No.

B 2425
September 2021

Long term trials with membrane bioreactor for

enhanced wastewater treatment coupled with
compact sludge treatment
-pilot Henriksdal 2040, results from 2020

Sofia Lovisa Andersson, Klara Westling, Sofia Andersson, Jesper Karlsson, Mayumi Narongin,
Andrea Carranza Munoz, Niclas Bornold, Christian Baresel

In cooperation with
Author: Sofia Lovisa Andersson, Klara Westling, Sofia Andersson, Jesper Karlsson, Mayumi Narongin,
Andrea Carranza Munoz, Niclas Bornold, Christian Baresel
Funded by: Stiftelsen IVL and Stockholm Vatten och Avfall
Photographer: Sofia Lovisa Andersson
Report number B 2425
ISBN 978-91-7883-305-4
Edition Only available as PDF for individual printing

© IVL Swedish Environmental Research Institute 2021

IVL Swedish Environmental Research Institute Ltd.
P.O Box 210 60, S-100 31 Stockholm, Sweden
Phone +46-(0)10-7886500 // www.ivl.se

This report has been reviewed and approved in accordance with IVL's audited and approved
management system.
Preface
This report presents work performed during 2020, within the long-term pilot study trials of municipal
wastewater treatment with Membrane Bioreactors (MBR) and sludge treatment including thermophilic and
mesophilic digestion. The study is carried out in cooperation between IVL Swedish Environmental Research
Institute and Stockholm Vatten och Avfall (Stockholm Water and Waste Company). The trials are performed
at the R&D pilot facility Hammarby Sjöstadsverk in Stockholm, Sweden and they are jointly financed by the
IVL foundation and Stockholm Vatten och Avfall.

Previous results from the project are presented in Swedish in Samuelsson et al. (2014), Westling et al. (2016)
and Andersson et al. (2017) for project year 1, 2 and 3, respectively. For project year 4, 5 and 6 the reports are
in English, see Andersson et al. (2019; 2020; 2021).
Table of contents
Summary ................................................................................................................................ 6

Sammanfattning..................................................................................................................... 8

Terminology ......................................................................................................................... 10

1 Introduction .................................................................................................................. 12

2 Background ................................................................................................................... 12

3 Description of the pilot plant........................................................................................ 14

3.1 Process description water line......................................................................................................................... 15
3.1.1 Incoming wastewater ............................................................................................................................. 15
3.1.2 Pre-treatment ........................................................................................................................................ 16
3.1.3 Biological treatment ............................................................................................................................... 17
3.1.4 Membrane tanks .................................................................................................................................... 18
3.2 Process description sludge treatment ............................................................................................................. 19
3.2.1 Thickening .............................................................................................................................................. 19
3.2.2 Digestion ................................................................................................................................................ 19
3.2.3 Dewatering ............................................................................................................................................. 20
3.3 Flow rate and load ........................................................................................................................................... 21
3.4 Chemicals ........................................................................................................................................................ 23
3.4.1 External carbon source .......................................................................................................................... 23
3.4.2 Precipitation chemicals .......................................................................................................................... 24
3.4.3 Chemicals for membrane cleaning ......................................................................................................... 24
3.4.4 Polymers ................................................................................................................................................ 25
3.5 Control system................................................................................................................................................. 25

4 Experimental plan year 2020 ........................................................................................ 26

5 Method ......................................................................................................................... 28
5.1 Sampling and analyses..................................................................................................................................... 28
5.2 Online measurements ..................................................................................................................................... 30
5.3 Evaluation parameters .................................................................................................................................... 31
5.3.1 Membrane performance ........................................................................................................................ 31
5.3.2 Sludge quality ......................................................................................................................................... 32
5.3.3 Anaerobic digestion ............................................................................................................................... 32

6 Results and discussion .................................................................................................. 33

6.1 Primary treatment ........................................................................................................................................... 33
6.1.1 Inlet screen............................................................................................................................................. 33
6.1.2 Efficiency of primary settler ................................................................................................................... 34
6.1.3 Screen and sieve – effect on trash content ............................................................................................ 35
6.1.4 Pre-treated wastewater ......................................................................................................................... 35
6.2 Nitrogen removal............................................................................................................................................. 36
6.2.1 Nitrification ............................................................................................................................................ 38
6.2.2 Denitrification ........................................................................................................................................ 41
 6.2.3 Comparison of different external carbon sources.................................................................................. 43
6.3 Phosphorus removal ........................................................................................................................................ 46
6.4 BOD reduction ................................................................................................................................................. 53
6.5 Membrane performance ................................................................................................................................. 54
6.5.1 Permeability ........................................................................................................................................... 54
6.5.2 Flux and TMP .......................................................................................................................................... 55
6.5.3 Membrane cleaning ............................................................................................................................... 57
6.5.4 Membrane aeration ............................................................................................................................... 61
6.5.5 cTOC and membrane performance ........................................................................................................ 62
6.6 Offgas measurements during NaOCl RC .......................................................................................................... 65
6.7 Sludge production and sludge properties........................................................................................................ 70
6.8 Sludge pilot ...................................................................................................................................................... 74
6.8.1 Feed characteristics ............................................................................................................................... 74
6.8.2 How low can we go – background and set-up ....................................................................................... 77
6.8.3 Results thermophilic trial ....................................................................................................................... 80
6.8.4 Results mesophilic trial .......................................................................................................................... 85
6.8.5 Comparison ............................................................................................................................................ 89
6.8.6 Method comparison for VFA and CH4 production.................................................................................. 92
6.9 Resource consumption .................................................................................................................................... 93

7 Related publications ..................................................................................................... 95

8 Bibliography .................................................................................................................. 96
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Summary
Henriksdal wastewater treatment plant (WWTP) in Stockholm is currently being extended and rebuilt for
increased capacity (from 0.8 to 1.6 million PE) and enhanced treatment efficiency (6 mg TN/L, 0.20 mg TP/L, 5
mg BOD7/L). The reconstruction includes retrofitting of the existing conventional activated sludge (CAS)
tanks with a new membrane bioreactor (MBR) process containing 1.6 million m2 of membrane area. It also
includes extended pretreatment and a new treatment step for thickening of primary sludge. Digestion of thick
sludge (~6% TS) will be done at thermophilic conditions, unlike today’s mesophilic operation, with high
organic load and relatively short retention time.

