operational conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . .
32
High-load contact-stabilization . . . . . . . . . . . . . . . . . . . . . . . . . . .
34
6.3.1
Operational conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . .
34
Low-load reference reactors . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
36
6.4.1
36
Operational conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7 Batch tests
7.1
39
Respirometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
39
7.1.1
Contact and stabilization phase characterization . . . . . . . . . . . . .
39
7.1.2
Influent characterization . . . . . . . . . . . . . . . . . . . . . . . . . . .
40
7.2
Biochemical methane potential test . . . . . . . . . . . . . . . . . . . . . . . . .
41
7.3
Fermentation test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
41
7.3.1
inoculum preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . .
42
7.3.2
buffer preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
42
7.3.3
HiCS Sludge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
42
8 Chemical analysis
43
8.1
Chemical Oxygen Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
43
8.2
Biochemical Oxygen Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . .
43
8.3
Total suspended solids, total solids, volatilizable suspended solids, volatilizable
solids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
43
Nitrogen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
44
8.4
Contents
8.5
8.6
8.7
8.8
8.4.1 Total Ammonia Nitrogen . .
8.4.2 Total Kjeldal Nitrogen . . . .
Short Chain Fatty Acids . . . . . . .
Biogas composition . . . . . . . . . .
Particle size distribution . . . . . . .
pH, DO and Pressure measurements
9 Methods
9.1 Yield calcuation . . . .
9.1.1 Observed yield
9.1.2 Maximum yield
9.2 COD balance . . . . .
III
xvii
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
44
44
44
45
45
45
.
.
.
.
47
47
47
47
48
Results
10 Continuous reactors: Performance indicators
10.1 Operational conditions . . . . . . . . . . . . . . . . . .
10.1.1 Reference reactors . . . . . . . . . . . . . . . .
10.1.2 A-stage reactors . . . . . . . . . . . . . . . . .
10.1.3 HiCS reactors . . . . . . . . . . . . . . . . . . .
10.1.4 Summarizing table . . . . . . . . . . . . . . . .
10.2 Biomass concentration, SRT and specific loading rate .
10.2.1 Reference Reactor . . . . . . . . . . . . . . . .
10.2.2 A-stage reactors . . . . . . . . . . . . . . . . .
10.2.3 HiCS reactors . . . . . . . . . . . . . . . . . . .
10.3 COD removal efficiency . . . . . . . . . . . . . . . . .
10.3.1 Reference reactors . . . . . . . . . . . . . . . .
10.3.2 A-stage reactors . . . . . . . . . . . . . . . . .
10.3.3 HiCS reactors . . . . . . . . . . . . . . . . . . .
10.4 Particle size distribution . . . . . . . . . . . . . . . . .
10.5 Yield . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10.6 COD Balance . . . . . . . . . . . . . . . . . . . . . . .
49
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
51
51
51
52
53
54
56
56
57
58
59
59
60
61
62
63
64
11 Batch tests
11.1 Respirometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.1.1 Influent characterization . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.1.2 HiCS cycle characterization . . . . . . . . . . . . . . . . . . . . . . . . .
67
67
67
68
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
xviii
Contents
11.2 Biochemical Methane Potential . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.3 Fermentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
IV
Discussion
70
73
75
12 Introduction
77
13 Reactors
13.1 COD removal efficiency .
13.1.1 Reference reactors
13.1.2 A-stage & HiCS .
13.1.3 Conclusion . . . .
13.2 Yield & Recoverable COD
13.2.1 Yield . . . . . . . .
13.2.2 Recoverable COD
13.2.3 Conclusion . . . .
.
.
.
.
.
.
.
.
79
79
79
80
81
82
82
86
87
14 Respirometry
14.1 Influent characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14.2 Contact and stabilization phase characterization . . . . . . . . . . . . . . . . .
89
89
90
15 Biochemical Methane
15.1 BMP . . . . . . . .
15.2 Fermentation . . .
15.3 Economical aspects
93
93
94
95
.
.
.
.
.
.
.
.
.
. .
. .
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Potential &
. . . . . . . .
. . . . . . . .
. . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Fermentation
. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . .
16 General Conclusion
97
17 Future Research
99
Bibliography
101
List of Figures
2.1
Scheme of a typical state-of-the-art CAS treatment plan. Other configurations
are possible. (Verstraete and Vlaeminck, 2011). . . . . . . . . . . . . . . . . . .
6
2.2
scheme of a ZeroWasteWater treatment plant. (Verstraete and Vlaeminck, 2011).
8
3.1
The kinetic theory of BOD removal according to Bunch and Griffin (1987). In
the first minutes, the BOD concentration in the wastewater drops dramatically
due to rapid adsorption of the BOD onto the biomass. The BOD concentration
then peaks again as it becomes hydrolyzed and detaches. The soluble BOD is
thereafter mineralized to CO2 , which permanently decreases the BOD in the
wastewater. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12
3.2
Simplified scheme of the A/B-process. Based on (Salomé, 1990). . . . . . . . .
14
3.3
Simplified graphical representation of the contact stabilization process. Image
based on (Metcalf and Eddy, 2003). . . . . . . . . . . . . . . . . . . . . . . . .
16
Fraction of the COD that will remain in the effluent (mSte ; left), be oxidized to
CO2 (mSo ; center) and trapped in the excess sludge (mSxv ; right) in function
of the COD. fns is the non biodegradable dissolved influent COD fraction. The
dotted lines show the influence of the non-biodegradable dissolved influent COD
fraction (fnp ). Adapted from (Van Haandel and Van Der Lubbe, 2007) . . . .
21
4.1
6.1
Two lab scale SBR reactors operated to simulate the A-stage and HiCS reactors. 29
6.2
Scheme of the SBR reactor used for all reactors. A = influent pump; B = effluent
pump; C = sludge waste pump; D = pH controller; E = level controller. . . .
30
6.3
Graphical representation of the SBR-cycles used to mimic an A-stage reactor .
34
6.4
Graphical representation of the SBR-cycles used to mimic HiCS reactors
. . .
35
6.5
SBR cycles of the two lab-scale SBR reactors operated to simulate two low-rate
reference reactors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
37
xix
xx
List of Figures
7.1
Schematic representation of the respirometric batch experiment. A = aeration;
B = DO probe connected to the microcontroller; C = Stirrer . . . . . . . . . .
40
10.1 The reactor VSS (upper), SRT (middle) and Specific loading rate (bottom) of
the reference reactors CAS (‚) and CS () in function of the total runtime of
the experiment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
56
10.2 The reactor VSS (upper), SRT (middle) and Specific loading rate (bottom) of
the A-stage reactors A-1 (‚) and A-2 () in function of the total . . . . . . . .
57
10.3 The reactor VSS (upper), SRT (middle) and Specific loading rate (bottom) of
the A-stage reactors A-1 (‚) and A-2 () in function of the total . . . . . . . .
58
10.4 CODtotal (‚) and CODdissolved (ˆ) removal efficiencies observed for the lab-scale
CAS and CS reactor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
59
10.5 CODtotal (‚) and CODdissolved (ˆ) removal efficiencies observed for the A-stage
reactors A-1 (left) and A-2 (right). . . . . . . . . . . . . . . . . . . . . . . . .
60
10.6 CODtotal (‚) and CODdissolved (ˆ) removal efficiencies observed for the HiCS
reactors HiCS-1 (left) and HiCS-2 (right). . . . . . . . . . . . . . . . . . . . .
61
10.7 Particle size distribution of influent and effluent of both HiCS reactors.
. . . .
62
10.8 Cumulative biomass growth in function of the cumulative substrate removal for
all reactors operated in this study. The observed yield is calculated by a lineair
fit trough the datapoints. A-1 (‚); A-2 (‚); HiCS-1 (˚); HiCS-2 (`) CAS (İ);
CS(IJ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
63
10.9 Percentage of incoming COD lost with the effluent, both particulate (effluent
part) and soluble (effluent diss), sampling and that can be harvested trough
waste sludge (waste). The balance is closed with the fraction of COD that
was not retrieved by any of the prior mechanisms (presumably respiration; see
discussion). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
65
11.1 exogenous OUR- and cumulative OU profile of freshly made influent. The bottom graph represents the first derivative of the OU. (1) rbCOD (2) sbCOD (3)
endogenous respiration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
68
11.2 Respirogram and COD profile of one full SBR cycle of reactor HiCS-1.
. . . .
69
11.3 Respirogram and COD profile of one full SBR cycle of reactor HiCS-2.
. . . .
70
. . . . . . . . . .
71
. . . . . . . . . . . . . . . .
72
. . . . . . . . . . . . . . .
73
11.7 percentage of substrate that is conversion into SCFA and CH4 . . . . . . . . . .
74
11.4 Specific biogas production of the A-stage and HiCS sludges.
11.5 Specific CH4 production of the two HiCS sludges.
11.6 Specific SCFA production of the two HiCS sludges.
List of Figures
xxi
13.1 Calculated daily observed yield of all lab-scale reactors that were operated. The
variation observed in the daily observed yield is much smaller than the ones
attained in the high-rate reactors. CAS (IJ); CS (); A-1 (‚), A-2 (˛); HiCS-1
(İ) and HiCS-2 (ˆ). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
84
14.1 exogenous OUR- and cumulative OU profile of freshly made influent. (I) OUrbCOD
(II) OUstorage tail (IIb & III) OUsbCOD . . . . . . . . . . . . . . . . . . . . . . .
90
xxii
List of Figures
List of Tables
2.1
Average removal efficiencies observed in Belgian wastewater treatment plants in
2008 (personal communication, Aquafin). SS = suspended solids . . . . . . . .
7
6.1
Composition of concentrated SYNTHES . . . . . . . . . . . . . . . . . . . . . .
31
6.2
Design and operational parameters of the lab-scale A-stage reactors . . . . . . .
33
6.3
Design and operational parameters of the two lab-scale HiCS reactors . . . . . .
36
6.4
Operational parameters of the two lab-scale reference reactors . . . . . . . . . .
38
7.1
Overview of BMP-test setup for both the A-stage and HiCS sludge. . . . . . . .
41
10.1 Statistical significance between the key parameters of CAS and CS reactor obtained with the Mann-Whitney-U non-parametric test in R. ‹ “ mildly significant (0.05 ą p ě 0.01) ; ‹‹ “ significant (0.01 ą p ě 0.001); ‹ ‹ ‹ “ strongly
significant (p ą 0.001). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
52
10.2 Statistical significance between the key parameters of the two A-stage reactors
obtained with the Mann-Whitney-U non-parametric test in R. ‹ “ mildly significant (0.05 ą p ě 0.01) ; ‹‹ “ significant (0.01 ą p ě 0.001); ‹ ‹ ‹ “ strongly
significant (p ą 0.001) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
53
10.3 Statistical significance between the key parameters of the two HiCS reactors
obtained with the Mann-Whitney-U non-parametric test in R.. ‹ “ mildly
significant (0.05 ą p ě 0.01) ; ‹‹ “ significant (0.01 ą p ě 0.001); ‹ ‹ ‹ “
strongly significant (p ą 0.001) . . . . . . . . . . . . . . . . . . . . . . . . . . .
54
10.4 Summarizing Table of the mean performance indicators and conditions of all the
reactors that were operated during this study. . . . . . . . . . . . . . . . . . .
55
10.5 attained cumulative VSS yield of all reactors and their corresponding maximum
yields. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
64
xxiii
xxiv
List of Tables
11.1 Respirometric observed yield based on soluble substrate of the stabilization,
contact phase of the HiCS reactors and the corresponding respirometric observed
yield of the cycle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11.2 Key ratio’s of the A- and HiCS-sludge. . . . . . . . . . . . . . . . . . . . . . .
11.3 Key parameters of BMP tests conducted on the A- and HiCS-sludge. . . . . .
70
71
72
15.1 Reactor details and corresponding biogas and fermentation revenue. . . . . . .
96
Part I
literature review
Chapter 1
Water - an abundant yet scarce
resource
Seventy percent of the earth is water, yet only 2.53% of the total water reserve on earth is
considered fresh water. Only 0.26% of the total fresh water reserve is stored in fresh water
lakes and rivers and is thus considered available for human consumption (Gleick, 1993). Due to
today’s increasing human population size and rapidly accelerating development, both urbanization and increasing standard of living, the global demand for fresh water is steadily growing
causing environmental stress on these fresh water reserves. This environmental stress factor
can be assessed with the concept ’water stress’, which is used as an indicator on the intensity
of pressure put on water resources by external drivers of change. Water stress on river basins
will increase with more than 50% by 2050 due to climate change and rapid socio-economical
development (Alcamo et al., 2007). Furthermore, more than one billion people will deal with
increased water stress by 2050 (Arnell, 2004).
In western countries, cities have to deal with increasing demands on energy and resources.
Currently, cities have a linear resource pathway, depending fully on the hinterland for energy,
heat and water supply. The increasing demand creates an incentive towards a circular pathway for energy, heat and water (Agudelo-Vera et al., 2012; Verstraete and Vlaeminck, 2011).
Wastewater is, as the name suggests, is still seen as waste stream. However, technology allows
us to extract and reuse (valuable) resources present in wastewater (Verstraete et al., 2009).
The water sector has therefore a big challenge ahead: tackling issues concerning water stress
while making water purification more sustainable and perhaps even economically profitable.
3
4
Water - an abundant yet scarce resource
Chapter 2
The activated sludge process
2.1
The basics
Celebrating its 100 year anniversary, the conventional activated sludge (CAS) process is the
standard in treating domestic and industrial wastewaters and the basics haven’t changed since.
Within this process, a diverse community of heterotrophic and autotrophic microorganisms,
known as the activated sludge, remove organic carbon (generally referred to as biological oxygen demand or BOD) and nutrients (N, P) present in the wastewater under aerated (and
anaerobic/anoxic) conditions.
The concept of CAS is remarkably simple; in its most basic configuration, wastewater is mixed
with the sludge in a continuously aerated basin. The bacteria utilize biodegradable organic
carbon and nutrients as substrate for maintenance, oxidizing the BOD present in the wastewater
to CO2 (mineralization), and for the creation of new biomass (reproduction). As the volume
of the basin is fixed, the influent flow rate creates a constant washout of the sludge into the
settler. Here, sludge and water is most commonly separated by gravitational sedimentation.
The purified water is discharged to surface waters and sludge is internally recycled back to the
reactor. To maintain a constant sludge concentration and counteract for growth, a part of the
sludge needs to be removed constantly (waste sludge) (Metcalf and Eddy, 2003).
2.2
State-of-the-art
The process above is an simplification of a state-of-the-art biological wastewater treatment
plant operating today. CAS sewage treatment plants today are complex biological processes
combining BOD removal with removal of nutrients through nitrification/denitrification and
5
6
The activated sludge process
enhanced biological phosphorous removal (Figure 2.1). After primary treatment with a primary settler to remove solids and some COD, the wastewater is treated in a basin which is
partly anaerobic, anoxic and aerobic for a complete degradation of nutrients and BOD. The
wastewater-sludge mixture is subsequently separated through a secondary settler and part of
sludge is recycled. A fraction of the effluent is also recycled to ensure proper denitrification
in the anoxic zone. The excess sludge from the primary and secondary settler is digested to
recover some of the energy present in the sludge as biogas (Verstraete and Vlaeminck, 2011).
Many variants on conventional activated sludge plants have been developed over the years. A
non-exhaustive list includes contact stabilization (CS), step-aeration, extended aeration and
sequencing batch reactors (SBR) (PaDEP).
Figure 2.1: Scheme of a typical state-of-the-art CAS treatment plan. Other configurations are possible.
(Verstraete and Vlaeminck, 2011).
2.3
Sustainability issues
Despite the fact that a CAS process plant is very efficient in removing organic carbon and
nutrients (Table 2.1), the overall sustainability of the process is increasingly being questioned.
Diffuse emissions of nitrous oxide (N2 O), a byproduct of nitrification and denitrification, a
potent greenhouse gas, substantially reduces ecological sustainability and carbon footprint of
the process (Colliver and Stephenson, 2000).
The main concern however is the low energy-efficiency and therefore cost-effectiveness of the
overall process. High aeration and pumping costs make that CAS treatment plants is not the
most cost-effective way to treat domestic sewage in many occasions (Tsagarakis et al., 2003).
A large-scale screening of CAS plants in Nordrhein-Westfalen, Germany, indicates that CAS
The future of wastewater treatment
7
Table 2.1: Average removal efficiencies observed in Belgian wastewater treatment plants in 2008 (personal communication, Aquafin). SS = suspended solids
BOD
COD
SS
N
P
96.6% 88.0% 94.8% 77.1% 84.4%
plants scores very low on total energy-efficiency. The plants have, on average, a total electricity
consumption of 33 kWh IE-1 year-1 , yet only 20% of the energy is recovered through anaerobic
digestion (Müller and Kobel, 2004).
Surprisingly, the total potential energy available in raw wastewater significantly exceeds the
amount of energy the treatment process demands. In 2000, the influent of the North Toronto
Treatment Plant in the United States contained approximately 9.3 times more energy than is
necessary to run the plant (Shizas and Bagley, 2004). The possibility to create more energy
efficient or even self-sufficient wastewater treatment plant (WWTP) becomes therefore feasible
in theory. Most of the usable caloric energy present in wastewater is incorporated in BOD, of
which a considerable fraction gets embodied within the sludge. The sludge can subsequently be
digested to harvest the energy (Wett et al., 2007). However, this practice is still relatively rare
in Europe, with exception of Sweden (Lantz et al., 2007). Moreover, the high sludge age used
at CAS treatment plants for nitrification vastly reduces the efficiency of the sludge digestion
(Bolzonella et al., 2005).
2.4
The future of wastewater treatment
To address these problems, a radical new WWTP design concept has been developed by Verstraete and Vlaeminck (2011). The ZeroWasteWater concept short-cycles wastewater and create energy positive wastewater treatment plants with a very low carbon footprint. In case
of a centralized sewer system, the ZeroWasteWater concept suggests short cycling water by
purifying it in a multistep approach (Figure 2.2).
The incoming wastewater is treated in an advanced pre-concentration step such as a high-rate
activated sludge process (HRAS) or chemical precipitation, which efficiently removes BOD
and maximizes sludge production. Subsequently, the concentrated sludge gets anaerobically
digested with an up-flow anaerobic sludge blanket (UASB) reactor to harvest biogas. This biogas can be converted in to electrical power with a combined heat and power (CHP) installation.
A net amount of 40 kWh IE-1 year-1 of electricity can be produced through anaerobic digestion.
This is four times higher than what an advanced CAS plant equipped with anaerobic digestion
can produce. The effluent of both pre-concentrator step and UASB reactor, both low in BOD,
8
The activated sludge process
gets subsequently treated to remove nitrogen. This is performed with the oxygen-limited autotrophic nitrification/denitrification (OLAND) process, as it has low energy consumption and
sludge production. The treated wastewater is thereafter purified to become potable through a
tertiary treatment using ultrafiltration and reverse osmosis.
Figure 2.2: scheme of a ZeroWasteWater treatment plant. (Verstraete and Vlaeminck, 2011).
High BOD loads are essential for the feasibility of a self-sufficient WWTP as present in the
ZeroWasteWater concept. Therefore, degradation of BOD before it arrives at the plant should
be avoided. Degradation of BOD in sewer systems can be avoided by changing the sewer
system from a slow, gravitational system to a fast vacuum or pressure system. However,
Western countries have an extensive sewer system with little room for immediate and drastic
changes. Concentrating the wastewater as it arrives at the ZeroWasteWater plant is a more
realistic measure that can be taken (Verstraete et al., 2009).
Recently, efforts have been made to increase the methane production potential of the sludge
by concentrating the sludge. Chemically enhanced primary treatment and biological sorption
and flocculation are the most promising techniques to achieve higher sludge concentrations
(Cardoen, 2011). A WWTP in Strass, Austria, has proven to be completely self-sufficient
in terms of energy. The WWTP takes advantage of biological sorption of COD with the
’adsorption-belebungsverfahren’ or A/B-process, followed by a nitrogen treatment with a very
The future of wastewater treatment
9
energy friendly partial nitrification and anaerobic ammonia oxidation (anammox) process. This
way the plant was able to energy independent by generating electricity which only accounts for
11% of the total caloric energy available in the wastewater (Wett et al., 2007).
10
The activated sludge process
Chapter 3
High-rate activated sludge
3.1
Introduction
High-rate activated sludge systems differ from low-rate activated sludge systems in terms of
sludge loading rate (Bx ). The loading of CAS systems is generally in order of 0.25 kgBOD
kgVSS-1 d-1 (PaDEP), while high rate activated sludge systems operate at loading rates of 2
to 10 kgBOD kgVSS-1 d-1 Boehnke (1977). High-rate systems operate at a much lower sludge
retention time (SRT) (0.1-1 day) than conventional systems (8-20 days). At these low residence
times, removal of BOD by processes such as biological adsorption and storage becomes more
important, as opposed to full mineralization to CO2 (Boehnke et al., 1998).
3.2
Biological Sorption
Biological sorption is the physicochemical adhesion of pollutants (i.e. organic particulates and
heavy metals) by microorganisms or flocs without using it for immediate aerobic or anaerobic
biodegradation (independent of metabolism) (Aksu, 2005). It has been proven to effectively remove slowly or non-biodegradable pollutants from waste streams (Aksu, 2005; Vijayaraghavan
and Yun, 2008). The concept of wastewater treatment by biological sorption originates from
the hypothesis that BOD removal in activated sludge reactors is a result of three distinct and
non-simultaneous processes (Figure 3.1:
First, the organic compound, both soluble and particulate, attaches to the floc through physicochemical adhesion (adsorption). Secondly, the adsorbed compounds are hydrolyzed by extracellular enzymes to smaller, soluble compounds. Finally, the degraded compounds are absorbed
into the cell and used for anabolism and catabolism (Bunch and Griffin, 1987).
11
12
High-rate activated sludge
Figure 3.1: The kinetic theory of BOD removal according to Bunch and Griffin (1987). In the first
minutes, the BOD concentration in the wastewater drops dramatically due to rapid adsorption of the BOD onto the biomass. The BOD concentration then peaks again as it
becomes hydrolyzed and detaches. The soluble BOD is thereafter mineralized to CO2 ,
which permanently decreases the BOD in the wastewater.
The adsorptive step is typically fast (15 to 30 minutes) in comparison with the latter two.
However, the peak related to utilization of soluble substrate has proven to be very difficult
detect and has yet to be confirmed in practice. Adsorption becomes easily exploitable for
high-load systems as the contact time for these systems is within the optimal range and the
impact of adsorption increases with higher sludge loading rates (Tan and Chua, 1997; Zhao
et al., 2000). Guellil et al. (2001) reported that with batch tests, on average 45% of the
non-settleable fraction of domestic wastewater was removed due to biosorption. Steady state
conditions were achieved after 20 to 40 minutes and a total adsorption capacity of 40-100 mg
COD g VSS-1 was observed. Biological sorption can also be quantified in terms of observed
yield, which should be higher than theoretically possible in an regular activated sludge system
as particulate organics are quickly transformed to sludge (Zhao et al., 2000).
Biological Sorption
13
Bio-sorption of organics is theorized to be possible due to the excretion of extracellular polymeric substances (EPS) by the microorganisms (Sheng et al., 2010). EPS is a complex highmolecular-weight mixture of polymers, which has a very high binding capacity due to the high
number of carboxyl and hydroxyl groups present. However, the role of EPS in bio-sorption,
-flocculation and sludge settling behavior is still intensely questioned and debated. Indeed,
certain studies show that the amount of EPS correlates positively with bioflocculation properties of the sludge (Urbain et al., 1993), while other researchers suggest a negative correlation
(Goodwin and Forster, 1985). The latter stresses the importance of the composition of the EPS
rather than the quantity. The amount of EPS production decreases in function of increasing
loading rates, with a sharp decrease around 1 gCOD gVSS-1 d-1 . Interestingly, Chao (1979)
observed that the electrokinetic potential (zeta potential) of sludge flocs remains constant with
increasing loading rates.
High-load activated sludge systems are naturally very dynamic systems, as they are characterized by very low SRTs and contact times. When sludge is subjected to these very dynamic
conditions, the microorganisms switch from a dominating growth response to a dominating
storage response (Daigger and Grady, 1982). This storage response is characterized by the
uptake and conversion of rapidly biodegradable chemical oxygen demand (rbCOD) to storage polymers in the form of polysaccharides and polyhydroxyalkanoates (PHA) (Majone et al.,
1999). Sludge that is subjected to an intermittent feeding regime (feast-famine), used in contact
stabilization or SBR-type systems, is more likely to show a storage response. Cech and Chudoba (1982) found that activated sludge fed with glucose on an intermittent regime exhibited
a glucose accumulation capacity of 600-750 mg gglucose -1 while those being fed on a continuous regime only prevailed to accumulate 174 mg gglucose -1 . Similar to biological sorption, the
storage response exhibits itself in a higher observed (heterotrophic) yield than commonly seen
in aerobic wastewater treatment systems (Gujer and Jenkins, 1975). PHA production is furthermore positively influenced by short SRTs. A more strict selection of PHA-accumulating
bacteria due to the quick turnover of the genetic community present at low SRT might be the
reason for this correlation (Chua et al., 2003).
Two well-known processes that make use of of bio-sorption and storage to achieve COD removal are the A-stage of the adsorption-belebungsverfahren or A/B-process, developed in the
1970’s by professor Bohnke, (Boehnke, 1977) and contact-stabilization (CS) process, a popular
wastewater treatment technique in the United States (Kayser, 2005).
14
3.3
High-rate activated sludge
A/B-process
Developed in 1977 at the university of Aachen by professor Boenhke, the A/B-process is a
two-stage activated sludge process (Figure ??). Largely resistant to toxic shocks and pH
fluctuations, the process was originally designed for treatment of industrial wastewater, but
has also been applied for treatment of domestic sewage (Boehnke et al., 1998). The process
consists of two stages: The first stage, the A-stage, is characterized by a very high sludge loading
rate ranging from 2-10 kgBOD kgVSS-1 d-1 , low contact time (15-30min) and a extremely low
sludge age (0.1-0.5 days)(Boehnke et al., 1998; Salomé, 1990). The effluent of the A-stage is
discharged into the second stage, the B-stage. This is operated at a low sludge loading rate
(0.1 kgBOD kgVSS-1 d-1 ) and high SRT (10-16 days) to ensure nitrification (Boehnke, 1977;
Wett et al., 2007). Denitrification can occur in the A- or B-stage depending on the design of
the plant.
Influent
Effluent
A-stage
B-stage
Return Sludge
Return Sludge
Waste Sludge
Waste Sludge
Figure 3.2: Simplified scheme of the A/B-process. Based on (Salomé, 1990).
The A-stage reactor has a chemical oxygen demand (COD) removal efficiency of 50-70%
(de Graaff and Roest, 2012; Salomé, 1990). A large fraction of the removed COD (50-70%) is
adsorbed on the sludge and removed trough sludge disposal. The remaining COD is removed
trough mineralization to CO2. This demonstrates the concept of bio-flocculation used in highrate activated sludge systems. However, WWTP Nieuwveer, a full-scale A/B-plant in the
Netherlands, achieves only 5-15% removal efficiency for rbCOD (de Graaff and Roest, 2012).
Contact Stabilization Process
15
It is observed that, due to the young nature of the sludge, fast-growing microorganisms present
in the sludge are capable of only using a small fraction of the readily biodegradable COD
(Haider et al., 2000, 2003). The adsorptive nature of the sludge also lowers the overall oxygen
demand of the process as adsorption is, unlike respiration, in most cases oxygen-independent.
A-stage reactors can therefore operate at dissolved oxygen (DO) levels close to zero (Boehnke
et al., 1998). Modern A-stage reactors generally operate at a DO level of 1-2 mg O2 L-1 ,
however the optimal DO level of the process has not been investigated so far (de Graaff and
Roest, 2012). This indicates that still little is known about the optimal process conditions of
the A-stage.
The low SRTs of A-stage processes do not support growth of nitrifiers, that typically require
residence times of 8 days (Ekama, 2008), so that nitrogen removal efficiencies in the A-stage
process are typically very low. Conversely, the low loading rate and high SRT in the B-stage is
ideal for nitrification of the A-stage effluent (Boehnke et al., 1998). The conventional aerated
heterotrophic nitrification used in the B-stage requires a fair amount of energy and aeration
cost. Aeration energy of a CAS plant with nitrification accounts for more than 60 to 70% of
the total energy consumption of the plant (Zessner et al., 2010). Therefore, an evolution of
conventional nitrification to state-of-art autotrophic one-stage partial nitrification/anammox
processes like OLAND is desirable from an energy-efficiency point of view (Wett et al., 2007).
Using this methodology, the energy requirements and subsequently the cost of the overall process can be drastically reduced. OLAND is 30-40% less expensive compared to conventional
nitrification/denitrification, mainly due to lower energy requirements (aeration) and the absence of carbon dosing necessary for denitrification (Fux and Siegrist, 2004).
3.4
Contact Stabilization Process
Since the invention of the activated sludge process in by Ardern and Lockett in 1914, researchers have been trying to increase the efficiency of the process to cope with the increasing
organic loads that the treatment plants have to deal with. In the 1950’s, a new process called
’biosorption process’ was developed to deal with these problems (Ullrich and Smith, 1951).
Later renamed as the ’contact stabilization’ (CS) process, it can be seen as a modification of a
conventional activated sludge system. As the effluent criteria in Europe became more stringent,
advanced CAS processes were favored over CS as the latter does not attain adequate nitrogen
mitigation (Maribo, 2009).
Recently, the CS process has regained interest from the scientific community as it fits well within
the scope of sustainability due to it’s bio-sorption capacities. To fully exploit the advantages
of CS, high-load contact stabilization process (HiCS) is being developed.
16
High-rate activated sludge
In contrast with the single aerated basin used in the CAS process, CS systems use two basins:
the contactor and stabilizer (Figure 3.3). Wastewater flows into the contactor where it is
mixed with sludge which is in a near-endogenous state for a relatively short time (15 to 60
minutes) and is subsequently separated in the settler (Rittman and McCarty, 2011; Liu et al.,
2009). Organic matter is predominantly removed via biological sorption and storage, as the
conditions in the contractor are considered to be very dynamic (Ramalho, 1997; Bunch and
Griffin, 1987). The return sludge is pumped in to the stabilizer which is the most distinct and
important feature of a CS process. The stabilizer is an aerated tank where the return sludge
is aerated for at least 1.5 hours to ensure a near-endogenous state.
Influent
Effluent
Stabilizer
Contactor
Return Sludge
Waste Sludge
Figure 3.3: Simplified graphical representation of the contact stabilization process. Image based on
(Metcalf and Eddy, 2003).
As contact stabilization works with a very short contact time, the CS process is optimized to
remove particulate and other quickly sorbable COD from a waste stream (Figure 3.1). The
overall performance of the process is therefore largely influenced by the influent characteristics,
more specific the solubility index (SI). SI is the ratio between soluble COD and total COD
present in the influent. Contact stabilization will attain increasing COD removal efficiencies
with decreasing SI. The COD removal efficiency of the conventional CS process can go up to
85-90% (Ramalho, 1997).
Regular CS plants operate at a specific loading rate of 0.2-0.6 kgBOD kgVSS-1 d-1 , similar to
Energy & Operational costs
17
the CAS process (0.25 kgBOD kgVSS-1 d-1 ) (PaDEP). However, to efficiently concentrate COD
in the waste sludge in order to achieve energy efficient sludge digestion, loading rates achieved
by conventional CAS and CS plants are too low (Verstraete and Vlaeminck, 2011). High-load
contact stabilization might have some distinct advantages over the A-stage to achieve highly
concentrated sludge: The stabilizer ensures that a strong concentration gradient is present
between the sludge and the wastewater. This stimulates the feast-famine response of the
sludge. Furthermore, the low SRT used in high-load systems might induce a fast selection of
these microorganisms are adapted to fast adsorption and storage of substrate Huang:2000. For
HiCS only one study is available, attaining very promising results on the topic Zhao et al.
(2000). The researchers designed a laboratory scale CS system to exploit the advantages at
higher specific loading rate (0.5-1 kgBOD kgVSS-1 d-1 ). Under optimal conditions the labscale HiCS reactor had a total COD removal efficiency of 70-80% and a yield coefficient of 0.69
kgVSS kgBOD-1 applied (approximately 0.79 kgVSS kgCOD-1 removed ) , which is in effect higher
than the yield reported in CAS systems (0.4 kgVSS kgCOD-1 removed ), but in the same order as
a full-scale A-stage plant (0.44 - 0.93 kgVSS kgCOD-1 removed ) Despite the fact that the applied
loading rate is too low to be considered truly high-rate according to the definition above, the
results are an incentive to investigate the possibility of a real HiCS system as an alternative
to the A-stage. Furthermore, no systematic comparison between an A-stage and HiCS reactor
has been made.
3.5
Energy & Operational costs
Adsorption and storage found in high-rate systems have an influence on the energy requirements and consequently the operational costs of the plant. As both processes are metabolism
independent (i.e oxygen-independent), the aeration requirements of a high-rate activated sludge
plant will be lower than the carbonaceous oxygen demand of a conventional activated sludge
plant that works solely on respiration (nitrification oxygen demand is ignored).
In a full-scale A-stage, the aeration energy requirements per unit of COD removed is between 0.039 kWh kgCOD-1 removed at WWTP Strass, Austria and 0.169 kWh kgCOD-1 removed
at WWTP dokhaven (de Graaff and Roest, 2012). However, a side note has to be made as the
energy requirements per unit of COD removed are also dependent on the plant design and the
efficiency of the aerators (both kL a and electrical efficiency). For example, the very low energy
requirement per unit of COD at Strass is due to rigorous energy-cuts as a result of many energy
benchmark studies and in Austria (Wett et al., 2007; Jonasson, 2007)
In 2009, WWTP Nieuwveer had a total energy requirements for aeration of 27.1 kWh IE-1 COD180
y-1 , which is 50% of the total energy requirements (51.4 kWh kWh IE-1 COD180 y-1 ). The total
18
High-rate activated sludge
aeration cost is estimated at e3.25 IE-1 COD180 y-1 , which is 53% of the total yearly operational
cost (e6,17 IE-1 COD180 y-1 ) (personal communication, Nieuwveer). Remarkably, the A-stage
only requires 7.87 kWh IE-1 COD180 y-1 (0.104 kWh kgCOD-1 removed ) , which is 30% of the plant’s
total aeration energy requirements. Thus, if the A-stage is to be combined with an energyefficient mainstream OLAND process, energy-positive sewage treatment might be feasible in
the near future. As a comparison, a typical CAS plant with nitrification consumes on average
44 kWh IE-1 COD180 y-1 and aeration requires more than 70% of the total plant’s total energy
requirements (Zessner et al., 2010).
To illustrate, at the WWTP in Strass, the electricity requirements dropped from 70 Wh
IE-1 COD110 to 60 Wh IE-1 COD110 after the implementation of the DEMON process for the
treatment of reject water originating from sludge dewatering, creating an energy surplus of
11% and thus making the whole plant self-sustainable in terms of energy (Wett et al., 2007).
Chapter 4
Operational Parameters for
activated sludge processes
4.1
Sludge retention time (SRT)
The sludge retention time (SRT) is a critical parameter in any activated sludge process. It
determines the overall sludge concentration, influences the biomass growth (yield) and imposes
whether nitrification is occurring or not (Ekama, 2008). There are multiple ways to define the
SRT, hence it is critical to clearly state which one is used to avoid confusion. In its most simple
version, the SRT of an activated sludge plant with is calculated by the ratio between the mass
in the sludge reactor (Xr Vr ) and the mass of sludge that is wasted per day (Xw Qw ) (Eq. 4.1).
SRT “
Xr Vr
Xw Qw
(4.1)
SRTreactor “
Xr Vr
Xw Qw ` Xe Qe
(4.2)
SRTsystem “
Xr Vr ` Xs Vs
Xw Qw ` Xe Qe
(4.3)
In a real world application, sludge separation is not perfect, therefore the most commonly used
Eq. is 4.2. Here, the settler is assumed not ideal and some sludge will leave with the effluent.
This has to be incorporated into the sludge balance. The most accurate definition of the SRT
for a wastewater treatment plant is given by Eq. 4.3. This approach incorporates the mass of
19
20
Operational Parameters for activated sludge processes
sludge present in the settler in the mass balance. However, it is not obvious to have an accurate
idea of the sludge mass in the settler. Note that the sludge concentration in the reactor or
thickened sludge does not to be known to accurately control the SRT for this method.
All the definitions above describe both the SRT in the aerobic and anoxic (settling and denitrification) zones. If one wants to relate the SRT with processes like aerobic heterotrophic and
autotrophic growth, the aerobic SRT is more relevant. The aerobic SRT van be defined as the
SRT multiplied by the faction of the sludge that is in an aerobic zone at any point in time
(fx,aer ) (Carlos et al., 1999).
SRTaer “
fx,aer Xr Vr
Xw Qw ` Xe Qe
(4.4)
In general, the effluent quality of the activated sludge plants increases with increasing SRT as
the diversity of micro-organisms increases (Ekama, 2009). However, higher SRTs are mainly
maintained when nitrification of the wastewater is necessary. Higher SRT also stimulates the
sludge to degrade micropollutants (Clara et al., 2005). The daily dosage of sludge wasted per
day is in direct correlation with the SRT (Eq.s 4.1-4.3). Lower SRT will increase the amount
of sludge needs to be disposed, which imposes a higher overall cost. Sludge disposal costs
accounts up to 40% of the overall treatment cost (Zessner et al., 2010). However, a very high
SRT will lead to elevated sludge concentrations in reactor if no sufficient reactor and settler
volume is present, which might induce settling problems.
For high-rate pre-treatment systems, nitrification is not required and furthermore sludge production has be to maximized to enhance biogas production. Lowering the sludge age increases
the fraction of total COD that will be present in the excess sludge and lower the fraction of
influent COD that will be oxidized in the overall process (Figure 4.1) (Van Haandel and Van
Der Lubbe, 2007). Lowering the SRT in combination with a high specific loading rate can
therefore be seen as a potential way to biologically concentrate the incoming waste stream.
A lower SRT leads to a young and fast-growing type of sludge which can be digested easily.
Bolzonella et al. (2005) calculated the relationship between SRT and specific gas production
(SGP) (in m3 kgVSfed -1 ) (Eq. 4.2) by testing the digestibility of the secondary sludge of four
different WWTPs in Italy. All four operate on a different SRT ranging from 8 to 45 days. The
exponential equation clearly indicates that anaerobic digestion will achieve higher SGP values
if the sludge is younger. Higher SGP leads to an increased energy efficiency, as more biogas
can be converted to heat and electrical energy.