To increase the knowledge of MBRs in Nordic conditions, Stockholm Vatten och Avfall (SVOA) and IVL
Swedish Environmental Research Institute have conducted long-term MBR studies in pilot scale at the R&D
facility Hammarby Sjöstadsverk, located on the premises of the Henriksdal WWTP. The MBR-pilot was taken
into operation in 2014 and was reconstructed to its current configuration in 2016. In 2017 the MBR pilot was
supplemented with a sludge treatment line in order to study different aspects of sludge digestion.

During 2020, the MBR-pilot was operated at a fixed inflow of 4.4 m3/h, which is 37% higher than the design
average flow, with glycerol as external carbon source for post-denitrification, and aluminum (PAX) instead of
Ferric (PIX) as complement to Ferrous (FeSO4) for phosphorous precipitation. This was done to test the
operational strategy for the first MBR-line in Henriksdal WWTP. The average effluent concentration of
nitrogen and phosphorus was 3.9 mg TN/L and 0.10 mg TP/L, respectively, which means that the effluent
requirements were met also this year. To achieve this, 13.7 g Fe2+/m3, 0.5 g Al3+/m3 and a glycerol dose
equivalent to 16.7 g COD/m3 wastewater was required. The low consumption of phosphorus precipitation
chemicals, 0.95 mole metal per mole of phosphorus removed, was mainly due to a high enhanced biological
phosphorus removal (EBPR) activity. In 2020 the phosphorous release rates (lab test which indirectly
quantifies EBPR activity) went up to above 10 g PO4-P/kg VSS,h, which is higher than all previous years and
indicate very good EBPR-activity. The iron and aluminum content in the activated sludge was 4.1 and 0.5%,
respectively. Two master thesis projects on EBPR were conducted to learn more about the process. Results
showed that most P-release took place in the pre-denitrification zone in the presence of nitrate. During spring,
ethanol was used instead of glycerol to see if EBPR was boosted, but no effect on the EBPR was observed.

The consumption of glycerol (external carbon source) was similar to values obtained for methanol in terms of
COD-dose (g/m3 ww), which was also confirmed by denitrification rate lab tests. This indicates that glycerol
can be a good temporary carbon source at Henriksdal WWTP until the methanol tanks are constructed.
However, the glycerol dosed is not a pure product and potential long-term effects on the concentration of
colloidal total organic carbon (cTOC) and indirectly membrane performance has to be evaluated further.
Ethanol had a higher denitrification rate (3.1-4.7 mg NO3-N/g VSS, h) compared to glycerol (2.4 mg NO3-N/g
VSS, h) and a lower specific COD consumption; 1.7-4.4 g COD/g NO3-Ndenitrified compared to 6.6 g COD/g NO3-
Ndenitrifed for glycerol in the lab tests. Nevertheless, when dosed in the pilot the consumption was almost
identical to the glycerol dose, 16.2 g COD/m3 while the specific dose was higher than for glycerol (2.9 g COD/g
NO3-Ndenitrified for ethanol and 1.8 g COD/g NO3-Ndenitrified for glycerol).

Average total sludge age during 2020 was 17.5 days and average aerated sludge age was 6.5 days.
Nitrification was complete with ammonia concentrations below 2 mg/L at all times except week 11, week 36
and week 50-51. Both week 11 and 38 the aeration was deliberately reduced to minimize foaming. In week 50-
51 the nitrification was inhibited by hypochlorite in combination with cold temperature after recover cleaning
was performed.

Like previous years, the membranes in membrane tank 1 (MT1) was cleaned with oxalic acid and the
membranes in MT2 with citric acid. Both membranes were also cleaned with hypochlorite. The membranes
6
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

were operated with an annular average net flux of approximately 22 lmh at an average TMP of 79 mbar. The
resulting permeability was 325 lmh/bar. In 2019, the cleaning scheme for maintenance cleaning (MC) with acid
was optimized to reach a minimum consumption of 50% of the recommended cleaning scheme for citric acid
and 25% for oxalic acid. In 2020, focus was put on reducing the use of sodium hypochlorite, which is
important since the hypochlorite supposedly wears most on membranes and cause them to age and also may
cause emissions of chlorine gas and formation of toxic chlorinated organic compounds. During the first 43
weeks, the aeration of the membranes was kept in Leap-Lo to facilitate evaluation of results from the chemical
cleaning trials. Acid cleaning was operated according to the optimized scheme most of the year. For 24 weeks,
MC with hypochlorite was turned off completely in both membrane tanks (normally done twice per week in
each MT). During the first 10 weeks the hypochlorite MC was replaced with water backpulses. The
permeability remained stable during this period. Thereafter both hypochlorite and water backpulses were
turned off and only acid MC was done. The permeability then decreased from almost 400 lmh/bar to below
175 lmh/bar (MT1 to as low as 121 lmh/bar). To minimize hypochlorite consumption without affecting the
membrane function, a filtration resistance-based method for hypochlorite MC was then tested during the last
10 weeks of the year (after restoring permeability). During this period, 4 hypochlorite MC:s in MT1 and 1 in
MT2 were carried out. This corresponds to a hypochlorite consumption that is 8 times lower than the
recommended cleaning scheme (2 times/week in each MT).