Sludge retention time (SRT)
21
Figure 4.1: Fraction of the COD that will remain in the effluent (mSte ; left), be oxidized to CO2 (mSo ;
center) and trapped in the excess sludge (mSxv ; right) in function of the COD. fns is the
non biodegradable dissolved influent COD fraction. The dotted lines show the influence of
the non-biodegradable dissolved influent COD fraction (fnp ). Adapted from (Van Haandel
and Van Der Lubbe, 2007)
SGP “ 0, 23e´0,028SRT
(4.5)
The SRT has a considerable influence on the bioflocculation and therefore settling properties
of the sludge. Lower SRT generally leads to a worse sludge flocculation (higher sludge volume
index, SVI) and poorer effluent clarification (Chao, 1979; Li and Yang, 2007). Low specific
loading rates and SRTs will lead to a higher growth rate of filamentous bacteria as it is theorized
that, due to the larger surface-to-volume ratio of the filamentous bacteria, filamentous bacteria
will outcompete floc-forming bacteria as they have a greater accessibility to the substrate
(Martins et al., 2004). EPS production of the sludge is also affected by the SRT, although
there is no real consensus regarding the influence. Some suggest that EPS production increases
with increasing SRT (Sesay et al., 2006), while others found that there was no change in total
EPS with changing SRT (Liao et al., 2001; Li and Yang, 2007). Both conclude that the ratio
of proteins to carbohydrates of the EPS increases when the SRT is raised.
22
4.2
Operational Parameters for activated sludge processes
Loading rate (Bx )
In wastewater treatment, the specific loading rate (Bx ), also known as the food to microorganism (F/M) ratio, is a critical design parameter. It is defined as the rate amount of
substrate addition (Si Qi ) per sludge mass present in the reactor (Xr Vr ) (Eq. 4.6) (Metcalf and
Eddy, 2003).
Bx “
Si Qi
Xr Vr
(4.6)
As Eq. 4.6 suggests, a change in loading rate can be achieved in three distinct ways: The
influent flow rate (Qi ) of the wastewater can change, which is commonly the case in a full
scale activated sludge plant. The influent characteristics might shift (i.e low or high strength
wastewater). A third way is to lower the sludge concentration in the reactor.
An increase in loading rate will lead to an increase in sludge production as more substrate
will become available (Ekama, 2008). In conventional activated sludge systems, the diurnal
variability in loading rate due to human activities or storm flood imposes a great challenge for
the operator (Dauphin et al., 1998).
4.3
Dissolved Oxygen level
The degradation of carbonaceous organics in an activated sludge plant is an aerobic process.
Dissolved oxygen (DO) is used by various aerobic micro-organisms to mineralize organic carbon
into CO2 and use the chemical energy for growth and reproduction (Slonczewksi and Foster,
2009). DO levels play a crucial role in many processes occurring in activated sludge. These
processes include BOD degradation and ammonia oxidation (nitrification), but also floc formation and the development of the microbial community (floc formers or filamentous bacteria)
(Metcalf and Eddy, 2003). Oxygen stress induced by low DO levels has a clear influence on
the settleability of the sludge. Martins et al. (2003) observed SVI levels over 250 ml g-1 in
lab scale SBR reactors when operated at DO levels below 1.1 mg L-1 . The researchers also
observed an amplified effect when very high loading rates (47 kgCOD kgTSS-1 d-1 were applied.
Aerating the reactor above 2 mg L-1 quickly solved the problem. A rise in filamentous bacteria
is commonly the reason for elevated SVI measurements and decreased settling properties (Palm
et al., 1980; Pernelle et al., 2001). Extreme oxygen deficiency might even induce deflocculation
of the sludge (Zhang and Allen, 2008).
Temperature
4.4
23
Temperature
As an activated sludge plant is nearly always operated outdoors, it is subjected to the daily
and seasonal temperature changes. These variations have certain implications considering the
design and operation of an activated sludge. Being a (micro)biological process, temperature
alters the maximum growth rate (µmax and decay rate (kd ) of the microorganisms following
the Arrhennius relation θT ´20 . The factor theta depends on the parameter influenced by
temperature and the species or genus of microorganisms itself (Metcalf and Eddy, 2003).
The most important implication of sewage plants - but less important in the scope of this thesis
- is that the minimal SRT required for nitrification (SRTmin,nitri ) will drastically change with
changing temperature. Lower temperatures imply a higher SRTmin,nitri , thus should not be
overlooked when designing and operating a CAS plant (Ekama, 2008).
As Eq. 4.7 suggests, the observed biomass yield (Yobs ) will also change in function of temperature.
Yobs “
Y
1 ` kd SRT
(4.7)
The observed biomass growth yield will go down with increasing temperature as the decay rate
increases. For high-rate systems where maximum sludge production is prioritized, higher temperatures are less desirable. This is in strong contrast with the ’B-stage’ in the ZeroWasteWater
framework, as the efficiency of partial nitration/anammox increases with increasing temperature (Dosta et al., 2008), thus a trade-off has to be made. The relative importance of the
temperature decreases with SRT as endogenous respiration decreases, hence the temperature
effect will not be very profound on high-rate systems. In addition, Sayigh and Malina (1978)
observed that Yobs did not alter at all between 4 ˝ C and 20 ˝ C. The oxygen consumption per
unit of substrate removed did, however, increase from 0.64 mgO2 mgCOD-1 soluble at 4 ˝ C to
1.38 mgO2 mgCOD-1 soluble at 34 ˝ C. A lower temperature might therefore be more desirable
from a energy-efficiency point-of-view.
24
Operational Parameters for activated sludge processes
Chapter 5
Research Question
High-rate systems are ideal to serve as an advanced pre-treatment to efficiently concentrate
the influent in the form of sludge. Two main technologies, the A-stage of the A/B-process
and high-load contact stabilization (HiCS) are proposed. Despite the fact that A/B reactors
have run in Germany for many years, little is still known about the effects of the operational
parameters on the process. In theory both processes should have an increased yield compared
to a conventional activated sludge. The HiCS is a new concept which may further increase the
yield as it is a HRAS variant of the contact stabilization process, which on its own has a higher
yield than a CAS reactor. The goal of this thesis is to map the effects of the SRT of a lab-scale
A-stage on the operational efficiency and yield of the lab-scale operation. This will be compared
to two HiCS reactors which run at a specific, but different contact time over stabilization time
(τ ), which might have a profound influence on the biomass yield of the reactor. In short, the
research questions are as followed:
• How does the SRT affect the maximum yield of a high-rate system operating at very
short SRT? Does it still follow the same trend as observed in Eq. 4.7, or does it shift due
to the effect of sorption and storage?
• Given an optimal SRT for HiCS, does τ influence the maximum yield? If it does, what
is the relationship between the two?
The efficiency will be quantified in terms of COD removal efficiency, yield and waste sludge
characteristics. The latter will be performed with a bio-methane potential (BMP) test to
quantify the digestibility of the sludge. The efficiency will be compared with each other to
conclude which operate in a superior fashion. Lastly, the contact and stabilization phase will
be characterized with respirometry, to gain more insight in the effects of the two phases.
25
26
Research Question
Part II
Materials, Methods and Reactors
Chapter 6
Reactors
6.1
General
All reactors were run in a sequencing batch reactor (SBR) configuration unless stated otherwise
(Figure 6.1 and 6.2).
Figure 6.1: Two lab scale SBR reactors operated to simulate the A-stage and HiCS reactors.
29
30
Reactors
D
pH
B
A
C
E
Figure 6.2: Scheme of the SBR reactor used for all reactors.
A = influent pump; B = effluent pump; C = sludge waste pump; D = pH controller; E =
level controller.
6.1.1
Materials
Pumping was perfumed with peristaltic pumps (Watson Marlow 505u, 313s, 503s, 603s; homemade with pump head). pH of both reactors was controlled using a Consort R305 pH controller.
Stirring was performed with a top-stirrer (Ika RW16 basic, 200RPM), aeration was achieved
with compressed air and a cylindrical aquarium aeration stone (Scalare Oxy-Tech, 20x80mm,
porosity not specified).
6.1.2
Influent
All continuous reactors ran on complex synthetic wastewater SYNTHES, which is representative for real domestic sewage (Aiyuk and Verstraete, 2004). All reactors were operated with a
tenfold dilution (800 mgCOD L-1 ) of concentrated SYNTHES (Table 6.1), which corresponds
to high-strength wastewater ((Metcalf and Eddy, 2003)). This was a necessary design choice
General
31
as, given the total cycle time of the reactors (see further), high-strength was required to run
the reactors under high-load conditions.
Table 6.1: Composition of concentrated SYNTHES .
Component
Amount (mg L-1 )
Chemical Compounds
Urea
NH4 CL
Na´acetate · 3 H2 O
peptone
MgHPO4 · 3 H2 O
K2 HPO4 · 3 H2 O
FeSO4 · 7 H2 O
CaCl2
1600
200
2250
300
500
400
100
100
Food ingredients
Starch
Milk powder
Dried yeast
Soy oil
2100
2000
900
500
Trace metals
Cr(NO3 )3 · 9 H2 O
CuCl2 · 2 H2 O
MnSO4 · H2 O
NiSO4 · 6 H2 O
PbCl2
ZnCl2
Overall estimated parameters
COD total
COD dissolved
COD particulate
15
10
2
5
2
5
8000
2500
5500
32
6.1.3
Reactors
Inoculation & Start-up
All continuous reactors were inoculated with A-sludge from WWTP Nieuwveer, Breda in such
a way that the mixed liquor suspended solid (MLSS) in the reactors was 2-3 g VSS L-1 . Twodays acclimatization period was maintained in which the sludge was not wasted, to allow an
adaptation.
6.1.4
SRT
The solid retention time (SRT) is a critical parameter in this thesis and should therefore be
well defined. A differentiation is made between the nominal sludge retention time SRTnom
(equation 6.1) and the actual sludge retention time SRTact (equation 6.2). Formula 6.1 is
used as a design parameter and was used to calculate the hourly MLSS volume that required
wasting whereas the 6.2 is an operational parameter and takes into account the amount of
sludge that leaves the system with the effluent. Note that if one applies wasting during the
react phase (MLSS wasting), the reactor biomass concentration is equal to the waste biomass
concentrations (Xr “ Xw ).
Vr
Xr Vr
“
Xw Qw
Qw
(6.1)
Xr Vr
Xr Vr
“
Xw Qw ` Xe Qe
Xr Qw ` Xe Qe
(6.2)
SRTnom “
SRTact “
6.1.5
Sampling, follow-up and analysis
Daily samples were taken on weekdays. In the react phase, 40mL of reactor content (approximately 2% of the total volume) was taken as reactor sample for TSS/VSS analysis and DO
and pH were measured. A large amount of influent volume was taken out of the influent barrel
for total suspended solids (TSS), volatile suspended solid (VSS) and COD analysis. Effluent
of both reactors was separately sampled in the decant phase and also used for TSS/VSS and
COD analysis.
6.2
6.2.1
A-stage
Operational conditions
Two A-stage reactors with different sludge retention times (SRT) (referred to as A-1 and A-2)
were operated simultaneously to ensure all operational conditions are equal except the SRT
A-stage
33
(Table 6.2). Reactor A-1 was controlled at a SRT of 0.5d whereas A-2 was controlled at a SRT
of 1.5 days. The reactors had a total operating volume of 2L. A constant volumetric loading
rate (Bv ) of 7.5 gBOD L-1 d-1 was applied using complex synthetic wastewater SYNTHES as
influent. The pH was controlled between 7.75 and 8.25 and the DO was kept above 2 mgO2
L-1 to ensure no oxygen limitation was present. The room was temperature controlled at 15
℃, which is the average temperature in Western Europe.
Table 6.2: Design and operational parameters of the lab-scale A-stage reactors .
A-stage
A-1
A-2
SRTnom
Bv
T
pH
DO
tcycle
VER
HRT
0.5
1.5
6
15
7.75-8.25
ą2
60
50
2
d
gBOD L-1 d-1
℃
mgO2 L-1
min
%
h
The total cycle time (tcycle ) of the SBRs was set on 1h. The volume exchange ratio (VER)
applied is 50%, which results in a nominal hydraulic retention time (HRT) of 2h. The cycle
started with a short idle time of 2 minutes to ensure no overlap of the decant and fill & react
phase is occurring. Next, the SBR is stirred, aerated and gradually filled with 1L of influent
for 30 minutes in the fill + react phase (influent flow rate = 2L h1 ). The reactor is stirred
and aerated for an additional 10 minutes in the react phase. In this phase, MLSS is wasted
to control the SRT. After a cumulative time of 42 minutes, the sludge is allowed to settle for
15 minutes, whereafter 1L of effluent is decanted in 3 minutes (effluent flow rate = 20L h-1 )
(Figure 6.3).
34
Reactors
Figure 6.3: Graphical representation of the SBR-cycles used to mimic an A-stage reactor
6.3
6.3.1
High-load contact-stabilization
Operational conditions
For HiCS systems the ratio of the contact time (tcont ) over stabilization time (tstab ) is a crucial
design parameter. For practical purposes this ratio will be defined as τ .
The high-load contact-stabilization (HiCS) reactors differ from the A-stage reactors in that
the SBR operates with 2 distinctly different phases to simulate a 2-stage CSTR design: the
contact phase and the stabilization phase. The contact phase is defined as the anoxic (not
aerated) fill phase in the SBR rotation. The stabilization phase is defined as an aerated idle
phase (V “ 21 Vreactor ; VER = 50%) at the start of the SBR cycle where the thickened sludge
is allowed to achieve a more endegenous state.
The two HiCS reactors (referred to as HiCS-1 and HiCS-2) operated at different τ and constant
SRTnom . HiCS-1 had a τ of 1 (tstab “ 20min; tcont “ 20min), while HiCS-2 had a τ of 17
(tstab “ 35min; tcont “ 5min). tcycle was set on 1 hour (Figure 6.4). The sludge had 15
minutes to settle, the effluent was drained in 3 minutes and the idle time was 2 minutes. The
same ratio between the fill + contact and contact phase was applied as the one used in the
A-stage for fill + react and react phase.
High-load contact-stabilization
35
Figure 6.4: Graphical representation of the SBR-cycles used to mimic HiCS reactors
HiCS-1 and HiCS-2 were operated simultaneously to ensure all operational conditions are equal
except τ (Table 6.3). A constant volumetric loading rate (Bv) of 6 gBOD L-1 d-1 was applied
using complex synthetic wastewater SYNTHES as influent. The pH was controlled between
7.75 and 8.25 and the DO was kept above 2 mgO2 L-1 to ensure no oxygen limitation was
present. The room was temperature controlled at 15 ˝ C, which is the average temperature in
Western Europe.
36
Reactors
Table 6.3: Design and operational parameters of the two lab-scale HiCS reactors .
HiCS
HiCS-1
HiCS-2
τ
SRTnom
Bv
T
pH
DO
tcycle
VER
HRT
6.4
6.4.1
1
1{7
1.5
6
15
7.75-8.25
ą2
60
50
2
d
gBOD L-1 d-1
℃
mgO2 L-1
min
%
h
Low-load reference reactors
Operational conditions
As low-load reference reactors to the A-stage and HiCS reactors, a conventional activated
sludge (CAS) and contact stabilization (CS) reactor were operated at lab-scale. Operational
parameters were chosen in such a way that both reactors reflected a full-scale installation on a
lab-scale scale in a SBR environment (PaDEP).
To achieve this, the HRT of the system was increased to 12 hours while keeping the VER equal
to their high-load counterparts (50 %). As a consequence, the tcycle was increased to 6 hours
for both reactors. The volumetric loading rate was decreased to 0.45 g BOD L-1 d-1 which is
equivalent to 0.3 g BOD L-1 d-1 at MLSS = 3 gVSS L-1 . The SRT was increased to 8 days for
both the CAS and CS setup.
Conventional activated sludge
The SBR design of the CAS reactor was designed in the same fashion as the A-stage. The
cycle started with an idle time of 10 minutes. The fill + react phase took place for 4.5 hours
followed by a react phase for 45 minutes. The sludge was allowed to settle for 30minutes and
the effluent is drained in 5 minutes (Figure 6.5).
Low-load reference reactors
37
Contact Stabilization
The contact stabilization is based on the HiCS reactor in terms of SBR design. The cycle starts
with an idle time of 10 minutes, whereafter the sludge is stirred and aerated in the stabilization
phase for 5 hours. Aeration is subsequently stopped and the reactor is allowed to fill with fresh
influent for 25 minutes in the fill + contact phase. The contact phase is set on 5 minutes
where the sludge and influent can react at full reactor volume. The sludge is then settled for
30 minutes and the effluent is subsequently drained in 5 minutes.
Figure 6.5: SBR cycles of the two lab-scale SBR reactors operated to simulate two low-rate reference
reactors.
A summary of the operational parameters for the reference reactors is given in Table 6.4
38
Reactors
Table 6.4: Operational parameters of the two lab-scale reference reactors .
Reference Reactors
CAS
CS
τ
SRTnom
Bv
T
pH
DO
tcycle
VER
HRT
1
10
-
8
1
15
7.75-8.25
ą2
6
50
12
d
gBOD L-1 d-1
℃
mgO2 L-1
h
%
h
Chapter 7
Batch tests
Next to continuous reactors, certain batch experiments were performed for characterization.
This is includes respirometry for the HiCS-cycle, influent and effluent characterization and
biochemical methane potential (BMP) and fermentation tests for sludge characterization
7.1
7.1.1
Respirometry
Contact and stabilization phase characterization
The oxygen uptake rate (OUR) and COD trend of the contact and stabilization phase of the
HiCS reactors is tested with respirometric setup at BioMATH, Faculty of Bioscience Engineering.
The reactor had a total volume of 2L and was temperature controlled with a water bath at 15
˝ C. The DO (Mettler Toledo) and pH (Mettler Toledo HA405-DXK-58/120) was logged every
second with Labview (American Instruments). The pH was controlled between 7.75-8.25 with
a acid/base pulse dosage pump (Gibson Minipuls).
The reactor was filled with one liter of sludge originating from HiCS-1 or HiCS-2 and the
SBR cycles of both reactors (see Figure 6.4) are simulated. Influent is pumped into the the
reactor vessel with a peristaltic pump (Watson Marlow 505u for HiCS-1; Homemade pump
with pump head for HiCS-2) and effluent is drained manually. A reactor sample is taken at
regular intervals and filtered over a 1.5 µm filter for particulate COD separation and 0.45 µm
for colloidal and soluble COD separation.
The respirometric yield was determined according to (Majone et al., 1999):
Yobs “ 1 ´
39
OU
S
(7.1)
40
Batch tests
where OU is the total oxygen uptake and S is the soluble COD removal.
7.1.2
Influent characterization
The (diluted) influent and effluent is characterized with respirometry in batch (Figure 7.1). A
1L flask was filled with 0.5L of conventional activated sludge from the CAS plant at Destelbergen, Ghent. The sludge was aerated overnight to bring it into an endogenous state. Next, 0.5L
influent was poured into the reactor and alternately aerated to measure the OUR. The DO
and temperature was logged with a membrane DO electrode (YSI 5700 series, Yellow Spring
International) which is connected to a microcontroller. The OUR was calculated by the microcontroller using the pulse width modulation (PWM) technique of the aeration cycle and
subsequently visualized with a computer. This method calculates the OUR based on the relation between the on- and off-intervals of the aeration cycle of the reactor. A full mathematical
background and calculations can be found in (Cantunda et al., 1999). The experiment was
stopped after 7 hours and the VSS of the reactor was measured in triplicate.