Recovery cleanings (RC) of the membranes were performed twice in 2020 (March-April and December) by
soaking the membranes in first sodium hypochlorite and then in citric or oxalic acid. During the soaking in
sodium hypochlorite, a separate measuring campaign of chlorine rich gas emission from the tank was
performed in March. The results showed that emissions occurred during a much longer period than expected
and extended measurement will therefore be conducted during next year.

Two experiments were conducted in the sludge pilot in 2020; a thermophilic and a mesophilic hydraulic
retention time (HRT) crash test where the HRT was gradually lowered until the anaerobic digestion processes
collapsed from microbial imbalance and subsequent acidification. Due to recurrent operational problems with
the thickener, it was bypassed throughout the experiments. The first trial started in July 2019 at thermophilic
conditions with HRT 9 days. The initial organic loading rate (OLR) was low, around 2 kg VS/m3,d but
increased with decreasing HRT. It was difficult to provoke a crash and the digester still performed relatively
well after 7 weeks at HRT 4 days and OLR >4 kg VS/m3,d. Since it was not technically possible to further lower
the HRT, glycerol and cooking oil was added in increasing amounts to the digester until the digester finally
crashed (pH dropped below 6, marginal biogas production) at an OLR of almost 14 kg VS/m3,d. A similar trial
at mesophilic conditions started in June 2020 with HRT 10 days and an OLR of 1.5-2 kg VS/m3,d. By the end of
2020 the digester had been operated at HRT 4 days and OLR >3 kg VS/m3,d for a couple of weeks without any
decrease in pH or reduction in specific methane production compared to normal values. After 5 days of
glycerol dosing the process crashed on the 15th of January 2021 with an OLR of 5.3 kg VS/m3,d. While both
thermophilic and mesophilic digestion showed stable performance down to 4 days HRT, thermophilic
operation managed a higher OLR before crashing and also showed a better overall performance (degradation
rate, methane production) at low HRTs.

The overall resource consumption in the pilot showed that the optimization of phosphorus precipitation and
membrane cleaning chemicals resulted in a significantly lower dosing than design values for the future
Henriksdal WWTP. The consumption of external carbon source was similar to the design, despite the lower
set point for nitrate in the pilot. The airflow demand, on the other hand, was significantly higher in the pilot,
which largely is explained by the configuration/smaller depth in the tanks.

7
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Sammanfattning
Henriksdals reningsverk i Stockholm byggs nu ut och om för ökad kapacitet (från 0,8 till 1,6 miljoner PE) och
för förbättrad reningsgrad (6 mg TN/L, 0,20 mg TP/L, 5 mg BOD7/L). Projektet inkluderar uppgradering av
den befintliga konventionella aktivslamprocessen till en ny membranbioreaktorprocess (MBR) med mer än 1,6
m2 installerad membranyta. Det inkluderar även utökad förbehandling och ett nytt steg för
primärslamförtjockning. Termofil rötning av tjockt slam (~6% TS) vid hög organisk belastning och relativt låg
uppehållstid kommer ersätta dagens mesofila rötning.

För att öka kunskapen om MBR-teknik i nordiskt klimat har Stockholm Vatten och Avfall (SVOA) och IVL
Svenska Miljöinstitutet genomfört långtidsstudier på en membranprocess i pilotskala på F&U-anläggningen
Hammarby Sjöstadsverk, som ligger i anslutning till Henriksdals reningsverk. MBR-piloten togs i drift 2014
och byggdes om till sin nuvarande utformning under 2016. År 2017 kompletterades MBR-piloten med en
slamlinje för att kunna studera olika aspekter av slamrötning.

Under 2020 drevs MBR-piloten med fast inflöde på 4,4 m3/h, vilket är 37% högre än designflödet, med glycerol
som kolkälla för efterdenitrifikation och aluminium (PAX) istället för trevärt järn (PIX) som komplement till
järnsulfaten för fosforfällning. Detta för att testa den driftstrategi som kommer tillämpas under de första årens
drift av fullskalelinjen på Henriksdals reningsverk. Utsläppsvärden för fosfor och kväve var som årsmedel
0,10 mg TP/l och 3,9 mg TN/l vilket innebär att de framtida reningskraven underskreds även detta år. För att
åstadkomma detta doserades i snitt 13,7 g Fe2+/m3 (doserades innan försedimenteringen), 0,5 g Al3+/m3 och
glycerol motsvarande 16,7 g COD/m3 avloppsvatten.

Den låga förbrukningen av fällningskemikalie för fosforrening, 0,95 mol metall per mol avskild fosfor,
berodde till stor del på en hög biologisk fosforreningsaktivitet (bio-P), trots avsaknad av en anaerob zon.
Under 2020 ökade fosforsläppshastigheten (indirekt mått på bio-P aktivitet) till 10 g PO4-P/kg VSS, h vilket är
högre än tidigare år och indikerar mycket väl utvecklad bio-P. Halten av järn och aluminium i slam var låg,
4,1% respektive 0,5%. Två examensarbeten utfördes inom ramen för projektet för att studera bio-P processen.
Resultaten visade bland annat att fosforsläppet skedde i fördenitrifikationszonen i närvaro av nitrat. Under
våren testades etanol som extern kolkälla för efterdenitrifikation istället för glycerol för att se om det
förstärkte bio-P aktiviteten, men ingen effekt kunde ses.