B
A
C
pc
microcontroller
Reactor
Figure 7.1: Schematic representation of the respirometric batch experiment.
A = aeration; B = DO probe connected to the microcontroller; C = Stirrer
Biochemical methane potential test
7.2
41
Biochemical methane potential test
Biochemical methane potential (BMP) test was performed to quantify the biogas production
of a specific type of sludge. The BMP test was performed in triplicate in 120mL penicillin
bottles. The total working volume was 40mL, which resulted in a headspace of 80mL (66,67
%). For the A-sludge, mesophilic anaerobic digestion (AD) sludge from mixed sources was
used as inoculum. For the HiCS-sludge granular AD sludge was used. The AD sludge was
diluted (with tap water) in the penicillin bottles until a total VSS content of 10 g L-1 (0.4
gVSSinoculum per penicillin bottle) was achieved. The specific load (Bx ) was 0.3 gCODsludge
gVSS-1 inoculum (0.12 g COD per bottle). This was achieved by measuring the COD of the
sludge (in triplicate) and diluting it with tap water accordingly. The pH is measured and the
penicillin bottles were subsequently capped and 50mL air was withdrawn to create a moderate
underpressure. The penicillin bottle were subsequently put at 34 ˝ C and the internal pressure
measured with at regular intervals. When the internal pressure is stabilized (i.e. all sludge has
been anaerobically digested), the biogas content (CH4 , CO2 and H2 ) was measured with gas
chromatography (compact GC). At the end of the experiment, the bottles are decapped and
the pH is measured again
A BMP test was conducted with sludge originating from reactors A-1, A-2, HiCS-1 and HiCS-2
(Table 11.3). For each type of inoculum, an inoculum control test was performed. For the
control, 0.4 gVSSinoculum was put into the bottle and filled with tap water to a total volume of
40mL.
Table 7.1: Overview of BMP-test setup for both the A-stage and HiCS sludge. .
Name
A-sludge
Control A-1
A-2
Inoculum Mesophilic AD-sludge
Volume
40
Bx
0.3
T
34
7.3
HiCS-sludge
Control HiCS-1 HiCS-2
Granular AD-sludge
40
0.3
34
mL
gCOD gVSS-1
℃
Fermentation test
A fermentation batch test quantifies the amount of short chain fatty acids (SCFA) that can
be produced due to fermentation of a specific type of sludge. The fermentation batch test
was performed in triplicate in 120mL penicillin bottles. Each experiment had a inoculum and
42
Batch tests
sludge control test in triplicate.
7.3.1
inoculum preparation
The inoculum was produced in 2 steps. First, 0.1L of anaerobic inoculum is mixed with
0.5L tap water and 0.4L A-sluge, originating from WWTP Nieuwveer, Breda. The mixture is
corrected to pH 7 using P-buffer and allowed to acclimatize at the desired temperature (34 ˝ C
for mesophilic inoculum) for 4 days. After 4 days 0.5 L of the mixture is discarded and 0.5L
of fresh A-sludge is added. The new mixture is again allowed to acclimatize at the desired
temperature for 4 days.
7.3.2
buffer preparation
As a buffer, a 2M K2 HPO4 /KH2 PO4 buffer at pH 7 is used. To achieve this, 11.7 g of KH2 PO4
and 30.9 g K2 HPO4 is dissolved in 1 L of distilled water.
7.3.3
HiCS Sludge
HiCS sludge fermentation test was performed on a specific basis. The total working volume was
80mL, which results in an initial headspace of 40mL (33.33 %) 10mL of mesophilic inoculum
was mixed with 30 mL of the least concentrated sludge and buffered with 40mL of P-buffer.
The more concentrated sludge was diluted with tap water to the COD concentration of the
lesser concentrated sludge and thereafter the same procedure is used. The inoculum- and sludge
control tests were set up in the same fashion as the A-stage control tests.
Chapter 8
Chemical analysis
8.1
Chemical Oxygen Demand
Chemical oxygen demand (COD) was measured using Nanocolor kits (Machery-Nagel). Three
types of kits were used (COD-160, COD-1500 and COD-15000) with an analysis range of 10 160 mgCOD L-1 , 100 - 1500 mgCOD L-1 and 1000 - 15000 mgCOD L-1 respectively.
8.2
Biochemical Oxygen Demand
Biochemical oxygen demand (BOD) is the amount of amount of oxygen used by a microbial
community at 20 ˝ C for a certain period of time.
In this thesis, BOD5 was measured using manometric respirometric BOD OxiTop method.
This test is based on the automatic measurement of the pressure depletion in a closed bottle at
20 ˝ C due to oxygen consumption by microorganisms (public health association , APHA). A
predetermined amount of sample (dependant on the expected BOD) was put into shaded glass
bottles and 1mL of freshly prepared inoculum was added. Next, mineral solutions (phosphate
buffer at pH 7.2, MgSO4 · 7 H2 O, CaCL2 and FeCL are added in accordance with 0.1 % of the
sample volume to avoid micronutrient depletion. The bottles are subsequently tightly screwed
with an OxiTop manometer (WTW GmbH; Germany) and put at 20 ˝ C.
8.3
Total suspended solids, total solids, volatilizable suspended
solids, volatilizable solids
Total solids (TS) were determined by drying the sample in a crucible for at least 12h at a
temperature of 105 ˝ C. Volatilizable solids were measured by ashing the dried sample in a
43
44
Chemical analysis
muffle oven (Nabertherm B150, Germany) for 1.5h at 550 ˝ C. Total suspended solids (TSS)
were determined by filtering a predetermined amount of sample over a 0.45 µm filter and drying
it at 105 ˝ C for at least 1h. The volatilizable suspended solids are determined by ashing the
dried filter at 550 ˝ C for 1.5h.
The TS and TSS (g L-1 ) are quantified by subtracting the weight of the crucible/filter from the
weight after drying at 105 ˝ C and dividing by the volume of the original sample used. The VS
and VSS are quantified by subtracting the weight of the crucible/filter after ashing at 550 ˝ C
from the weight after drying at 105 ˝ C and also dividing by the volume of the original sample
used.
8.4
8.4.1
Nitrogen
Total Ammonia Nitrogen
Total ammonia nitrogen (TAN) was analyzed by steam destination according to the standard
methods (public health association , APHA). 20 mL of sample was brought inside a Kjeldahl
tube so that the nitrogen content is between 0.1 and 6 mg NH+
4 ´N. Approximately 0.4g of
MgO was added to convert all ammonium to ammonia. The ammonia wasW subsequently
distilled (Vapodest 30s, Gerhardt) and condensed into a boric acid indicator solution, which is
at pH 5.3. The total ammonia in the sample is determined with a back-titration of the boric
acid with a 0.02 M HCL solution.
8.4.2
Total Kjeldal Nitrogen
Total kjeldal nitrogen (TKN) was analyzed by steam destination according to the standard
methods (public health association , APHA). Kjeldal nitrogen (which is the combination both
ammonia nitrogen and organically bound nitrogen) was determined by sample destruction at
400 ˝ C for 1.5 hours in the presence of sulphuric acid and a catalyst containing copper sulfate
and potassium sulfate. Next, NaOH was added automatically to convert all ammonium to
ammonia, which is subsequently distilled in boric acid at pH 5.3. The TKN in the sample is
determined with a back-titration of the boric acid with a 0.02 M HCL solution.
8.5
Short Chain Fatty Acids
Short Chain Fatty Acids (SCFA) were determined using gas chromatography ((30 m x 0.32
mm x 0.25 µm; Agilent, Belgium) and a flame ionization detector (FID). Samples were treated
with sulfuric acid (50% v/v) and sodium chloride. 2-methyl hexanoic acid is used as internat
Biogas composition
45
standard (IS). SCFA’s present in the liquid sample and the internal standard (IS) are subsequently extract into ether. The prepared sample(1 µL) was injected at 200 ˝ C with a split
ratio of 60 and a purge flow of 3 mL min-1 . The oven temperature increased by 6 ˝ C min-1
from 110 ˝ C to 165 ˝ C where it was kept for 2 min. The flame ionization detector (FID) had
a temperature of 220 ˝ C. The carrier gas was nitrogen at a flow rate of 2.49 mL min-1 .
8.6
Biogas composition
Biogas composition was analyzed with a Compact GC (Global Analyser Solutions, Breda, The
Netherlands), equipped with a Porabond precolumn and a Molsieve SA column. Concentrations
of CH4 , CO2 and H2 were determined using a thermal conductivity detector with a lower
detection limit of 1 ppmv for each gas component.
8.7
Particle size distribution
To measure the particle size distribution (PSD), a Mastersizer S (Malvern, Malvern, UK) with
a glass flask as a small volume dispersion unit was used. Before the actual measurement, each
sample was added so that the obscuration was between 10 and 30 % for sludges and between 1
and 10% for influents and effluents. The real and imaginary refractive indices for the particles
were set at 1.596 and 0.100. The dispersant refractive index was 1.333. Results were obtained
using the ’polydisperse’ analysis model in the Mastersizer model
8.8
pH, DO and Pressure measurements
pH was measured with a consort C5010 multimeter. The pH electrode was calibrated weekly
between pH 7 and pH 10 DO was measured using a Hach-Lange HQ 30d in mg O2 L-1 The
internal pressure of penicillin bottles was measured with a UMS infield 7 with a tensiometer
attached.
46
Chemical analysis
Chapter 9
Methods
9.1
9.1.1
Yield calcuation
Observed yield
To calculate the observed biomass yield (Yobs in gVSSproduced gCODconsumed -1 ), the cumulative
biomass growth in the reactor over t days (Xr,growth,t ) is plotted in function of the cumulative
COD removal. Biomass growth was determined by calculating the daily sludge balance (Eq.
9.1).
Xr,growth,t “
pXr,t ´ Xr,t´1 qVr
` Xe,t Qe ˚ ∆t ` Xr,t Qw ` Xr Vsa
∆t
(9.1)
COD removal was calculated determining the daily substrate balance. The substrate balance
is designed in such a way that all effluent particulate COD is considered sludge. Substrate in
the effluent is therefore considered to only consist of soluble COD so that the substrate balance
is given by Eq 9.2.
CODremoved,t “ CODtot,i,t Qi ´ CODdiss,e,t Qe
9.1.2
(9.2)
Maximum yield
The observed yield is dependent on the SRT. To get a more representative idea of the biomass
of the system, the ´ SRT independent ´ maximum maximal yield (Y) was calculated from
the observed yield using the following Eq. 9.3. bh,T was the temperature corrected endogenous
decay rate (bH “ 1.24θT ´20 ; d-1 ; θ “ 1.029) and fH the endogenous residue fraction (0.2) (?).
47
48
Methods
Y “
9.2
Yobs p1 ` bH,T SRT q
1 ` bH,T fH SRT
(9.3)
COD balance
The amount of COD that is wasted per unit of time was expressed as:
Sw,t “ Xr,t Qw fcv ∆t
(9.4)
fcv is the COD/VSS ratio of the sludge and was experimentally determined for every reactor.
The amount of COD that is lost with the effluent was expressed as:
Se,t “ Stot,e,t ∆t
(9.5)
The amount of COD that is sampled was calculated as:
Ssample,t “ Xr,t Vs fcv
(9.6)
The fractions of the balance are calculated by making the fraction cumulative and dividing by
the cumulative influent COD load.
Part III
Results
Chapter 10
Continuous reactors: Performance
indicators
10.1
Operational conditions
10.1.1
Reference reactors
The reference reactos ran for a total time of 21 days to set a baseline for the high-rate reactors.
Both operated at an average SRTact of 10.15˘4.96 days and 9.12˘2.49 days for CAS and
CS respectively (Table 10.4). No significant difference in specific loading rate (0.59˘0.16
gBOD gVSS-1 d-1 for CAS versus 0.40˘0.11 gBOD gVSS-1 d-1 for CS) was observed (Table
10.1). Biomass concentration was found to be significantly different higher in the CAS reactor
(1.64˘0.45 gVSS L-1 than the CS reactor (1.77˘0.49 gVSS L-1 ). Incoming suspended solids
(SS) were identical for both reactors (0.12˘0.03 gVSS L-1 ), while effluent suspended solids of
the CS reactor were slightly higher (0.09˘0.03 gVSS L-1 ) compared to its CAS counterpart
(0.11˘0.02 gVSS L-1 ).
The difference in removal efficiency of total and dissolved COD was mildly significant (0.05 ą
p ě 0.01) between the two reactors. The CAS reactor attained higher removal efficiencies for
both CODtot (74˘12%) and CODdissolved (85˘8%) than the CS reactor (53˘21% for CODtot
and 66˘17% for CODdissolved . The CAS reactor performed better (although not significantly)
for CODparticulate removal aswel (56˘21%) in comparison to the CS reactor (12˘51%). The
latter was very dynamic with periods of sludge washout (i.e negative removal efficiencies), hence
the large standard deviation.
Dissolved oxygen in the reactors was well above the threshold of 2 mgO2 . pH was controlled
51
52
Continuous reactors: Performance indicators
between 7.75-8.25 and both reactors remained between those two boundaries. Ambient temperature of the lab was controlled at 15˝ C ,hence both reactors operated at the same temperature.
Table 10.1: Statistical significance between the key parameters of CAS and CS reactor obtained with
the Mann-Whitney-U non-parametric test in R.
‹ “ mildly significant (0.05 ą p ě 0.01) ; ‹‹ “ significant (0.01 ą p ě 0.001);
‹ ‹ ‹ “ strongly significant (p ą 0.001).
Parameter
SRTact
Bx
Xr
% CODrem,tot
% CODrem,diss
% CODrem,part
10.1.2
Significance p-value
´
´
‹‹
‹
‹
´
0.7209
0.3865
0.0041
0.0148
0.0104
0.1049
A-stage reactors
The A-stage reactors were operated for 14 days at a SRT of 0.41˘0.06 d and 1.39˘0.83 d for
A-1 and A-2 respectively (Table 10.4). Reactor biomass concentration of A-1 was significantly
larger than the one observed in A-2 (Table 10.2). As a result, the specific loading rate (Bx ) of
A-1 was also significantly larger (11.73˘4.75 gBOD gVSS-1 d-1 and 2.94˘1.06 gBOD gVSS-1
d-1 ), as the reactor was operated on the basis of volumetric loading rate. Suspended solids in
the influent are equal for both reactors (both batches were fed from the same influent barrel),
while the effluent suspended solids was higher for reactor operated at a higher SRT (0.13˘0.06
gVSS) than one at lower SRT (0.10˘0.02 gVSS).
A-2 attained a higher CODtot removal (51˘17%) compared to A-1 (38˘19%). The same
trend is observed for CODdiss (62˘25% and 32˘19% respectively). Particulate COD removal
fluctuated a lot and did not attain high efficiencies for both reactors. In addition, no significant
difference was observed (33˘30% and 19˘42% respectively).
Operational conditions were similar for both reactors and in range of the predesignated values.
pH was controlled between 7.75-8.25 and the average DO was above 2 mgO2 L. Temperature
was slightly different for both reactors.
Operational conditions
53
Table 10.2: Statistical significance between the key parameters of the two A-stage reactors obtained
with the Mann-Whitney-U non-parametric test in R.
‹ “ mildly significant (0.05 ą p ě 0.01) ; ‹‹ “ significant (0.01 ą p ě 0.001);
‹ ‹ ‹ “ strongly significant (p ą 0.001)
Parameter
SRTact
Bx
Xr
% CODrem,tot
% CODrem,diss
% CODrem,part
10.1.3
Significance p-value
‹‹‹
‹‹‹
‹‹‹
‹
‹
´
4 · 10-5
0.0002
0.0002
0.0172
0.0173
0.7125
HiCS reactors
The HiCS reactors were run for 16 days to see how it compared to a lab-scale A-stage and CS
reactor. The SRT of HiCS-1 and HiCS-2 was 1.02˘0.06 days and 1.22˘0.44 days respectively
(Table 10.4). There was no significant difference between both reactors in terms of SRT (Table
10.3). The specific loading rate applied on HiCS-1 (5.41˘1.85 gBOD gVSS-1 d-1 was found to
be significantly higher than the one applied on HiCS-2 (1.92˘0.40 gBOD gVSS-1 d-1 ). HiCS2 attained a significantly higher average biomass concentration throughout the experiment.
Influent suspended solids was higher for A-1 in this setup as the sampling technique was
different. A large standard deviation on the average influent SS was observed for HiCS-1.
The average effluent VSS was lower for HiCS-1 (0.09˘0.02 gVSS L-1 ) compared to HiCS-2
(0.12˘0.04 gVSS L-1 ).
COD removal efficiency was low for both reactors. HiCS-1 achieved on average 33˘15% CODtot
removal efficiency and 27˘15% CODdissolved removal efficiency. HiCS-2 attained a higher ´
though not significantly higher ´ average CODdissolved removal efficiency (44˘16%), but lower
average CODtot removal efficiency (30˘17%) due to an overall lower CODparticulate removal
efficiency (20˘54% for HiCS-2 compared to 31˘32% for HiCS-1). Both CODparticulate removal
efficiency had large fluctuations and therefore a large standard deviation is observed.
Dissolved oxygen in the reactors was well above the threshold of 2 mgO2 . pH was controlled
between 7.75-8.25 and both reactors remained between those two boundaries. Ambient temperature of the lab was controlled at 15˝ C hence both reactors operated at the same temperature.
54
Continuous reactors: Performance indicators
Table 10.3: Statistical significance between the key parameters of the two HiCS reactors obtained with
the Mann-Whitney-U non-parametric test in R..
‹ “ mildly significant (0.05 ą p ě 0.01) ; ‹‹ “ significant (0.01 ą p ě 0.001);
‹ ‹ ‹ “ strongly significant (p ą 0.001)
Parameter
SRTact
Bx
Xr
% CODrem,tot
% CODrem,diss
% CODrem,part
10.1.4
Summarizing table
Significance p-value
´
‹‹‹
‹‹
´
´
´
0.4418
0.0006
0.0042
0.6454
0.0830
0.8785
CAS
SRTact
Bx
Reactor VSS
Influent SS
Effluent SS
% CODrem,tot
% CODrem,diss
% CODrem,part
DO
pH
T
CS
A-1
A-2
HiCS-1
1.39 ˘ 0.83
2.94 ˘ 1.06
1.02 ˘ 0.06
5.41 ˘ 1.85
1.22 ˘ 0.44 d
1.92 ˘ 0.40 gBOD gVSS-1 d-1
0.95 ˘ 0.31
0.15 ˘ 0.10
0.10 ˘ 0.02
2.35 ˘ 0.65
0.15 ˘ 0.10
0.13 ˘ 0.06
1.67 ˘ 0.46
0.26 ˘ 0.21
0.09 ˘ 0.02
3.32 ˘ 0.83 gVSS L-1
0.18 ˘ 0.04 gVSS L-1
0.12 ˘ 0.04 gVSS L-1
38 ˘ 23
32 ˘ 19
19 ˘ 42
51 ˘ 17
62 ˘ 25
33 ˘ 30
33 ˘ 15
27 ˘ 14
31 ˘ 32
10.15 ˘ 4.96 9.12 ˘ 2.49 0.41 ˘ 0.06
0.59 ˘ 0.16 0.40 ˘ 0.11 11.73 ˘ 4.75
1.64 ˘ 0.45 1.77 ˘ 0.49
0.12 ˘ 0.03 0.12 ˘ 0.03
0.09 ˘ 0.03 0.11 ˘ 0.02
74 ˘ 12
85 ˘ 8
56 ˘ 21
53 ˘ 21
66 ˘ 17
12 ˘ 51
Operational conditions
Table 10.4: Summarizing Table of the mean performance indicators and conditions of all the reactors that were operated during this
study.