Förbrukningen av glycerol (extern kolkälla) var likvärdig med resultaten från tidigare försök med metanol,
sett till COD-dos (g/m3 avloppsvatten), något som också bekräftades med denitrifikationstester på labbet.
Resultaten visar att glycerol kan vara en fullvärdig ersättare för metanol på Henriksdals reningsverk fram till
dess metanolanläggningen byggts och driftsatts. Glycerolen är dock inte en ren produkt och potentiella
långtidseffekter på mängden kolloidalt organiskt material (cTOC) i processen och deras effekt på
membrandriften måste utvärderas mer ingående. Etanol uppvisade en högre denitrifikationshastighet vid
labbtester (3,1-4,7 mg NO3-N/g VSS, h) jämfört med glycerol (2,4 mg NO3-N/g VSS, h) och även en lägre
specifik COD-förbrukning, 1,7-4,4 g COD/g NO3-Ndenitrifierat jämfört med 6,6 g COD/g NO3-Ndenitrifierat för
glycerol. Trots detta var förbrukningen av etanol när den doserades i piloten 16,2 g COD/m3 vilket är nästan
identiskt med glyceroldoseringen samtidigt som den specifika dosen var lite högre än den för glycerol (2,9 g
COD/g NO3-Ndenitrifierat för etanol och 1,8 g COD/g NO3-Ndenitrifierat för glycerol).

Slamåldern i processen år 2020 var i snitt 17,5 dagar och den luftade slamåldern 6,5 dagar. Fullständig
nitrifikation med utgående ammoniumhalter på under 2 mg/l uppnåddes alla veckor utom vecka 11, 36 och
50-51. Under både vecka 11 och 38 minskades luftflödet medvetet i försök att minska flyslambildning. Under
vecka 50-51 inhiberades nitrifikationen av hypoklorit i kombination med kallt vatten i samband med
återställningsrengörning (RC), något som aldrig tidigare inträffat.

8
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Liksom tidigare år tvättades membranen i membrantank 1 (MT1) med oxalsyra och MT2 med citronsyra. Båda
membranen rengjordes även med hypoklorit. Membranen kördes med ett årsmedelflux på ungefär 22 lmh vid
ett TMP-medel på 79 mbar. Detta resulterade i en permeabilitet på ungefär 325 lmh/bar (årsmedelvärde).
Under 2019 optimerades syraförbrukningen vid underhållsrengöring (MC) till motsvarande 50% av
specificerad mängd citronsyra och 25% av specificerad mängd oxalsyra. Under 2020 låg fokus på att minska
förbrukningen av hypoklorit. Detta är viktigt eftersom hypoklorit är den tvättkemikalie som sliter mest på
membranen och får dem att åldras samtidigt som den potentiellt kan ge upphov till utsläpp av klorgas och
giftiga klororganiska föreningar. Under årets första 43 veckor låstes membranluftningen i Leap-Lo läge för att
underlätta utvärderingen av hypoklorittvätten. Samtidigt kördes syratvätten enligt det optimerade schemat
från året innan. Under 24 veckor användes ingen hypoklorit alls i piloten (jämför med normalt 2 tvättar per
vecka och membran). De första 10 veckorna, då hypoklorit MC ersattes med vatten, höll sig permeabiliteten
på en jämn nivå. När sedan backpulser med både hypoklorit och vatten stängdes av och enbart MC med syra
användes, sjönk permeabiliteten från ca 400 lmh/bar till under 175 lmh/bar (i MT1 till så låt som 121 lmh/bar)
över 13 veckor. Slutsatsen var att antalet hypoklorittvättar kan minskas rejält men inte helt uteslutas. För att ta
reda på exakt hur ofta hypoklorittvättar måste genomföras för att upprätthålla membranens drift och funktion
så infördes en motståndsbaserad styrning av hypoklorit MC som testades under årets sista 10 veckor (efter att
permeabiliteten återställts). Under denna period genomfördes 4 MC med hypoklorit i MT1 och en i MT2.
Jämfört med mängden hypoklorit som skulle åtgå vid standardrengöring (2 ggr/vecka i varje MT) var
förbrukningen med denna metod 8 ggr mindre.

Återställningsrengöring (RC) genomfördes vid två tillfällen under 2020, en i mars/april och en i december,
genom att blötlägga membranen först i hypoklorit och sen i syra. Under RC med hypoklorit i mars
genomfördes mätningar av klorhaltiga gaser i frånluften. Resultaten visade att klorhaltiga gaser förekom
under en längre tid än förväntat varför utökade mätningar kommer utföras vid RC nästa år.

Två experiment genomfördes i slampiloten under 2020, ett termofilt och ett mesofilt kraschtest där
uppehållstiden (HRT) i rötkammaren minskades stegvis tills rötningsprocessen kollapsade till följd av
surrötning. På grund av återkommande driftproblem med förtjockaren tidigare år beslutades att förbileda
förtjockningen under försöken. Första försöket startade i juli 2019 med termofil rötning vid 9 dagars
uppehållstid. Den organiska belastningen (OLR) var då låg, omkring 2 kg VS/m3,d men ökade sedan med
minskande HRT. Det var svårt att framkalla en krasch och rötningen fungerade fortfarande relativt bra efter 7
veckor vid 4 dagars HRT och en OLR på >4 kg VS/m3, d. Eftersom det inte var tekniskt möjligt att sänka
uppehållstiden i rötkammaren mer så tillsattes istället glycerol och matolja i ökande mängder för att på så sätt
öka OLR. När rötkammaren till slut kraschade (indikerades av pH<6 och marginell gasproduktion) var OLR
nästan 14 kg VS/m3, d.

Ett likadant försök genomfördes sedan vid mesofila förhållanden. Vid starten i juni 2020 var HRT 10 dagar
och OLR 1,5-2 kg VS/m3,d. Vid slutet av 2020 hade rötkammaren drivits vid HRT 4 dagar och OLR>3 kg
VS/m3, d under några veckor utan att pH minskat eller den specifika metanproduktionen förändrats jämfört
med värden vid normal drift. Efter tillsats av glycerol under 5 dagar kraschade reaktorn den 15 januari 2021
vid ett OLR på 5,3 kg VS/m3, h. Trots att både termofil och mesofil rötning uppvisade stabil drift ner till 4
dagar HRT, kunde den termofila processen klara en högre OLR innan kraschen samtidigt som den gav bättre
resultat i form av nedbryningsgrad, metanproduktion mm vid låga HRT.