HiCS-2
30 ˘ 17
44 ˘ 16
20 ˘ 54
%
%
%
8.76 ˘ 0.62 9.12 ˘ 0.41 3.22 ˘ 2.83 2.91 ˘ 2.72 3.79 ˘ 2.31 3.45 ˘ 3.39 mgO2 L-1
8.07 ˘ 0.24 8.09 ˘ 0.12 7.99 ˘ 0.38 7.89 ˘ 0.23 7.90 ˘ 0.35 8.01 ˘ 0.14
14.60 ˘ 0.2 14.6 ˘ 0.2 14.01 ˘ 0.98 13.36 ˘ 2.83 14.65 ˘ 0.15 14.65 ˘ 0.15 ˝ C
55
56
Continuous reactors: Performance indicators
10.2
Biomass concentration, SRT and specific loading rate
10.2.1
Reference Reactor
3
2.5
2
1.5
1
0.5
0
Bx
g BOD gVSS-1 d
SRT
(d)
Reactor VSS
(gVSS/L)
The biomass in both the CAS and CS reactor steadily increased to 2.5 gVSS L-1 and 1.5 gVSS
L-1 respectively (Figure 10.1). At day 8, both reactors experienced a wash-out of biomass
which dropped to 1 gVSS L-1 . The biomass thereafter restored itself, however the CS reactor
attained higher stable biomass concentrations than the CAS reactor. Remarkably, before day
8 the CAS reactor had more biomass than the CS reactor. The SRT was achieved solely on
biomass washout in the effluent. No sludge wasting was performed. Therefore the SRT was
heavily influenced by the effluent SS. SRT of both reactors followed the same trend and reached
a minimum at day 8 due to washout. The SRT never exceeded 10 days. The specific loading
rate applied on both reactors was relatively constant and did not significantly differ from each
other.
12
10
8
6
4
2
0
1
0.8
0.6
0.4
CAS
CS
0.2
0
0
5
10
15
20
Time (d)
Figure 10.1: The reactor VSS (upper), SRT (middle) and Specific loading rate (bottom) of the reference reactors CAS (‚) and CS () in function of the total runtime of the experiment.
SRT A-2
SRT A-1
Table 1 Bx CAS
Table 1 Bx CS
Biomass concentration, SRT and specific loading rate
10.2.2
57
A-stage reactors
The biomass concentration of reactor A-1 remained relatively stable for the duration of the
experiment and slightly decreased in the in the second period of the experiment at day 10
(Figure 10.2). The SRT of A-1 was successfully controlled and stable throughout the whole
experiment. The specific loading rate applied to A-1 was around 6 gBOD gVSS-1 d-1 and later
peaked to 14.38 gBOD gVSS-1 d-1 . The waste pump of reactor A-2 malfunctioned between
day 6 and 8. In this period, the reactor VSS peaked to 3.34 gVSS L-1 at day 8 and the SRT
increased to 3.13 days. Little effect was observed on the specific loading rate. When the waste
pump was again functional, the sludge concentration stabilized at around 2.7 g VSS/L and the
SRT remained stable as well at 1 day.
Bx
4
A-1
A-2
3
2
1
3.5
3
2.5
2
1.5
1
0.5
0
(gBOD gVSS-1 d-1)
SRT
(d)
Reactor VSS
(gVSS/L)
A-2 waste
pump off
15
10
5
0
0
5
10
15
Time (d)
Figure 10.2: The reactor VSS (upper), SRT (middle) and Specific loading rate (bottom) of the A-stage
reactors A-1 (‚) and A-2 () in function of the total
SRT A-2
SRT A-1
A
58
10.2.3
Continuous reactors: Performance indicators
HiCS reactors
The biomass concentration of HiCS-2 was consistently larger throughout the whole runtime
of the experiment (Figure 10.3). Both reactors also achieved to maintain relatively stable
biomass concentration. No effect on the biomass concentration was observed when the waste
pump failed between day 6 and 7. The SRT however did increase to 1.80˘0.21 days for HiCS-1
and 2.17˘0.03 days for HiCS-2 in this period. The SRT remained relatively stable for the rest
of the reactors’ runtime. The specific loading rate applied to HiCS-1 was higher than the one
applied to HiCS-2, but was stable for both reactors.
waste pump
faillure
Reactor VSS
(gVSS/L)
5
HiCS-1
HiCS-2
4
3
2
1
0
2.5
SRT
(d)
2
1.5
1
0.5
Bx
(gBOD gVSS-1 d-1)
0
7
6
5
4
3
2
1
0
0
5
10
15
Time (d)
Figure 10.3: The reactor VSS (upper), SRT (middle) and Specific loading rate (bottom) of the A-stage
reactors A-1 (‚) and A-2 () in function of the total
SRT A-2
SRT A-1
VSS_HiCS J
VSS_HiCS K
COD removal efficiency
59
10.3
COD removal efficiency
10.3.1
Reference reactors
Both reactors showed an increase in total and dissolved removal efficiency throughout the
runtime of the experiment (Figure 10.4). CAS achieved a maximum CODtot and CODdissolved
removal efficiency of 88% and 95% respectively at day 14. The CS reactor experienced a big
increase in COD removed after 10 days (roughly 1 time the SRT) which resulted in a much more
dynamic pattern. The CS achieved a maximum CODtot and CODdissolved removal efficiency of
75% and 92% respectively at day 16.
CS
CAS
1
COD removal efficiency (-)
0.8
0.6
0.4
0.2
CODtotal
CODdissolved
0
0
5
10
15
20
0
5
10
15
20
Days (d)
Figure 10.4: CODtotal (‚) and CODdissolved (ˆ) removal efficiencies observed for the lab-scale CAS
and CS reactor.
60
10.3.2
Continuous reactors: Performance indicators
A-stage reactors
The COD removal performance of the A-stage reactors is visualized in Figure 10.6. Both
reactors showed an increase in total and dissolved removal efficiency throughout the runtime of
the experiment. A-1 achieved a maximum CODtot and CODdissolved removal efficiency of 52%
and 61% respectively at day 11. The daily removal efficiency is fairly dynamic throughout the
whole experiment and did not achieve a stable value. The removal efficiency of reactor A-2
showed an upward trend the first 5 days and reached a maximum CODtot and CODdissolved
removal efficiency of 82% and 94% respectively at day 9. Compared to A-1, the removal
efficiency of A-2 is dynamic and did not achieve a stable value. CODparticulate removal efficiency
is not shown in the graph as it was not independently measured.
A-2
A-1
1
CODtotal
CODdissolved
COD removal efficiency (-)
0.8
0.6
0.4
0.2
0
0
5
10
15
0
Time (d)
5 A-2 waste 10
pump off
15
Figure 10.5: CODtotal (‚) and CODdissolved (ˆ) removal efficiencies observed for the A-stage reactors
A-1 (left) and A-2 (right).
COD removal efficiency
10.3.3
61
HiCS reactors
COD removal efficiency was very dynamic during the runtime. Both reactors showed a incremental increase in CODdissolved , which peaked at 47% at day 11 for HiCS-1 and 64% on day
13. The same upward trend is observed for CODtotal removal efficiency in HiCS-1. No such
trend is observed for HiCS-2.
HiCS-2
HiCS-1
1
CODtotal
CODdissolved
COD removal efficiency (-)
0.8
0.6
0.4
0.2
0
0
5
10
15
0
5
10
15
Time (d)
Figure 10.6: CODtotal (‚) and CODdissolved (ˆ) removal efficiencies observed for the HiCS reactors
HiCS-1 (left) and HiCS-2 (right).
62
10.4
Continuous reactors: Performance indicators
Particle size distribution
A volume weighted particle size distribution was made of fresh influent and effluent from both
HiCS-reactors to observe the shift in particle sizes (Figure 10.7). The influent show a left
tailed distribution with a mean diameter over volume (D[4,3]) of 65.13µm. The distribution of
the both effluent samples shifted towards the left indicating that smaller particles became more
abundant or larger particle had been removed. The mode of HiCS-2 effluent (22.84µm) was
larger than HiCS-1 effluent (27.30µm). However, the mean diameter over volume of HiCS-1
and HiCS-2 is respectively 27,47µm and 34.79µm. A noticeable difference is observed at 100
µm, where the HiCS-2 has a relative larger abundance of particles that HiCS-1 (1.06%).
Volume fraction
(%)
10
8
6
Influent
Effluent HiCS-1
Effluent HiCS-2
4
2
cummulative volume
fraction (%)
0
0.1
100
1
10
100
1000
100
1000
80
60
40
20
0
0.1
1
10
µ
Particle Size ( m)
Figure 10.7: Particle size distribution of influent and effluent of both HiCS reactors.
Yield
63
10.5
Yield
The CAS reactor had the lowest observed yield (Figure 10.8) but the maximum yield (Eq.
9.3 was higher than the A-stage reactors (Table 10.5). The observed yield of the CS reactor
was lower than the HiCS-reactors, but attained a higher maximum yield than HiCS-1. HiCS-2
attained the highest observed and maximum yield of all reactors that were operated. A-2 had
the lowest theoretical yield.
120
A-1
A-2
Linear fit A-1
Linear fit A-2
CAS
Linear fit CAS
Cumulative biomass growth
(gVSS)
100
80
60
40
20
0
120
100
HiCS-1
HiCS-2
Linear fit HiCS-1
Linear fit HiCS-2
CS
Linear fit CS
80
60
40
20
0
0
50
100
150
200
250
Cumulative COD removal
(gCOD)
Figure 10.8: Cumulative biomass growth in function of the cumulative substrate removal for all reactors operated in this study. The observed yield is calculated by a lineair fit trough the
datapoints. A-1 (‚); A-2 (‚); HiCS-1 (˚); HiCS-2 (`) CAS (İ); CS(IJ)
64
Continuous reactors: Performance indicators
Table 10.5: attained cumulative VSS yield of all reactors and their corresponding maximum yields.
Yobs
CAS
A-1
A-2
CS
HiCS-1
HiCS-2
10.6
0.27
0.39
0.29
0.31
0.47
0.67
˘
˘
˘
˘
˘
˘
Y
0.01
0.02
0.01
0.01
0.04
0.03
0.53
0.42
0.34
0.65
0.55
0.79
˘
˘
˘
˘
˘
˘
0.09 gVSSprod
0.02
0.03
0.07
0.05
0.04
gCODcon -1
”
”
”
”
”
COD Balance
The CAS reactor achieved the highest fraction of COD that was not retrieved of all reactors
(71%) (Figure 10.9). As both reference reactors did not require sludge wasting to keep the
SRT below a certain threshold, no COD was recovered. HiCS-2 attained the highest fraction
of COD that can be recovered (18%). Moreover, most of the COD of this reactor is lost with
the effluent. Non-retrievable fraction was low compared to the other high-rate reactors. The
HiCS-1 reactor scored lowest on the COD recovery potential for high-rate reactors (8%).
COD Balance
65
100%
90%
Fraction of influent COD
80%
70%
Not retrieved
Sample
Effluent diss
Effluent Part
Waste
60%
50%
40%
30%
20%
10%
0%
CAS
A-1
A-2
CS
HiCS-1
HiCS-2
Figure 10.9: Percentage of incoming COD lost with the effluent, both particulate (effluent part) and
soluble (effluent diss), sampling and that can be harvested trough waste sludge (waste).
The balance is closed with the fraction of COD that was not retrieved by any of the prior
mechanisms (presumably respiration; see discussion).
66
Continuous reactors: Performance indicators
Chapter 11
Batch tests
11.1
Respirometry
11.1.1
Influent characterization
Respirometry is a quick and accurate tool to characterize various wastewater- and sludgespecific parameters (Spanjers and Vanrolleghem, 1995). It can be used to determine how
different COD fractions (rbCOD and sbCOD) relate to each other (Sperandio and Etienne,
2000). This way, the representability of the synthetic wastewater can be tested and see how it
compares actual wastewater.
The reactor was spiked with 500ml influent containing 770 mg L-1 COD (48% in soluble form),
which brought the total initial COD in the reactor to 167.5mg COD. The COD/BOD5 ratio of
the influent was experimentally determined at 0.75, thus the initial BOD5 in the reactor was
estimated at 125.5 mg BOD5 . The reactor had a biomass concentration of 2.97˘0.08% gVSS
L-1 . The load was therefore 0.11 gCOD gVSS-1 . The endogenous respiration of the sludge
was 10.19˘0.51 mgO2 h-1 . The total oxygen uptake (OU) in 7 hours was 124.5˘13.79 mgO2
(Figure 11.1). The OUrbCOD ( 1 ) was 43.87˘9.62 mgO2 and 35% of the total OU, while the
OUsbCOD ( 2 ) was 80.63˘9.82, which was 64% of the total OU. The fractionation between
rbCOD and sbCOD was estimated based on the largest (negative) value obtained from the first
derivative of the OU, i.e., the place where oxygen uptake slows down due to the depletion of
rapidly degradable substrate (Figure 11.1). In total, 74.3% of the COD was degraded during
the 7 hours of contact time. Therefore, 26% of the total COD was considered rbCOD, while
48% was sbCOD. In total, 26% of the COD was not degradable. The specific oxygen uptake
rate (SOUR) of the rbCOD fraction was 34.5˘1.9 mgO2 gVSS-1 h-1 .
67
Batch tests
1
150
2
3
150
100
100
50
50
Cumulative OU
d(OUR)/dt
0
2000
Cumulative OU
(mg O2)
Oxygen uptake Rate (OUR)
(mgO2/h)
68
0
1000
0
−1000
d(OUR)/dt
−2000
0
1
2
3
4
5
6
7
Time (h)
Figure 11.1: exogenous OUR- and cumulative OU profile of freshly made influent. The bottom graph
represents the first derivative of the OU.
(1) rbCOD (2) sbCOD (3) endogenous respiration.
11.1.2
HiCS cycle characterization
To characterize the contact and stabilization phase of both HiCS reactors, respirometry was
used in combination with a COD profile. In the specific oxygen uptake rate (SOUR) profile
in Figure 11.2 and 11.3, four distinct phases were separated. The experiment was carried
out at two different biomass concentrations. The biomass concentration for HiCS-1 was 1.72
gVSS L-1 . The biomass concentration for HiCS-2 was 4.48˘0.22 gVSS L-1 . Note that due to
unforeseen circumstances, the sludge from the HiCS-2 reactor experienced a short pH shock (pH
11) the day before the experiment, which might have killed a reasonable amount of biomass.
The VSSalive might be lower than estimated. The volumetric gas transfer coefficient (kL a) was
experimentally determined in triplicate and was 0.59 min-1 for HiCS-1 and 0.48 min-1 for HiCS
2. In total 48 mg CODcol and 794 mg CODsol was added.
The stabilization phase was characterized by a first period of elevated SOUR (175 mgO2 gVSS-1
h-1 for HiCS-1 and 50 mgO2 gVSS-1 h-1 ) and subsequently a period of declining OUR. The
contact phase ´ where aeration was turned of ´ showed a spike in the SOUR (300 mgO2
gVSS-1 h-1 for HiCS-1 and 110 mgO2 gVSS-1 h-1 ) and then quickly declines to 0 (aeration was
turned off) for the remainder of the cycle.
Respirometry
69
The COD profile of HiCS-1 showed a downwards trend in the stabilization phase, but unlike
HiCS-2, did not achieve a stable value after 20 minutes. During the contact phase, where
in total 1L of influent was pumped into the reactor, of A-1, the COD steadily increased and
remained stable during the settle phase. HiCS-2 experienced an initial very steep increase in
COD during the contact phase at 35 minutes. Both reactors attained a larger CODcol fraction
at the end of the contact phase than the amount of CODcol that was pumped into the reactor.
As the CODcol was not measured in an accurate way, the respirometric yields calculated in
Table 11.1 are solely based on soluble substrate and were therefore underestimations (sorption
is ignored).
Overall, HiCS-1 achieved a greater respirometric observed yield than HICS-2 (Table 11.1) .The
yield in the contact phase is higher than the stabilization phase for both reactors, with a larger
discrepancy for HiCS-2 than HiCS-1.
Stabilization
Contact
Settle
Decant
350
500
300
250
0.85 gCOD gCOD-1
300
200
150
200
mgCOD in reactor
SOUR
(mg O2 gVSS-1 h-1)
400
Yresp,diss =
100
CODcoll+diss
50
DO = 0 mg L-1
CODdiss
100
SOUR
0
0
10
20
30
40
50
0
60
Time (min)
Figure 11.2: Respirogram and COD profile of one full SBR cycle of reactor HiCS-1.
70
Batch tests
Stabilization
Contact
Settle
Decant
350
500
CODcoll
CODdiss
400
SOUR
SOUR
(mg O2 gVSS-1 h-1)
250
Yresp,diss =
300
0.51 gVSS gCOD-1
200
150
200
DO = 0 mg L-1
mg COD in reactor
300
100
100
50
0
0
10
20
30
40
50
0
60
Time (min)
Figure 11.3: Respirogram and COD profile of one full SBR cycle of reactor HiCS-2.
Table 11.1: Respirometric observed yield based on soluble substrate of the stabilization, contact phase
of the HiCS reactors and the corresponding respirometric observed yield of the cycle.
HiCS-1
HiCS-2
Stab contact cycle Stab contact cycle
Yresp,diss 0.83
11.2
0.88
0.85 0.68
0.91
0.8 COD/COD
Biochemical Methane Potential
The sludges originating from the high-rate reactors were tested for anaerobic digestibly and
conversion to methane. Characterization of the sludges was performed to know the parameters
which might influence the digestive process (Table 11.2)
The BMP test was stopped when the pressure inside the penicillin bottles was stable, which
indicated that the conversion was finished (Figure 11.4). However, HiCS-2 was suddenly more
active when the test was stopped, but was stopped nonetheless due to time constraints. The
kinetics of the biogas production were different for the A-stage sludges and HiCS sludges were
different. Note that the HiCS sludges were digested with another type of inoculum (Table
11.3).
Biochemical Methane Potential
71
Table 11.2: Key ratio’s of the A- and HiCS-sludge.
A-1 A-2 HiCS-1 HiCS-2
COD/VS
1.55 1.18
22
16
COD/TKN
COD/TAN
17
44
1.86
10
32
1.82 gCOD gVS-1
16 gCOD gTKN-N-1
71 gCOD gTAN-N-1
A-1
A-2
HiCS-1
HiCS-2
Specific biogass production
mL biogass gCODsubstrate-1
500
400
300
200
100
0
0
20
40
60
80
Time (d)
Figure 11.4: Specific biogas production of the A-stage and HiCS sludges.
The sludges originating from the A-stage reactors converted 100% into methane, with only a
minor difference in biogas composition (CH4 /CO2 ratio). The HiCS sludges only converted 4244% into methane. The CH4 /CO2 ratio of the HiCS sludge was higher, indicating the relative
biogas composition was more methane-rich (Table 11.3).
The specific methane production was similar for all sludges (Figure 11.5 Table 11.3). A1 achieved the most methane production of all tested sludges and the A-sludge specifically.
HiCS-1 had the best methane production concerning the HiCS-sludges.
72
Batch tests
0,35
speecific methane production
(L methane/g VS substrate)
0,3
0,25
0,2
0,15
0,1
0,05
0
A-1
A-2
HiCS-1
HiCS-2
Figure 11.5: Specific CH4 production of the two HiCS sludges.
Table 11.3: Key parameters of BMP tests conducted on the A- and HiCS-sludge.
A-1
A-2
HICS-1
HiCSC-2
Sp. CH4 production 0.27 ˘ 0.04 0.24 ˘ 0.06 0.27 ˘ 0.04 0.24 ˘ 0.06 L CH4 gVS-1 sub
CH4 /CO2 ratio
3.27 ˘ 0.16 3.01 ˘ 0.26 4.48 ˘ 0.09 4.64 ˘ 0.17 ´
% COD to CH4
106 ˘ 5
124 ˘ 22
44 ˘ 4
42 ˘ 4
CODCH4 /CODsub
Fermentation
11.3
73
Fermentation
A fermentation test was set up beside a classical BMP to see how well the HiCS sludge could
be fermented into short-chain fatty acids (SCFA) at mesophilic temperatures. The two sludges
fermented with nearly the same specific SCFA production (0.63˘0.05 gCODSCFA gVSsub -1 for
HiCS-1 sludge and 0.64˘0.13 gCODSCFA gVSsub -1 for HiCS-2 sludge) (Figure 11.6). No results
of the A-stage sludges are reported as the protocol was not optimized at that time.
0,9
0,8
speecific SCFA production
(mg COD/g VS)
0,7
0,6
0,5
0,4
0,3
0,2
0,1
0
HiCS-1
HiCS-2
Figure 11.6: Specific SCFA production of the two HiCS sludges.
The overall recovery was low. Only 35% was of the sludge was fermented into SCFA and
CH4 (COD/COD-basis). Acetate was the most abundant SCFA present (13.9˘0.7% and
15.1˘0.3% for HiCS-1 and HiCS-2 sludge respectively). A small methane production was
noted (0.35˘0.19% and 0.19˘0.02%)
74
Batch tests
100%
90%
Substrate conversion (%COD)
80%
70%
Not recovered
Methane
isovalerate
butyrate
isobutyrate
propionate
Acetate
60%
50%
40%
30%
20%
10%
0%
HiCS-1
HiCS-2
Figure 11.7: percentage of substrate that is conversion into SCFA and CH4 .
Part IV
Discussion
Chapter 12
Introduction
The modern problems considering the energy, resource and cost efficiency of wastewater treatment are multidisciplinary and therefore require a multidisciplinary approach to be tackle. The
most important design change that has to be made is split of the degradation of carbonaceous
pollutants and nitrification in two distinctly different stages: the high-rate activated sludge
(HRAS) stage and a partial-nitration/anammox B-stage (Verstraete and Vlaeminck, 2011).
This gives researchers and decision makers more flexibility to increase the efficiency of each
stage separately.
This thesis has elaborated on the first stage of the treatment process. Two types of HRAS-stage
reactors, the A-stage and High-load contact stabilization (HiCS), were systematically tested
and compared in terms of COD removal efficiency (Section 13.1), yield (Section 15) and sludge
digestibility (Section 15). Further on, as the HiCS process is a novel concept under high-rate
conditions, the contact and stabilization phase is characterized trough respirometry (Section
14).
77
78
Introduction
Chapter 13
Reactors
13.1
COD removal efficiency
The COD removal efficiency of a reactor is an important indicator of performance. Generally,
the higher the COD removal, the better. As the HRAS-stage in the two-stage purification
process as a HRAS-stage is the first part of the two-stage AB process (see Introduction), the
effluent should have specific characteristics. To make sure that partial-nitration and anammox
can work with full potential, a low COD/N-ratio is desirable (Hulle et al., 2010). As a consequence, the HRAS-reactor should be as efficient as possible to remove COD from the waste
stream.
13.1.1
Reference reactors
The CAS reference reactor achieved the highest average CODtotal (74˘12%) and CODdissolved
(85˘8%) removal efficiency of all reactors that were operated. It also has the lowest reported
standard deviation on these removal efficiencies, indicating that the reactor was the most stable
one operated. The achieved removal efficiencies are lower than those operating at full scale
(Table 2.1).
The COD removal efficiency (53˘21%) of the reference CS reactor was on average 25% lower
than the CAS reactor and did generally fluctuate more. The removal efficiency did however
show a positive trend indicating that the sludge was still adapting to the different conditions
(i.e contact and stabilization phase). Table 10.4 reveals that the removal efficiency drastically
increased between day 9 and 11, which is about the one time the SRT of the reactor. This
suggests that the sludge was indeed still adapting to the new conditions and that after one
complete turnover, the sludge was more adapted to treat the wastewater. The average COD
79
80
Reactors
removal efficiency from day 12 till the end of the experiment is 66˘10% for CODtotal and
76˘14% for CODdissolved , which is in the same range as a pilot-scale plant in Colombia (86%),
which ran at a higher temperature (Sarria et al., 2011).
Despite operated at a SRT that allows nitrification, no nitrification was observed. The reactor
was inoculated with A-sludge from WWTP Nieuwveer, Breda. This sludge is very young (SRT
= 0.65 days) and will have a very low initial abundance of ammonia oxidizing bacteria (AOB)
and nitrite oxidizing bacteria (NOB). This might explain that, for the runtime of the reactor,
no sufficient of AOB and NOB was observed, even though the conditions to support their
growth was present.
13.1.2
A-stage & HiCS
The A-stage reactor operating at a higher SRT (1.39˘0.83 d; A-2) achieved a higher removal
efficiency (51˘17%) than the one operating at a very low SRT (0.41˘0.06 d; A-1; 38˘23%).
The significantly higher specific loading rate which is a result of the low biomass concentration
induced by the SRT might be the reason for the lower removal efficiencies observed for A-1.
When working with a constant volumetric loading rate (Bv ), the specific loading rate can
be seen as a consequence of the low SRT. Assuming steady state conditions and 100% COD
removal, the biomass in the reactor can be approximated with Eq. 13.1 (Ekama, 2008).
Xr “ Bv Yobs SRT
(13.1)
Therefore, at constant volumetric loading rate, the biomass concentration is solely dependent
on the observed yield and SRT. One can assume that a decrease in SRT is directly responsible
for a decrease in removal efficiency as the steady-state biomass concentration (and therefore the
specific loading rate) is lower. The very low SRT of 0.41 days applied in A-1 induced a severe
turn-over stress on the microorganisms, which was too high to maintain an adequate biomass
concentration in the reactor. Compared to the COD removal efficiency full-scale installation at
Breda which runs at an SRTact of 0.39 days and attains a COD removal efficiency of 51%, the
lab-scale A-stage reactor operating at a SRT of 1.5 days attained equal efficiency, whereas A-1
has worse efficiency. However, COD removal at Breda consists more of CODparticulate removal
(50%) than CODdissolved (10-15%) compared to the lab-scale A-stage reactors (A-1: 32˘19%
CODdiss removal and 19˘42 CODpart ; A-2: 62˘25% CODdiss and 33˘30% CODpart ) (de Graaff
and Roest, 2012). The influent at Breda might consist of soluble components which are more
complex and therefore harder to degrade than acetate, which is the main soluble component
COD removal efficiency
81
of the used synthetic wastewater. The particulate COD at Nieuwveer might also be relatively
bigger than the starch particulates in the synthetic wastewater. COD removal by means of
sedimentation might therefore be of greater relative importance as a removal technique at
Nieuwveer than in the lab-scale reactors.
Similar to the reference reactors, the HiCS reactor performed worse in terms of COD removal
efficiency compared to the A-stage. Whereas HiCS-1 (τ =1) achieved a slightly better CODtot
removal efficiency (33˘15%) compared to HiCS-2 (τ = 17 ; 30˘17%), HiCS-2 achieved a higher,
but non-significant, CODdissolved removal efficiency (44˘16% for HiCS-2 compared to 33˘15%
for HiCS-1). A longer stabilization phase might be the reason for higher removal efficiency,
however Figure 11.2 and 11.3 indicate that it generally takes 10 minutes to degrade all the
soluble rbCOD. Furthermore, at the end of the contact phase, HiCS-2 had a more profound
removal of soluble COD while having a lower SOUR at that moment, suggesting that storage
of rbCOD might be more pronounced with a shorter contact time and larger concentration
gradient. This is in line with the general consensus that a strong concentration gradient implies
a larger storage effect (Majone, 1996; Daigger and Grady, 1982). Particulate COD removal was
slightly larger for HiCS-1 (31˘32%) than HiCS-2 (20˘54%), however the standard deviations
on these averages are too large to draw any profound conclusion on the fact that a longer
contact time has an effect on the adsorption of particulate COD onto the cell wall. The
PSD profile (Figure 10.7) of HiCS-1 shows a more profound decrease in the larger fraction of
particles, with a relatively large difference around 100 µm, suggesting that 20 minutes might
be better for a more complete adsorption of particles. Total COD removal efficiency did not
significantly differ between the two HiCS-reactor, which might suggest that τ has no effect on
removal efficiency. The removal efficiency might therefore be dependent on other operational
or design parameters, like SRT or HRT.
13.1.3
Conclusion
As expected, the reference reactor obtained the best COD removal efficiencies. The A-stage
reactors attained moderately well COD removal efficiencies compared to full-scale operations.
A-2 would therefore be the best reactor if one only considers removal efficiency of the reactors.
The HiCS reactors achieved relatively poor removal efficiencies, however storage presumably
makes a major contribution towards the CODsoluble removal efficiency. The COD removal
performance of the HICS reactor might be τ -independent, which opens the possibility to choose
a specific τ where sludge production is maximized and look at other parameters to optimize
the COD removal efficiency.
82
13.2
Reactors
Yield & Recoverable COD
The HRAS-stage primary function in the ZeroWasteWater platform is to maximize sludge production, while efficiently removing COD from the waste stream. The Yield ´ and consequently
the amount of COD that can be harvested as sludge ´ is another very important parameter for
determining which HRAS-reactor is best suited to be implemented into the new framework.
13.2.1
Yield
This thesis will work with the maximum yields calculated rather than the observed yield. The
maximum (heterotrophic) yield (Y) is the maximum biomass growth that is ´ stoichiometrically ´ possible by mere respiration of the carbonaceous organics to CO2 and is moreover SRT
independent. Any experimentally obtained maximum yield higher than the maximum yield
reported in literature for respiration, is an indication that another COD removal mechanism
different from respiration (i.e. storage, adsorption) is contributing towards the COD removal.
The maximum yield attained with the baseline CAS reactor can be seen as a baseline for
the yield obtained solely by respiration. The attained maximum yield of the CAS reactor
(0.53˘0.09 gVSSproduced CODremoved -1 ) is similar to the 0.40-0.50 gVSSproduced CODremoved -1
commonly reported in literature (Strotmann et al., 1999; Sykes, 1975; Cokgor et al., 1998; Dold
and Marais, 1986). Therefore, any reported Y greater than 0.53 gVSSproduced CODremoved -1
will indicate that adsorption of particulate COD and/or storage of rbCOD may be present to
some extent in this particular SBR setup.
The A-stage reactors were operated to test whether there is a difference in yield caused by a
higher occurrence of storage and/or sorption compared to the CAS reactor and whether this
possible difference is amplified at a lower SRT. The results however are primarily against the hypothesis because the A-stage attained a maximum yield which is lower (0.42˘0.02 gVSSproduced
CODremoved -1 at SRT = 0.41 days and 0.34˘0.03 gVSSproduced CODremoved -1 at 1.39 days) than
the baseline reactors. Thus storage and/or sorption are most likely not present as a COD removal mechanism. The full scale A-stage reactors in The Netherlands do exhibit a larger
than usual maximum yield (0.46-0.86 gVSSproduced CODremoved -1 ) implying that storage and
adsorption might be a main driver for COD removal in those reactors (de Graaff and Roest,
2012). One can therefore assume that storage and/or sorption may happen in an A-stage stage
reactor. The reason why it wasn’t observed in these particular lab-scale reactores and why the
maximal yield is even (considerably) lower is multidimensional.
One can question the method used to calculate the yield. The biomass yield is calculated by
Yield & Recoverable COD
83
dividing the cumulative sludge balance by the cumulative substrate balans. To be able to apply
this method of estimation, some assumptions have to made, which might bias the result of the
estimation:
All effluent SS is considered sludge and therefore contribute towards the sludge balance. This
assumption implies that there is always a biomass yield even when no biomass is present in
the reactor. For example, if a reactor with no biomass is fed with only particulate COD, the
effluent would contain those particulates. Applying the first assumption, those particulates will
be considered sludge instead of substrate. The sludge balance will therefore be positive and
there will be an apparent COD removal as the substrate balance is based on effluent soluble
COD. As a result a ’fictitious’ biomass yield is created and therefore decreasing the accuracy of
the estimation. This assumption has an effect on all reactors and does imply that all yields are
overestimated as the effluent VSS will also be to some degree particulate substrate. However,
the CAS reactor runs at a higher HRT than the A-stage reactor. Hydrolysis of particulates
and subsequent removal will presumably be more pronounced.
The amount of COD removed is measured for only one cycle is measured and is then assumed
to be constant until the next measurement. Whereas the difference in biomass concentration in
the reactor can be estimated with good accuracy on a daily basis, the COD removal efficiency
can be relatively dynamic throughout the difference cycles. This assumption however, may
imply why the maximum yield A-stage reactor is lower than the CAS reactor. HRAS systems
are far more dynamic due to their low HRT and SRT, which can lead to fluctuating COD
removal. For example, the A-stage reactor that operated at a SRT of 0.41˘0.06 days had more
than two complete sludge turnovers between the COD sampling points. Figure 13.1 shows
the daily yield of all reactors calculated with equation 9.1 and 9.2. The daily yields obtained
from the CAS and CS reactor are more stable and show nearly no deviation from the mean,
while the high-rate reactors show a big variability in daily observed yield. This implies that
the precision with which the yield of the reference reactors is calculated is higher than that of
the A-stage reactors. This difference in precision is not observable in the standard deviation,
as the yield is calculated in a cumulative way with linear regression
The sampling method for COD analysis was different for the A-stage reactor in comparison
with the other reactors. The influent sample for COD analysis was sampled directly from the
influent barrel, whereas the influent of the other reactors was directly sampled from the exit
of influent tubing in the reactor. Not all particulates in the influent barrel might reach the
reactor (due to sedimentation and/or adhesion to the tubing wall). The COD concentration
entering the A-stage reactor might therefore be lower than estimated, resulting in an apparent
84
Reactors
Daily Yield
gVSSproduced gCODconsumed
2
1.5
1
0.5
0
−0.5
CAS
CS
0
5
10
15
2
1.5
1
0.5
0
−0.5
2
1.5
1
0.5
0
−0.5
20
A-1
A-2
HiCS-1
HiCS-2
0
5
10
15
Days (d)
Figure 13.1: Calculated daily observed yield of all lab-scale reactors that were operated. The variation
observed in the daily observed yield is much smaller than the ones attained in the highrate reactors. CAS (IJ); CS (); A-1 (‚), A-2 (˛); HiCS-1 (İ) and HiCS-2 (ˆ).
COD removal which lowers the yield.
Perhaps a more accurate and precise way to estimate the observed yield is trough a combination
of continuous respirometry and a rigorous COD sampling campaign. The observed yield (in
gCOD gCOD-1 ) can be estimated using the ratio of the oxygen uptake (OU) and CODdissolved
removal (S) (Eq 13.2) (Majone et al., 1999).
Yobs “ 1 ´
OU
S
(13.2)
A major downside of this approach would be that it is only viable with simple rbCOD and
can only account respiration and storage into the estimation. If one could estimate the rate
of CODparticulate hydrolysis to CODdissolved (khydro , h-1 ) and know the initial CODparticulate
Yield & Recoverable COD
85
concentration (Xs,i ) and total CODparticulate removal rate, a modification could be made to
make the estimation viable for complex wastewater and be able to integrate sorption into the
observed yield calculation. However, this approach does allow a separation between a storage
and sorption yield. Furthermore, this type of yield determination will be very hard to integrate
in a continuous system as COD cannot be measured on-line.
The A-stage reactor running at a lower SRT achieved a higher maximal yield than the one
operating at a higher SRT, which indicates that a lower SRT might triggers a more noticeable
adsorption and storage effect. However, as the maximum yield is still lower than the CAS
reactor, this is highly unlikely. Still, in theory, the lower the SRT, the higher the microbial
turnover, which in its turn induces a more dynamic environment. Furthermore the low biomass
concentration in the reactor increases the specific loading rate. The combination of the elevated
dynamic environment and the high specific loading rate most likely stimulates the bacteria to
go more towards a storage/adsorption response compared to the A-stage running at higher
SRT (Daigger and Grady, 1982).
The maximum yield attained in the CS reactor is marginally higher (0.65˘0.07 gVSSproduced
CODremoved -1 ) than 0.53 gVSSproduced CODremoved -1 indicating a minor amount of sorption and
storage occurred in the reactor. It demonstrates that contact-stabilization ´ even at low loading
rates ´ is fundamentally a more dynamic process and therefore presumably exploiting sorption
and storage as a COD removal technique. This is in line with idea that contact stabilization
operates on the basis op rapid adsorption of particulates (Ullrich and Smith, 1951; Ramalho,
1997). This observation is the key element that led towards the incentive of designing and
testing a high-rate variant of the CS-process, the HiCS process: (1) whether the same remains
true under high-rate conditions and (2), does the high-rate environment amplify the effect.
HiCS-1, which operated at τ = 1, achieved a maximum yield of 0.61˘0.04 gVSSproduced
CODremoved -1 . This is only marginally larger and moreover well within the error range of
maximum yield attained with the CAS reactor. However, the yield is lower than the CS reactor operating at higher SRT and lower loading rate. This is peculiar as the high load and shorter
cycle time increase the yield as a more dynamic environment is created. HiCS-2 which operated
at τ = 17 , attained the highest maximum yield (0.79˘0.04 gVSSproduced gCODremoved -1 ) of all
reactors that were operated in this study, confirming the hypothesis that the high rate variant
did show an increase in yield and the high-rate environment somewhat amplifies the effect.
This maximum is the same than the only other HiCS-type (operating at an SRT of 6.8 days)
of reactor that was operated in literature (0.69 gVSS gBODapplied , which is approximately 0.79
gVSS gCODremoved ) (Zhao et al., 2000). They did not state what the HRT in the stabilization
86
Reactors
tank was, hence it is impossible to calculate τ . As the HiCS reactors operate on a different
SRTaer , one might suspect that the difference in yield is solely because of the difference in SRT
where the biomass is allowed to grow. However, since the maximum yield values are already
corrected for differences in SRT, there must therefore be another factor explaining the difference in Y between the two reactors. This parameter is most likely τ , as the difference in contact
over stabilization largely determines the dynamics and concentration gradient of the contact
phase. A smaller τ leads to a larger gradient, which leads to a larger storage effect (Daigger
1
and Grady, 1982). The baseline CS reactor operated at τ = 10
, which might be the reason
that the yield is higher than HiCS-1 operated at τ = 1 assuming that the yield increases with
decreasing τ . The latter stresses the paramount importance of τ as a critical design parameter.