Resursförbrukningen i piloten visade att optimering av fosforfällning och membranrengöring resulterade i

betydligt lägre relativ förbrukning än designvärden för framtida Henriksdals reningsverk. Förbrukningen av
extern kolkälla vad snarlik designvärdet trots lägre börvärdet för utgående nitrat i piloten. Luftningsbehovet å
andra sidan var betydligt högre i piloten är i designen för framtida Henriksdal vilket troligtvis förklaras av
utformningen av reaktorerna med relativt grunda tankar.

9
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Terminology
AD Anaerobic Digestion
Anoxic Process condition without dissolved oxygen, but available NO3
Anoxic zone Non-aerated zone
AOX Adsorbable organic halogens (mg/L)
ATEX Atmospheres Explosibles
BOD7 Biochemical Oxygen Demand, 7 days (mg/L)
BR1 to BR6 Biological reactor 1 to 6, sampling points
COD Chemical Oxygen Demand (mg/L)
cTOC collodial Total Organic Carbon (mg/L)
DDMS Dewatered digested mixed sludge, sampling point
DMS Digested mixed sludge, sampling point
DO Dissolved Oxygen (mg/L)
DS Daily composite sample (flow proportional)
EBPR Enhanced Biological Phosphorus Removal
EFF Effluent water, sampling point
EOX Extractable organic halogens (mg/L)
Fe Iron (mg/L)
F/M ratio Food to Mass, incoming substrate in relation to the amount of microorganisms
(kg BOD7/kg SS, d)
Flux Flow rate per unit area (L/(m2·h)). Flux is a measurement of the load on the membranes
Fouling Clogging of the pores in the membranes, causing reduced flow rate through the membranes.
In this report we use Fouling for both organic clogging and inorganic precipitation on
membranes (sometimes referred to as scaling).
GS Grab sample
Hepta Iron(II)sulfate heptahydrate
H-dal Short for Henriksdal WWTP
IN Influent wastewater, sampling point
Mesophilic Temperature condition in anaerobic digester, in this project 37 °C
MBR Membrane BioReactor, bio reactor with membrane separation
MLD Million litres per day
MT1 Membrane tank 1 (of 2), sampling point
MT2 Membrane tank 2 (of 2), sampling point
MC Maintenance cleaning
MS Mixed sludge (PS+WAS), sampling point
NIT Nitrification zone
NH4-N Ammonium nitrogen (mg/L)
NO2-N Nitrite nitrogen (mg/L)
NO3-N Nitrate nitrogen (mg/L)
Org-N Organically bound nitrogen (mg/L)
PA Pre-aeration
PE Population equivalent (defined as 70 g BOD7 per person and day)
Permeability Flux per TMP (L/(m2·h·bar)). Permeability is a measure of how well a specific flux permeates
the membranes. The permeability gradually decreases with time due to fouling
Permeate The treated wastewater that has passed through the membranes
PFAS Perfluorinated Alkylated Substances
PIX PIX 111, brand name of iron(III)chloride solution
PO4-P Phosphate phosphorus (mg/L)

10
 Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Pre-DN Pre-denitrification (Anoxic)

Post-DN Post-denitrification (Anoxic)
PS Primary sludge, sampling point
PTW Primary treated water, water after primary settler, sampling point
RAS Return activated sludge, sampling point
RAS-DeOx Zone where return activated sludge (RAS) is led for reduction of DO concentration
RC Recovery cleaning
RWD Reject water from sludge dewatering, sampling point
RWT Reject water from sludge thickening, sampling point
Scouring air Constant air flow around the membranes to reduce fouling
SED Pre-sedimentation (Primary settler)
SFA/SFA 2040 Stockholms Framtida Avloppsvattenrening år 2040 (name of reconstruction project) 1
SS Suspended Solids (mg/L)
SVOA Stockholm Vatten och Avfall
Thermophilic Temperature condition in anaerobic digester, in this project 55 °C
TOC Total Organic Carbon (mg/L)
TMP Transmembrane pressure (mbar). The pressure difference between two sides of a membrane,
shows how much force is needed to push water through a membrane
TN Total nitrogen (mg/L)
TP Total phosphorus (mg/L)
TMS Thickened mixed sludge, sampling point
TS Total Solids (%)
TSS Total Suspended Solids (mg/L)
TTF Time To Filter (s)
VS Volatile Solids (% of TS)
VSS Volatile Suspended Solids (mg/L)
WAS Waste activated sludge, sampling point
WS Weekly composite sample
WWTP Wastewater Treatment Plant

1 www.stockholmvattenochavfall.se/en/sfa-start/
11
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

1 Introduction
This report presents the results from year 2020 (project year 7), of the pilot scale trials with membrane
biological treatment of municipal wastewater with Membrane BioReactor (MBR, operated since 2014) and
associated sludge treatment (operated since 2018). The project is carried out in cooperation between IVL
Swedish Environmental Research Institute and Stockholm Vatten och Avfall AB at the R&D facility
Hammarby Sjöstadsverk, in Stockholm, Sweden. In the trials, an activated sludge process with a unique
process configuration is combined with membrane filtration to reach both a high degree of purification and
operational stability. Project years 2014-2019 have been presented in separate reports earlier.

In the initial chapters (2-3), the project background and the configuration of the pilot plant are described. An
overview of the experimental plan is presented in chapter 4, followed by a method description in chapter 5.
Finally, all results are presented and discussed in chapter 6.

No major alterations to the configurations have been made in 2020, and thus chapter 2 and 3 are similar as in
the previous report. However, the flowrates and load to the pilot for 2020 are summarized in section 3.3. Also,
the method section is similar as described in the previous reports.