A badly chosen τ will most likely operate worse in terms of yield.
13.2.2
Recoverable COD
Using the COD mass balance, one can calculate what percentage of the incoming COD can
be ’harvested’ as sludge trough wasting. As for practical purposes wasting was performed in
the react phase of the SBR cycle (MLSS wasting), this percentage is strongly dependent on
(1) the applied SRT, as a lower SRT implies a higher wasting volume (Eq. 6.2) and (2) the
yield, as the biomass concentration in the reactor will be higher at given SRT (Eq. 13.1). The
COD fraction that was not recovered can be completely assigned to respiration if one assumes
that the balance is 100%. This is however an approximation as no real respirometric data was
acquired to estimate respiration.
The reference reactors did not require any wasting to maintain the SRT on the desired level.
All harvestable COD was therefore lost with the effluent. Due to the relatively small reactor
volume and large VER, even a minor amount of effluent SS heavily influenced the SRT. In
full scale CAS operations however, sludge wasting is required to accurately maintain a desired
SRT. The high-rate reactors did require sludge wasting to maintain their low SRT. HiCS2, which operated at τ = 71 and attained the highest yield of all reactors, also attained the
highest percentage of COD (18%) that is recoverable. This is lower than the full-scale A-stage
at Nieuwveer where 26% of the incoming COD in incorporated into the sludge. However it
is hard to attain a valid comparison between the two due to the shear difference in reactor
volume. The A-stage reactors have lower recovery percentages than the HiCS-2, which is in
line with the lower biomass yield observed in these reactors. A-1 and A-2 have nearly equal
potential to recover COD (12%), which indicates that the higher waste flow rate neutralizes the
lower biomass concentration in A-1 and vice versa. Given an upscaled version of the reactors
where sludge loss with the effluent is less pronounced or an advanced sludge/liquid separation
Yield & Recoverable COD
87
technique (i.e membrane separation) is used, the particulate effluent COD can also potentially
be recovered. HiCS-2 will have a total recovery potential of nearly 50% when all particulate
COD is recovered. Remarkably, the CS-reactor has a higher recovery potential than HiCS1 when including particulate effluent COD. This is another confirmation that τ is a crucial
parameter in terms of yield and subsequent COD recovery.
The percentage of incoming COD that is used for respiration can be used as an indication for
storage and adsorption effects. As expected, the CAS reactor exhibited the largest respiration
as the reactor was running very stable. The high-rate reactors had a smaller percentage of
COD than the CAS reactor. However, as respiration is still higher than 30% (30% of the COD
that is respired is more or less a biological constant (Van Haandel and Van Der Lubbe, 2007)),
no conclusions can be drawn as respiration was not independently measured. HiCS-2 did show
a respiration rate lower than 30%, which do might indicate sorption and/or storage was present
in that reactor.
13.2.3
Conclusion
HiCS-2, which ran at τ = 71 , achieved the highest biomass yield and percentage of recoverable
COD and is therefore, from a pure sludge production point of view, the best choice. The A-stage
reactors did not show any form of increased yield due to sorption and/or storage and the SRT
does not seem to have a profound effect on the maximum yield. The method by which the yield
was estimated is not accurate as all effluent SS is considered sludge. Moreover, the estimation
is also not very precise for highly dynamic systems as the A-stage or HiCS because the COD
removal resolution is low. Despite this shortcoming, the CS and HiCS reactor performed better
than the CAS and A-stage respectively, suggesting that the increased dynamics of the contact
and stabilization phase provide an increased yield, presumably due to sorption and/or storage.
τ seems to be a very important design parameter here as is heavily influences the maximum
yield. First indications lead to the hypothesis that a smaller tau correlates with a higher yield,
1
almost had an equal yield than the high-rate
as the baseline CS reactor operating at τ = 10
HiCS-reactor operating at τ = 1. Further research into the optimization of τ is therefore the
first objective.
88
Reactors
Chapter 14
Respirometry
14.1
Influent characterization
The CAS sludge was able to degrade 74.3% of the influent COD, which is the same percentage
as the experimentally determined BOD/COD ratio. Approximately 99.1% of the estimated
BOD was degraded, confirming that all the biodegradable components were degraded. 48% of
the total OU was sbCOD, which is in range of the 40-60% commonly reported in literature
(Sperandio and Etienne, 2000). In total, 26% of the total COD was considered rbCOD, which
is on the upper side reported in literature (6-25%) for real domestic wastewater (Ekama et al.,
1986; Lesouef et al., 1992; Kappeler and Gujer, 1992). However, as 48% of the COD initially
added was in soluble form and therefore considered rbCOD, 22% should be considered not
biodegradable. The rbCOD fraction of SYNTHES is in the form of acetate (Table 6.1), which
is under normal circumstances 100% biodegradable. Furthermore, acetate has been found to
easily stored instead of immediately respirated (Majone, 1996). This leads to the assumption
that the rbCOD fraction is underestimated.
When only acetate is used as available substrate, storage leads to ’tailing’ of the respirogram
as the substrate is first stored and only later respired (Dircks et al., 1999; Karahan-Gul et al.,
2002). As a consequence, when rbCOD is estimated solely by looking to a sharp decease in the
first derivative may lead to an underestimation of the fraction. When a complex substrate is
used, the tail gets masked and cannot be clearly distinguished from respiration of hydrolyzed
particulates. Between minute 30 and 85, the OUR decrease is more pronounced than in the
later stage of the experiment, suggesting that a storage tail might be present. Assuming
the hydrolysis of particulates follows zero-order kinetics (Dennis and Irvine, 1981; Tsuno et al.,
1978), the presumed storage tail can be visualized (Figure 14.1). Some suggest that hydrolysis
follows first-order kinetics (Henze and Mladenovski, 1991) or surface limiting reactions kinetics
89
90
Respirometry
Oxygen uptake Rate (OUR)
(mgO2/h)
(Dold et al., 1980), which will classify II and IIb as ’rapidly hydrolysable’ COD. A possible way
to determine if II is indeed a storage tail is to run the experiment with only influent filtered
over a 0.45 µm filter so that all COD present is rbCOD.
150
100
50
I II
IIb
III
d(OUR)/dt
0
2000
1000
0
−1000
d(OUR)/dt
−2000
0
1
2
3
4
5
6
7
Time (h)
Figure 14.1: exogenous OUR- and cumulative OU profile of freshly made influent.
(I) OUrbCOD (II) OUstorage tail (IIb & III) OUsbCOD
Regardless the possible underestimation of rbCOD due to the storage tail which wasn’t accounted for, the COD fractions of SYNTHES is representative for real domestic sewage. Although the non-biodegradable fraction is relatively low and the rbCOD is relatively high, none
differ in great amounts from real wastewater.
14.2
Contact and stabilization phase characterization
Respirometry in combination with a COD profile is an excellent tool to characterize how the
contact and stabilization phases behave and potentially differ between the two HiCS configurations. This will give more insight into choosing the optimal τ and check if there are any
physiological differences between the two sludges.
HiCS-1, which operated at a contact time of 20 min (τ = 1), did not reach a steady state COD
concentration as both the COD and SOUR did not achieve a stable value. However, the COD
concentration at the end of both stabilization periods are equal, which might indicate that τ
Contact and stabilization phase characterization
91
= 1 did reach the background value in 20 minutes if one assumes that both sludges have equal
potentials to degrade COD. HiCS-2 achieved a stable value relatively fast, most likely because
the biomass concentration was higher and therefore the substrate-to-biomass ratio is lower.
The attained respirometric yield (based solely on soluble COD, Eq. 13.2) of the cycle of HiCS-1
is similar to the respirometric yields observed for the stabilization and contact phase of the
cycle. The elevated yield in the contact phase indicates that the sludge was presumably still
using storage as a COD removal technique. This is peculiar, as one might expect that the
stabilization phase is a relatively stable phase where no storage is present. The respirometric
yield of the contact stage is marginally higher indicating that storage might be a little bit more
pronounced in the contact phase. Nothing can be said about the possibility of adsorption, as
the respirometric yield is calculated with soluble substrate only. Interestingly, the SOUR of
the rbCOD degradation of HiCS-1 sludge is 3.5 times higher than the SOUR of the sludge
with operated at τ = 71 and even 5 times higher than the CAS sludge treating the same
influent. This suggests that sludge subjected to τ = 1 is can degrade the same amount of
substrate much faster and efficiently. This makes sense, as the bacteria have little time per
cycle to grow and replicate. Together with the very high sludge turnover imposed by the
low SRT, the microbial community might have changed to a fast-metabolizing community as
slow-metabolizing bacteria are not able to cope with these conditions and will therefore have
no chance to grow. However, as the HiCS-2 sludge experienced a short and unforeseen pH
shock to pH 11 the day before the experiment, the lower SOUR might just be the effect of
overestimated active biomass. A large fraction of the biomass might be dead. Characterization
of the sludges’ kinetic parameters (ko and µmax ) in combination with the characterization of
the endogenous respiration might confirm or reject this hypothesis.
The observed yield of the stabilization phase of HiCS-2 is marginally lower than the respirometric yield calculated for the entire cycle. The yield calculated over the short contact phase
is much higher, which indicates that a large fraction of the yield is most likely gained in short
contact phase due to presumably rapid storage and sorption. The dynamics of the contact
phase HiCS-2 are most likely fast (i.e short duration) and strong (big concentration gradient)
enough to trigger a storage response and rapid adsorption. The accumulated PHA and adsorbed colloids are thereafter presumably degraded under stable conditions in the stabilization
phase. However, storage can not be estimated with respirometry in the contact phase due to
the lack of aeration. Other techniques, like polymer formation quantification might be a more
viable option (Majone, 1996).
From a stabilization point of view, τ =
1
7
is the better option, however the very short contact
92
Respirometry
phase makes it more prone to shock loads and possible volumetric fluctuation due to dry
weather flow (DWF) and wet weather flow (WWF). Whereas this thesis only tested the practical
extremes, optimization of τ between 1 and 71 might give a more practical solution towards
upscaling. A combination of an elevated storage respons with a growth respons might give
increased yields and give more consistent effluent quality with variable influent flows.
Chapter 15
Biochemical Methane Potential &
Fermentation
15.1
BMP
Whereas maximizing sludge production is important, the sludge has to be easily and fully
digestible to have maximize biogas production. A biochemical methane potential test quantifies
the amount of biogas that can be produced given a specific load. The A-stage sludge was found
to be 100% digestible (COD/COD-basis) which is higher than the 90-95% for A-sludge from
the full-scale A/B-plant at Breda (personal communication). As all tests were performed with
the same COD load, both A-stage sludges performed equal on a COD basis (as 100% of the
sludge is digested). Being 100% digestible, the non-significant difference observed in Figure
11.5 is solely based on a different VS/COD ratio between the sludges. The CH4 /CO2 ratio
of the biogas was 2.8˘0.1 for A-1 sludge and 2.7˘0.2 for A-2 sludge indicating the biogas
composition was equal. As both sludges are very young and treated with the same influent,
the relation formulated by Bolzonella et al. (2005) concerning the decrease in digestibility with
increasing SRT does not stand when the SRT drops below one day. As the endogenous residue
fraction of young sludge is already very low (VSS/TSS close to 1), the inert fraction of the
sludge is low, which results in a methane yield close to 100% as observed for both sludges.
The BMP test of the HiCS-sludge was performed with another AD-inoculum (granular instead
of floccular), however a recent large inter-laboratory study found no significant effect of the type
of AD-inoculum on final result of the BMP test, however the digestion kinetics are different
for different inocula (Raposo et al., 2011). This is clearly observable in Figure 11.4 where the
granular inoculum has much faster kinetics than the floccular inoculum used. The latter was
very inactive and required a very long HRT before the biogas buildup was stable. The reason
93
94
Biochemical Methane Potential & Fermentation
for this inactivity is unknown.
The HiCS sludge was only 42-44% digestible, which is remarkably low considering that the Astage sludge at same SRT digested 100%. This might have several reasons. (1) the plateau phase
of the biogas buildup only seemingly, meaning that more biogas would have been produced if
the HRT was prolonged. (2) The inoculum does not have the same capacities to fully convert
the substrate COD into biogas. The lack of a positive control (i.e. glucose) does not allow to
check this. (3) The morphology of the HiCS sludge induced by the contact and stabilization
phase regime is very different from the A-stage, which might result in a type of sludge that is
not well convertible into methane. The CH4 /CO2 ratio of the HiCS biogas was higher than that
of the A-stage, indicating that the biogas is relatively more rich in methane. If one could find a
way to increase digestibility, this higher percentage of methane will be an advantage. However
the specific methane production was nearly equal for all test. The different COD/VSS ratio is
the main driver here, as it is much larger for HiCS sludge than for A-sludge. In conclusion, all
high-rate sludges had similar digestibility based on VS.
15.2
Fermentation
Within the scope of the recently developed carboxylate platform, (mesophilic) fermentation
of biomass to short-chain (volatile) fatty acids (SCFA) and further processing in bio-refineries
might be more profitable than methane production (Agler et al., 2011). The overall fermentation efficiency (0.64 gCODSCFA gVSsubstrate ) is high compared to other studies fermenting
primary or secondary sludge (0.20-0.30 gCODSCFA gVSSsubstrate ) (Ahn and Speece, 2006; Zhang
et al., 2009). Note that these values are obtained at different pH and fermentation might be
very sensitive to applied pH (Zhang et al., 2009). The notable difference in SCFA yield can
therefore be the result of the different pH. However, the SRT of the waste activated sludge is
not given, but it is safe to assume the CAS plants run at a much higher SRT than 1 day. The
influence of the sludge’s SRT on fermentation has not been investigated so far, nor is any study
on the fermentation of HRAS sludge been conducted. If the same trend in terms of better
biodegradability with decreasing SRT can be observed with fermentation as it is observed with
digestion, the increased SCFA yield is a logical result of this assumption. No difference between
the two type of HiCS-sludge was observed. Methane yield (0.19-0.35% CODCH4 /CODsubstrate )
was low and undesirable from a fermentation point of view. The chosen pH (7) might be reason for this methane production. When fermenting waste activated sludge, Zhang et al. (2009)
observed the highest methane yield at pH 7. There are strong indications in literature that the
SCFA yield increases with increasing pH (Chen et al., 2007; Yuan et al., 2006), which creates
the incentive to further optimize the protocol of the fermentation test.
Economical aspects
15.3
95
Economical aspects
To check whether fermentation is a viable alternative to anaerobic digestion in terms of costeffectiveness, the total revenue of a year of operation can be calculated. Looking at the reactor
that can attain the highest revenue in terms of sludge production and subsequent sludge processing and assume that sludge washout is fixed so that the effluent quality will increase,
HiCS-2 would be the best option. The calculations will therefore be based on the results of
the HiCS-2 reactor. To create a more realistic view in terms of industrial market price, the
reactor is upscaled to the same volume as the A-stage of WWTP nieuwveer, which is 3500 m3 .
Furthermore, to simplify the calculations, certain assumptions were made:
• The plant will have a constant volumetic loading rate throughout the year. Fluctuations
in terms of COD concentration and DWF/WFF are ignored. The COD recovery efficiency
remains constant throughout the year.
• The cost to operate the digester is not considered here, since this is outside the scope of
the research and estimated to be equal among the different types of sludge. Furthermore,
the calculations are made purely on methane basis which implies that methane extraction
from the biogas is 100%.
• Fermentation revenue is based purely on the acetate yield of the process, other SCFA are
ignored. The extraction of acetate from the matrix is 100% and does not cost anything.
The total combustion energy of CH4 is 21.6 MJ kWh-1 and 1 kWh equals 3.6 MJ. Given
an electrical efficiency of 35%, 2.1 kWh m-3 can be produced. The current market price
of electricity is 0.08 e kWh-1 , which results in a total price of 0.168 e m-3 . The Flemish
government also grants green power certificates (dutch: groene stroom certificaten) at a price
of 90 e 1000kWh-1 . The current market price of 1 ton of acetate can be estimated at 500 e
(personal communication).
Given a sludge methane production of 0.25 m3 CH4 kgCOD-1 and sludge acetate production
of 0.13 kg acetate kgCOD-1 , the total revenue for both sludge treatment pathways can be
estimated (Table 15.1).
Given the simplifications and assumptions above, fermentation is less cost-effective than digestion and subsequent transformation to energy. The combination of a HRAS systems in
combination with advanced fermentation and SCFA extraction as a profitable bio-refinery is at
currently not very feasibly with current efficiencies. However, this is only based on the acetate
production efficiency. Acetate is a much more versatile product, chain elongations are possible
and is currently a hot research topic (Agler et al., 2011). The other SCFAs might also be
96
Biochemical Methane Potential & Fermentation
Table 15.1: Reactor details and corresponding biogas and fermentation revenue.
sludge production
Volume
Vol. Loading Rate
COD production
Recovery efficiency
Waste COD production
3500
8.34
1.08 · 107
18
1.94 · 106
m3
kgCOD m-3 d-1
kgCOD y-1
%
kgCOD y-1
Biogas revenue
Market price
Green power certificates
total revenue
0.168 e m-3 CH4
90 e 1000kWh-1
1.73 · 105 e y-1
Fermentation revenue
Market price
total revenue
0.5 e kg-1 acetate
1.26 · 105 e y-1
extracted and separated electrochemically (Bailly et al., 2001). To put the obtained revenue
in perspective: A conventional activated sludge plant would have a total cost of 2.94 million
euros if it would treat the same amount of IE (Zessner et al., 2010).
Chapter 16
General Conclusion
The general conclusions are briefly summarized below:
COD removal efficiency
• The low-rate reference reactors obtained the highest COD removal efficiencies.
• The A-stage reactor operating at SRTact = 1.39˘0.83 days achieved the best COD removal efficiency of the high-rate reactors.
• The HiCS reactors achieved overall low removal efficiencies but the difference between
the two was not significant. The COD removal efficiencies may therefore be independent
of the contact time over stabilization time ratio τ .
Yield
• The method used to calculate the yield might not be suited to very accurately estimate
the yield due to the two assumptions made. Furthermore, the precision of the estimation decreases when calculated for high-rate reactors, as they are very dynamic in COD
removal efficiency.
• The A-stage reactor did not achieve a higher yield than the baseline reactor, indicating
that, under these conditions, no storage and/or storage effects are present as a method
to remove COD.
• The HiCS reactor did achieve a higher yield, which might indicate storage and sorption.
τ has been found to profoundly influence the yield, which is believed to increase with
decreasing τ .
97
98
General Conclusion
• For the HiCS reactor operating at τ = 1, there is still an elevated respirometric yield
observed in the stabilization phase. This is unwanted, as a mineralization response in
the stabilization phase is required to optimally remove COD in the contact phase with
storage and sorption. The latter is observed in the HiCS reactor operating at τ = 17 ,
explaining the large reactor yield.
BMP & fermentation
• The A-sludge was found to be 100% converted into CH4 , while the HiCS sludge only
converted for 42%. There was however no difference in specific methane production on a
VS basis.
• The HiCS sludge had a specific SCFA production of 0.64 gCODSCFA gVSsubstrate , which
is a higher than any fermentation test on secondary sludge in literature. The very low
age of the sludge might be the reason increased fermentability.