2 Background
Within the project Stockholm’s Framtida Avloppsrening (SFA, Stockholm’s future wastewater treatment), the
Henriksdal wastewater treatment plant (WWTP) in Stockholm, Sweden, is being extended and rebuilt for
increased capacity and enhanced treatment efficiency. The decision to extend and rebuild the plant is based on
several factors such as; (i) SVOAs WWTP in Bromma (which is already over loaded with very limited space
available for extension) will be decommissioned in 2025 to give space to new housing areas, and the
wastewater will be led to the Henriksdal WWTP in a new 14 km long sewage tunnel, (ii) the population in the
Stockholm region is increasing at a high rate, resulting in an increased influent load, and, (iii) the Swedish
Environmental Court has decided to sharpen the effluent requirements on the WWTPs in the Stockholm
region, which demands more efficient wastewater treatment processes.

The new process configuration at the Henriksdal WWTP has been designed for a capacity of 1.6 million
population equivalents (PE) which is about twice as much as today. The design maximum flow of the
biological treatment is 10 m3/s which is equivalent to 850 MLD. In addition, the treatment process has been
designed to reach low concentrations of organic matter and nutrients in the effluent (5 mg BOD7/L, 6 mg TN/L
and 0.20 mg TP/L). The extension of the plant will include new primary treatment, new primary settlers and a
new treatment step for thickening of primary and waste activated sludge. The reconstruction will include
retrofitting of the existing conventional activated sludge (CAS) tanks with a unique MBR-process containing
1.6 million m2 of membrane area. The first MBR-line, out of seven, will be taken into operation in 2021 and the
retrofitting of all seven lines will take an additional 6-8 years. The sand filters, currently used as a final
polishing step for phosphorus removal, will in the future be used for treatment of the flow bypassing the
biology at high flows. In order to increase the capacity and quality of the sludge treatment, digestion of thick
sludge (~6% TS) will be done at thermophilic conditions instead of mesophilic digestion of thin sludge (~3-
3.5% TS). Design data for the future Henriksdal WWTP can be found in Table 1, Table 2 and Table 3.

The MBR technology is well-known internationally with long term experiences from both industrial and
municipal wastewater treatment plants. In Italy and Germany, relatively large municipal WWTPs with MBR-
technology have been in operation for around 15 years (Brepols, 2010; Judd, 2020). In USA, China, Japan,
12
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

South Korea, France, Great Britain and Spain, there are several large MBR-plants (50,000-80,000 PE) which
have been in operation for 5-10 years (Judd and Judd Limited, 2017). The largest MBR-plant in operation
today is Beihu WWTP in Hubei, China (commissioned in 2019), designed for an average inflow of 9.3 m3/s,
which is significantly larger than the capacity of the future Henriksdal WWTP (design average 6.1 m3/s).
Europe’s largest MBR in operation is currently Seine Aval in France (commissioned in 2016). The plant, with a
design average inflow of 2.6 m3/s (www.thembrsite.com, 2021-01-05), is also the largest MBR installation using
SUEZ's ZeeWeed ultrafiltration (UF) hollow-fiber membranes as used in this project.

Challenges for the future MBR-process at the Henriksdal WWTP include:

 High seasonal variations in wastewater temperature and inflow, affecting both the membrane
performance and nitrogen removal.
 To meet the low effluent requirements for phosphorus (0.20 mg TP/L and 27 tons TP/year) by means
of pre- and simultaneous precipitation without affecting membrane performance.
 To minimize resource consumption.

There are MBR-plants in the USA, eg. Broad Run and King William County in Virginia, Ruidoso in New Mexico
and Cauley Creek and Yellow River in Georgia, that reach very low effluent nutrient concentrations, 0.05-0.10
mg TP/L and below 6 mg TN/L, without final polishing steps (Pellegrin et al., 2015). Phosphorus removal at
these plants is achieved by a combination of enhanced biological phosphorus removal (EBPR) and
precipitation using a trivalent metal ion (Al3+ or Fe3+). However, none of these treatment plants use ferrous
(Fe2+), which is planned to be utilized at the Henriksdal WWTP, or have as low incoming water temperatures
as the Henriksdal WWTP.

Membrane filtration requires aeration and chemicals for maintenance and cleaning of the membranes.
Standard protocols for air scouring and chemical cleaning of the membranes exist. However, each plant is
unique, and the cleaning schedule can and should be optimized for the local conditions to save resources.

The future doubled treatment capacity at Henriksdal WWTP will also affect the sludge treatment. The load on
the digesters is expected to double but the digester volume was, at the time of the experiments, not planned to
be expanded (during 2021 it was decided to construct one more digester, increasing the total volume with
20%). Consequently, digestion must be performed with higher organic load and shorter hydraulic retention
time. To manage this, not only the waste activated sludge but also the primary sludge will be thickened, and
digestion will be performed at thermophilic conditions. There are several uncertainties regarding the sludge
handling, including: function of thickening of fine particulate MBR-sludge, stability of the digestion process,
biogas production potential, smell, pumping of thick sludge, and function of dewatering of thermophilic
digested sludge.

To increase the knowledge on membrane technology for wastewater treatment in Nordic conditions, SVOA
and IVL decided in 2013 to conduct long-term pilot scale studies at the R&D facility Hammarby Sjöstadsverk,
located on the premises of the Henriksdal WWTP. In 2017, the project team decided to supplement the MBR-
pilot with a sludge treatment line in order to study the future digestion process.

13
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

3 Description of the pilot plant

The pilot plant is designed to be a small copy of the future Henriksdal WWTP plant, scale
1: 6,700. The incoming wastewater is pumped from the Henriksdal inlet with a mean flow of around 3.2 m3/h.
Primary treatment comprise of a fine screen (6 mm), pre-aeration, a primary settler and a fine sieve (2 mm).
The biological treatment consists of pre- and post-denitrification followed by two parallel membrane tanks.
The return activated sludge (RAS) passes a deoxygenation zone (RAS-Deox). The purpose of this zone is to
lower the oxygen concentrations in the RAS stream not to disturb the pre-denitrification. The sludge treatment
consists of mechanical thickening, anaerobic digestion, and mechanical dewatering. The pilot plant process
set-up is shown in Figure 1. All equipment in the pilot has been linked to a control system and process control
is highly automated.

Figure 1. Flow scheme of the pilot WWTP.

The reactor volumes of the pilot plant and the function of each reactor are specified in Table 1 together with a
comparison to the future Henriksdal WWTP design.

Table 1. Reactor volumes in the pilot WWTP compared to the future Henriksdal WWTP.
Pilot Future Scale factor
Tank (m3) H-dal (m3) H-dal/Pilot Specification
Pre-treatment
PA (sand trap) 0.7 2 460 - Pre-aeration. Dosing point 1 Fe2+.
SED 3.3 30 000 9 200 Primary settler. Withdrawal of primary sludge.
Membrane bioreactor (MBR)
BR1 4.8 33 500 7 000 Anoxic conditions. Stirred. Pre-denitrification.
BR2 4.8 33 500 7 000 Anoxic conditions. Stirred. Pre-denitrification.
BR3 4.8 40 000 8 300 Flex. Stirred/(aerated). Pre-
denitrification/(nitrification).
BR4 4.8 31 000 6 500 Aerated. Nitrification. Dosing point 2 Fe2+.
BR5ox 1.5 10 000 6 700 Aerated. Nitrification.
BR5Deox 3.3 15 000 4 500 Deox. Stirred.
BR6 4.8 24 000 5 000 Anoxic conditions. Stirred. Post-denitrification.
Dosing external carbon. Dosing point 3 Fe3+ or Al3+.
MT1 1.45 9 750 6 700 Membrane tank. Aerated.
MT2 1.45 9 750 6 700 Membrane tank. Aerated.

14
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Pilot Future Scale factor

Tank (m3) H-dal (m3) H-dal/Pilot Specification
RAS-Deox 2.7 18 000 6 700 Deoxygenation of the RAS. Stirred. Addition of reject
water (RWD). Withdrawal of WAS (before addition
of RWD).
Summary MBR
Total MBR 34.4 224 500 6 500 BR1-6, MT1-2, RAS-Deox

Sludge treatment
MS tank 0.4 1 060 2 650 Tank for PS + WAS before thickening. Stirred.

Digester 5.9* 38 000 6 500 Anaerobic digestion volume. Stirred.

DMS tank 0.2 9 000 45 000 Circulation mixing. Tank for digested mixed sludge
before dewatering.
*The volume is set by choosing the liquid level in the digester and can be increased or decreased.

3.1 Process description water line

3.1.1 Incoming wastewater
Incoming wastewater to the pilot plant is pumped from the Danviken channel, one of five inlet tunnels to
Henriksdal WWTP plant. The pilot influent contains a 10-20% higher concentration of organic matter
(measured as BOD7) than the combined average inflow to the Henriksdal WWTP. It also has about 60% higher
BOD7-concentration than the inlet to Bromma WWTP. The combined inlet from Henriksdal and Bromma will
make up the future inlet to the Henriksdal WWTP, after reconstruction. The incoming flow rate to the pilot
plant is proportional to the projected inflow to the Henriksdal WWTP year 2040. However, flow variations in
the pilot inflow are proportional to the current actual inflow variations to Henriksdal WWTP, as the pilot
inflow is controlled by a signal from flow meters in the full-scale plant.

Since the influent to the pilot is set by a scaled down flow rate, and not a scaled down load, the resulting
incoming load on the pilot plant is higher than the corresponding design load for the Henriksdal WWTP, year
2040, see Table 2. In 2020, the inlet concentrations of nitrogen and phosphorous to the pilot were lower than
normal from March onward. This resulted in a yearly average concentration of total nitrogen 17% lower than
2019 and a total phosphorus concentration 26% lower than 2019. Also, at Henriksdal WWTP the inflow was
lower than previous years, resulting in a load that was around 12% lower for nitrogen, 18% lower for
phosphorus and 17% lower for BOD. This was presumably caused by the corona pandemic, leading to more
people working from home and therefore fewer working people in the office areas dominating central
Stockholm that are serviced by Henriksdal WWTP.

In addition, incoming wastewater to the pilot has a higher temperature than incoming wastewater to
Henriksdal. Previous years, the incoming wastewater was during some periods cooled in heat exchangers to
compensate for this. However, due to continuous problems with clogging, cooling was during 2020 limited to
the nitrate recirculation (flow from BR5 to BR1). The processes have been tested during cold inlet
temperatures previous years. The heat exchanger was controlled to maintain a temperature in MT1
corresponding to 1°C higher than inlet to Henriksdal, since temperature normally increases from inlet to
biology in Henriksdal with about 1°C. The temperatures in the incoming wastewater to Henriksdal and to the
pilot are presented in Figure 2. On average the temperature of the inlet water to the pilot was 19.0°C, which is
2.2°C higher than the influent wastewater to Henriksdal (16.8°C). The daily average temperature in the pilot

15
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

inlet varied between 12.8°C and 25.0°C. With cooling on the nitrate recirculation using the temperature in
MT1, process temperature could be controlled to match the temperatures in Henriksdal WWTP +1°C. On
average the temperature in MT1 was 17.8°C.

Temperature
26

24

22
Temp. °C

20

18

16

14

12
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Henriksdal influent Pilot influent MT1
Figure 2. Influent temperatures to the MBR pilot (dotted line) and Henriksdal WWTP (black line) together
with temperature in the Membrane tank 1 (MT1) during 2020.