To conclude, the HiCS reactor operating at τ = 71 had the most potential in this study, whereas
no indication of sorption and or storage has been observed for the A-stage nor any profound
influence of the SRT on the yield. It had the highest biomass yield and there is room for
COD removal optimization. A HiCS-type of reactor as an advanced HRAS system is therefore
certainly plausible. Further research has be conducted to find the optimal τ and improve the
overall COD removal efficiency.
Chapter 17
Future Research
For future research on this topic, some suggestion are made:
• Test other operational parameters on the A-stage to investigate which one can induce
a higher yield and therefore presumable storage and/or adsorption. HRT would be the
first parameter to look at, as the storage response of the sludge is mainly based on highly
dynamic conditions, which a small HRT might deliver. Next, DO might be an interesting
parameter as under oxygen stress, the microorganisms might show an increased storage
response.
• Optimize τ to achieve the largest yield that might be possible in a HiCS configuration.
A smaller τ presumably leads to a higher yield.
• Look into other crucial parameter of the HiCS process, like HRT, that might improve the
COD removal efficiency of the system.
• Improve the method to estimate the reactor biomass yield. This may be achieved by
minimizing the potential error made by the assumptions (i.e separate particulate substrate
from sludge in the effluent and measure COD more often) or by applying respirometry
as a technique to estimate the yield.
• Optimize the fermentation protocol. Fermenting at a higher pH might improve fermentation yields even more. Also test the fermentation capabilities of the A-stage sludge and
check if there is an observable difference in SCFA production.
99
100
Future Research
Bibliography
Agler, M.T., Wrenn, B.A., Zinder, S.H., Angenent, L.T., 2011. Waste to bioproduct conversion with undefined mixed cultures: the carboxylate platform. Trends Biotechnol 29, 70–8.
doi:10.1016/j.tibtech.2010.11.006.
Agudelo-Vera, C.M., Leduc, W.R., Mels, A.R., Rijnaarts, H.H., 2012. Harvesting urban resources towards more resilient cities. Resources, Conservation and Recycling 64, 3–12.
Ahn, Y.H., Speece, R.E., 2006. Elutriated acid fermentation of municipal primary sludge.
Water Res 40, 2210–20. doi:10.1016/j.watres.2006.03.022.
Aiyuk, S., Verstraete, W., 2004. Sedimentological evolution in an uasb treating synthes, a
new representative synthetic sewage, at low loading rates. Bioresour Technol 93, 269–78.
doi:10.1016/j.biortech.2003.11.006.
Aksu, Z., 2005. Application of biosorption for the removal of organic pollutants: a review.
Process Biochemistry 40, 997–1026.
Alcamo, J., Flörke, M., Märker, M., 2007. Future long-term changes in global water resources
driven by socio-economic and climatic changes. Hydrological Sciences 52, 247–275.
public health association (APHA), A., 1992. Standard methods of water and wastewater. 18th
edition ed., American public health association and American water works association and
Water Environment Federation.
Aquafin, . Persoonlijke communicatie.
Arnell, N.W., 2004. Climate change and global water resources: Sres emissions and socioeconomic scenarios. Global Environmental Change 14, 31–52.
Bailly, M., Roux-de Balmann, H., Aimar, P., Lutin, F., Cheryan, M., 2001. Production processes of fermented organic acids targeted around membrane operations: design of the concentration step by conventional electrodialysis. Journal of Membrane Science 191, 129–142.
101
102
Bibliography
Boehnke, B., 1977. Das adsorption-belebungsverfahren. Korrespondenz Abwasser 24, 33–58.
Boehnke, B., Schulze-Rettmer, R., Zuckut, S.W., 1998. Cost-effective reduction of highstrength wastewater by adsorption-based activated sludge technology. Water Engineering
and Management 145, 31–34.
Bolzonella, D., Pavan, P., Battistoni, P., Cecchi, F., 2005. Mesophilic anaerobic digestion of
waste activated sludge: influence of the solid retention time in the wastewater treatment
process. Process Biochemistry 40, 1453–1460.
Bunch, B., Griffin, D.M., 1987. Rapid removal of colloidal substrate from domestic wastewaters.
Water Polution Control federation 59, 957–963.
Cantunda, S.Y.C., Deep, G.S., Haandel, A.V., Freire, R.C.S., 1999. Feedback control method
for estimating the oxygen uptake rate in activated sludge systems. IEEE TRANSACTIONS
ON INSTRUMENTATION AND MEASUREMENT 48, 864–869.
Cardoen, D., 2011. Up-concentration techniques for zero-waste water treatment. Master’s
thesis. Universiteit Gent.
Carlos, D.M., Filipe, C., Grady, L., 1999. Biological Wastewater Treatment. 2nd edition ed.,
Marcel Dekker.
Cech, J.S., Chudoba, J., 1982. Influenve of accumulation capacity of activated sludge microorganisms on kinetics of glucose removal. Water Research 17, 659–666.
Chao, A.C., 1979. Influence of process loading intensity on sludge clarification and thickening
characteristics. Water Research 13, 1213–1223.
Chen, Y., Jiang, S., Yuan, H., Zhou, Q., Gu, G., 2007. Hydrolysis and acidification of waste
activated sludge at different phs. Water Res 41, 683–9. doi:10.1016/j.watres.2006.07.030.
Chua, A.S.M., Takabatake, H., Satoh, H., Mino, T., 2003. Production of polyhydroxyalkanoates (pha) by activated sludge treating municipal wastewater: effect of ph, sludge
retention time (srt), and acetate concentration in influent. Water Res 37, 3602–11.
doi:10.1016/S0043-1354(03)00252-5.
Clara, M., Kreuzinger, N., Strenn, B., Gans, O., Kroiss, H., 2005. The solids retention time-a
suitable design parameter to evaluate the capacity of wastewater treatment plants to remove
micropollutants. Water Res 39, 97–106. doi:10.1016/j.watres.2004.08.036.
Bibliography
103
Cokgor, E.U., Sozen, S., Orhon, D., Henze, M., 1998. respirometric analysis of activated sludge
behaviour i. assessment of the readily biodegradable substrate. Water Research 32, 461–475.
Colliver, B., Stephenson, T., 2000. Production of nitrogen oxide and dinitrogen oxide by
autotrophic nitrifiers. Biotechnology Advances 18, 219–232.
Daigger, G.T., Grady, C.L., 1982. The dynamics of microbial growth on soluble substrates.
Water Research 16, 365–382.
Dauphin, S., Joannis, C., Deguin, A., Bridoux, G., Ruban, G., Aumond, M., 1998. Influent
flow control to increase the pollution load treated during rainy periods. Water Science and
Technology 37, 131–139.
Dennis, R.W., Irvine, R.L., 1981. a stoichiometric model of bacterial growth. Water Research
15, 1363–1373.
Dircks, K., Pind, P.F., Mosbak, H., Henze, M., 1999. Yield determination by respirometry the possible influence of storage under aerobic conditions in activated sludge. Water SA 25,
69–74.
Dold, P.L., Ekama, G.A., Marais, G.V.R., 1980. the activated sludge process 1. a general model
for the activated sludge process. Progress in water technology 12, 47–77.
Dold, P.L., Marais, G.V.R., 1986. evaluation of the general activated sludge model proposed
by the iawprc task group. Water Science and Technology 18, 63–89.
Dosta, J., Fernández, I., Vázquez-Padı́n, J.R., Mosquera-Corral, A., Campos, J.L., MataAlvarez, J., Méndez, R., 2008. Short- and long-term effects of temperature on the anammox
process. J Hazard Mater 154, 688–93. doi:10.1016/j.jhazmat.2007.10.082.
Ekama, G.A., 2008. Biological Wastewater Treatment: Principles, Modelling and Design. IWA
publishing.
Ekama, G.A., 2009. The role and control of sludge age in biological nutrient removal activated
sludge systems. Water Science and Technology .
Ekama, G.A., Dold, P.L., v. R. Marais, G., 1986. procedures for determining influent cod
fractions and the maximum specific growth rate of heterotrophs in activated sludge systems.
Water Science and Technology 18, 91–114.
Fux, C., Siegrist, H., 2004. Nitrogen removal from sludge digester liquids by nitrification/denitrification or partial nitration/anammox. Water Science and Technology 50, 19–26.
104
Bibliography
Gleick, P.H., 1993. Water in Crisis: A Guide to the World’s Fresh Water Resources. Oxford
University Press.
Goodwin, J.A.S., Forster, C.F., 1985. A further examination into the composition of activated
sludge surfaces in relation to their settlement characteristics. Water Research 19, 527–533.
de Graaff, M., Roest, K., 2012. Inventarisatie van AB-systemen - optimale procescondities in
de A-trap. Technical Report. STOWA.
Guellil, A., Thomas, F., Block1, J., l Bersillon, Ginestet, P., 2001. Transfer of organic matter
between wastewater and activated sludge flocs. Water Research 35, 143–150.
Gujer, W., Jenkins, D., 1975. The contact stabilization process - oxygen utilization, sludge
production and efficiency. Water Research 9, 553–560.
Haider, S., Svardal, K., Vanrollenghem, P., Kroiss, H., 2003. The effect of low sludge age on
wastewater fractionation (ss, si). Water Science and Technology 47, 203–209.
Haider, S., Vanrollenghem, P., Kross, H., 2000. Low sludge age and its consequences for
metabolisation, storage and adsorption of readily biodegradable substrate, in: Proceedings
1st World Congress of the International Water Association.
Henze, M., Mladenovski, C., 1991. hydrolysis of particulate substrate by activated sludge under
aerobic, anoxic and anaerobic conditions. Water Research 25, 61–64.
Huang, J.C., Li, L., 2000. An innovative approach to maximize primary treatment performance.
Water Science and Technology 42, 209–222.
Hulle, S.W.V., Vandeweyer, H.J., Meesschaert, B.D., vanrollenghem, P.A., Dejans, P., Dumoulin, A., 2010. Engineering aspects and practical application of autotrophic nitrogen
removal from nitrogen rich streams. Chemical Engineering Journal 162, 1–20.
Jonasson, M., 2007. Energy Benchmark for Wastewater treatment processes - a comparison
between Sweden and Austria. Master’s thesis. Lund University.
Kappeler, J., Gujer, W., 1992. estimation of kinetic parameters of heterotrophic biomass under
aerobic conditions and characterization of wastewater for activated sludge modelling. Water
Science and Technology 25, 125–139.
Karahan-Gul, O., Artan, N., Orhon, D., Henze, M., van Loosdrecht, M., 2002. Respirometric
assessment of storage yield for different substrates. Water Science and Technology 46, 345–
352.
Bibliography
105
Kayser, R., 2005. Chapter 3: Activated sludge Process. In Environmental Biotechnology:
Concepts and Applications. Wiley.
Lantz, M., Svensson, M., Bjornsson, L., Borjesson, P., 2007. The prospects for an expansion of
biogas systems in sweden – incentives, barriers and potentials. Energy Policy 35, 1830–1843.
Lesouef, A., Payraudeau, M., Rogalla, F., Kleiber, B., 1992. optimizing nitrogen removal
reactor configurations by on site calibration of the iawprc activated sludge model. Water
Science and Technology 25, 105–123.
Li, X.Y., Yang, S., 2007. Influence of loosely bound extracellular polymeric substances (eps)
on the flocculation, sedimentation and dewaterability of activated sludge. Water Research
41, 1022–1030.
Liao, B.Q., Allen, D.G., Droppo, I.G., Leppard, G.G., Liss, S.N., 2001. Surface properties of
sludge and their role in bioflocculation and settleability. Water Res 35, 339–50.
Liu, S.G., Ni, B.J., Wei, L., Tang, Y., Yu, H.Q., 2009. Contact-adsorption-regenerationstabilization process for the treatment of municipal wastewater. Journal of Water and Environment Technology 7, 83–89.
Majone, M., 1996. Influence of storage on kinetic selection to control aerobic filamentous
bulking. Water Science and Technology 34, 223–232.
Majone, M., Dircks, K., Beun, J.J., 1999. Aerobic storage under dynamic conditions in activated sludge processes. state of the art. Water Science and Technology 39, 61–73.
Maribo, P., 2009. Chapter 6: biological purification processes. In Wastewater treatment: A
practical guide for design and dimensioning of plants for wastewater treatment-Compendium
for BTWWTP. Technical Report. The University College of Aarhus.
Martins, A.M.P., Heijnen, J.J., van Loosdrecht, M.C.M., 2003. Effect of dissolved oxygen
concentration on sludge settleability. Appl Microbiol Biotechnol 62, 586–93. doi:10.1007/
s00253-003-1384-6.
Martins, A.M.P., Pagilla, K., Heijnen, J.J., van Loosdrecht, M.C.M., 2004. Filamentous bulking
sludge–a critical review. Water Res 38, 793–817. doi:10.1016/j.watres.2003.11.005.
Metcalf, Eddy, 2003. Wastewater Engineering: Treatment and Reuse. 4th ed., McGraw-Hill
Education.
106
Bibliography
Müller, E., Kobel, B., 2004. Energetische bestandsaufnahme an klärlanlagen in nordrheinwestfalen mit 30 millionen einwohnerwerten – energie-benchmarking und sparpotenziale.
Korrespondenz Abwasser 51, 625–631.
PaDEP, . Module 17: The Activated Sludge Process Part III. Pennsylvania Department of
Environmental Protection.
Palm, J.C., Jenkins, D., Parker, D.S., 1980. Relationship between organic loading, dissolved
oxygen concentration and sludge settleability in the completely-mixed activated sludge process. Water Polution Control federation 52, 2484–2506.
Pernelle, J.J., Gaval, G., Cotteux, E., Duchène, P., 2001. Influence of transient substrate
overloads on the proliferation of filamentous bacterial populations in an activated sludge
pilot plant. Water Res 35, 129–34.
Ramalho, R.S., 1997. Introduction to Wastewater Treatment Processes. Academic Press, Inc.
Raposo, F., Fernandez-Cegri, V., la Rubia, M.D., Borja, R., Beline, F., Cavinato, C., Demirer,
G., Fernandez, B., Fernandez-Polanco, M., Frigon, J., Ganesh, R., Kaparaju, P., Koubova,
J., Mendez, R., Menin, G., Peene, A., Scherer, P., Torrijos, M., Uellendahl, H., Wierinck,
I., de Wilde, V., 2011. Biochemical methane potential (bmp) of solid organic substrates:
evaluation of anaerobic biodegradability using data from an international interlaboratory
study. J. Chem Technol Biotechnol 86, 1088–1098.
Rittman, B.E., McCarty, P.L., 2011. Environmental Biotechnology: Principles and Applications. McGRAW-HILL.
Salomé, A.A., 1990. AB-systemen: Een invertarisatie. Technical Report. STOWA.
Sarria, N.V., Victoria, J.R., Lozada, P.T., Parra, C.M., 2011. Performance of a contact stabilization process for domestic wastewater treatment of cali, colombia. Dyna 168, 98–107.
Sayigh, B.A., Malina, J.F., 1978. Temperature effects on the activated sludge process. Water
Pollution Control Federation 50, 678–687.
Sesay, M.L., Ozcengiz, G., Dilek Sanin, F., 2006. Enzymatic extraction of activated sludge
extracellular polymers and implications on bioflocculation. Water Res 40, 1359–66. doi:10.
1016/j.watres.2006.01.045.
Sheng, G., Yu, H., b, X.L., 2010. Extracellular polymeric substances (eps) of microbial aggregates in biological wastewater treatment systems: A review. Biotechnology Advances 28,
882–894.
Bibliography
107
Shizas, I., Bagley, D.M., 2004. Experimental determination of energy content of unknown
organics in municipal wastewater streams. J. energy engineering 130, 45–53.
Slonczewksi, J.L., Foster, J.W., 2009. Microbiology: An evolving science. 2nd ed., W.W.
Norton and Company.
Spanjers, H., Vanrolleghem, P.A., 1995. Respirometry as a rapid tool for characterization of
wastewater and activated sludg. Water Science and Technology 31, 105–114.
Sperandio, M., Etienne, P., 2000. estimation of wastewater biodegradable cod fractions by
combining respirometric experiments in various so/xo ratios. Water Research 34, 1233–146.
Strotmann, U.J., Geldern, A., Kuhn, A., Gendig, C., Klein, S., 1999. evaluation of a respirometric test method to determine the heterotrophicyield coefficient of activatedsludge bacteria.
Chemosphere 38, 3555–3570.
Sykes, R.M., 1975. Theoretical heterotrophic yields. Water Pollution Control Federation 47,
591–600.
Tan, K.N., Chua, H., 1997. Cod adsorption capacity of the activated sludge - its determination
and application in the activated sludge process. Environmental Monitoring and Assessment
44, 211–217.
Tsagarakis, K., Mara, D., Angelakis, A., 2003. Application of cost criteria for selection of
municipal wastewa- ter treatment systems. Water Air Soil and Pollution 142, 187–210.
Tsuno, H., Goda, T., Somiya, I., 1978. kinetic model of activated sludge metabolism and its
application to the response of qualitative shock load. Water Research 12, 513–522.
Ullrich, A.H., Smith, M.W., 1951. The biosorption process of sewage and waste treatment.
Sewage and Industrial Wastes 23, 1248–1253.
Urbain, V., Block, J.C., Manem, J., 1993. Bioflocculation in activated sludge: an analytic
approach. Water Research 27, 829–838.
Van Haandel, A., Van Der Lubbe, J., 2007. Handbook Biological Waste Water Treatment:
Design and optimisation of activated sludge systems. Uitgeverij Quist.
Verstraete, W., de Caveye, P.V., Diamantis, V., 2009. Maximum use of resources present in
domestic “used water”. Bioresource Technology 100, 5537–5545.
108
Bibliography
Verstraete, W., Vlaeminck, S.E., 2011. Zerowastewater: short-cycling of wastewater resources
for sustainable cities of the future. International Journal of Sustainable Development and
World Ecology 18, 253–264.
Vijayaraghavan, K., Yun, Y.S., 2008. Bacterial biosorbents and biosorption. Biotechnology
Advances 26, 266–291.
Wett, B., Buchauer, K., Fimmi, C., 2007. Energy self-sufficiency as a feasible concept for
wastewater treatment systems, in: Proceedings of the IWA Leading Edge Technology Conference.
Yuan, H., Chen, Y., Zhang, H., Jian, S., Zhou, Q., Gu, G., 2006. Improved bioproduction of
short-chain fatty acids (scfas) from excess sludge under alkaline conditions. Environmental
Science and Technology 40, 2025–2029.
Zessner, M., Lampert, C., Kroiss, H., Lindtner, S., 2010. Cost comparison of wastewater in
danubian countries. Water Science and Technology 62, 223–230.
Zhang, P., Chen, Y., Zhou, Q., 2009. Waste activated sludge hydrolysis and short-chain fatty
acids accumulation under mesophilic and thermophilic conditions: effect of ph. Water Res
43, 3735–42. doi:10.1016/j.watres.2009.05.036.
Zhang, Y., Allen, D.G., 2008. The effect of short-term dissolved oxygen transients on activated
sludge. Water Quality Research Journal of Canada 43, 93–102.
Zhao, W., Ting, Y., Chen, J., Xing, C., Shi, S., 2000. Advanced primary treatment of ww
using bio-flocculation-adsorption sedimentation. Acta Biotechnologica 20, 53–64.