3.1.2 Pre-treatment
The pre-treatment steps in the pilot consisted of a 6 mm punch hole sieve (with screen capture rates similar to
3 mm step screen, UKWIR (2015)), a pre-aeration tank with Ferrous dosing and a vertical flow primary settler,
with a surface area of 1.13 m2 and a water depth of 4.3 m (scale 1:9,200 compared to the future Henriksdal
design), followed by a 0.6 mm punch hole drum sieve before the biology, see Figure 3. The small hole size of
the drum sieve was chosen to enable the study of clogging tendencies. In June 2020, the 0.6 mm fine sieve
before the biology was changed to a 2 mm punch hole drum sieve, which is the same hole size as in future
Henriksdal WWTP, to study tendencies of accumulation of trash in the biological sludge.

Figure 3. Photo of the fine sieve installation.

16
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

3.1.3 Biological treatment

The biological treatment consisted of six identical biological reactor tanks, BR1-6, see Figure 4. All tanks were
equipped with stirrers and BR3, BR4 and BR5 were equipped with membrane disc aerators. BR5 was divided
into two zones where the first one was aerated and the second one was stirred. The biological process was
operated with pre-denitrification, nitrification and post-denitrification with addition of primarily glycerol as
an external carbon source. The oxygen-rich return activated sludge (RAS) flow (4×Qin) passed a specific RAS-
Deox zone where RAS was mixed with ammonium-rich reject water from digested sludge dewatering before
recirculation to the pre-denitrification zone. Waste activated sludge (WAS) was taken out from the return
sludge stream, after the membrane tanks and prior to the RAS-Deox. Precipitation chemicals for phosphorus
removal was dosed in BR4 and BR6.

Figure 4. Photo of the top of biological treatment tanks BR2-4

The biological treatment set-up was almost identical to the design of the future Henriksdal WWTP in scale
1:6,700, with few minor exceptions. The deox zone in BR5 and the post-denitrification zone in BR6 were
slightly over dimensioned compared to the full-scale design. The discrepancy depends on the size of the
existing tanks in the pilot plant and the difficulties in creating zones within the tanks. When setting up the
pilot, a correct volume of the aerated zones for nitrification was given priority (BR4 and BR5ox), as the size of
this zone will be crucial for the nitrogen removal during winter.

Another difference between the pilot and the future Henriksdal WWTP is that the pilot lacks a RAS-channel.
Instead, the RAS flowed directly from the membrane tanks into the RAS-Deox from where it was pumped
back into BR1. In the full-scale plant, the RAS will flow into a RAS-channel by gravity and then be pumped
into the RAS-Deox zone from where it will flow to the pre-denitrification zone by gravity. The volume of the
RAS-channel will be small (HRT ~ 2 minutes) which puts a lot of pressure on the RAS-pumps. One of the
questions for the future Henriksdal WWTP is how to control the RAS-pumps to maintain a steady level in the
RAS-channel and to avoid flooding or creation of waves through the plant. This could not be tested in the
pilot since the RAS-Deox volume is much larger (HRT ~ 10 min) and the water level did not change as fast
with changes in inflow. Table 1 shows the size of the treatment volumes in the pilot plant compared to the
design of the future full-scale system at Henriksdal.

17
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

3.1.4 Membrane tanks

In the pilot, hollow fiber membrane from Suez with a nominal pore size of 0.04 μm was used (ZeeWeed 500D-
Leap). The membrane pilot was made up of two cassettes (2.5 m x 1.0 m x 0.34 m) consisting of three
membrane modules each, see Figure 5, immersed in two separate tanks. Each module had a membrane area of
34.4 m2 and consisted of membrane fibers fastened at the top and bottom of the cassette frame. The filtered
water (permeate) was transported on the inside of the fibers to connections in both the bottom and the top of
the module. The membranes were kept clean during operation by aeration from below (air scouring). As
shown in Figure 5c, the membranes were not completely tensioned between the top and bottom, so that the
air bubbles cause the fibers to move and thus more easily remove sludge stuck on the membrane fibers.

The two membrane cassettes in the pilot were operated in parallel but independently of each other to enable
comparisons of different operational strategies.

b)
a)

c) d)

Figure 5. The membrane during installation of the pilot. a) Membrane cassette with three membrane
modules, b) cassette lowered into the tank, view from above, c) mounting and aeration equipment at the
bottom of the cassette, d) permeate connections (yellow) at the top of the cassette.

The total membrane surface area in the pilot (206 m2) corresponds to the design membrane surface installed in
six treatment lines (out of seven in total) in the full-scale plant. There are two reasons for this. First, the setup
corresponds to two standard design pilot cassettes from the manufacturer. Secondly, the design max flow rate
to the biological treatment, according to the full-scale design of the future Henriksdal WWTP, could be treated
even if one of the seven treatment lines are out of operation.

In the future Henriksdal WWTP, each treatment line (a total of seven) will have 12 membrane tanks of which
each can be taken into and out of operation depending on the influent flow rate. Each membrane tank is
18
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

equipped with 12 cassettes, with 48 modules in each cassette. This provides good flexibility and an
opportunity to always have a constant flux across the membrane surface. In the pilot there are only two
membrane tanks and six modules, which gives less flexibility. At design average flow rate (see Table 2) and
normal operation, a membrane area of approximately 160 m2 would have to be in operation in the pilot, which
corresponds to 4.7 modules. However, the pilot could only be operated with three or six modules in
operation, as a pilot cassette contains three modules. To enable operation at a constant flux, the pilot was
equipped with permeate recirculation. This means that the flow through the membranes was higher than the
inflow by having a partial flow of the permeate recycled back to the membrane tank. This mode of operation
was chosen since it did not affect the concentration of sludge or pollutants around the membranes but
allowed operators to control the membrane operation by adjusting the flux.

The airflow requirement for membrane cleaning in the pilot plant is higher than the future airflow according
to the Henriksdal design since both cassettes in the pilot plant must be in operation most of the time. In future
Henriksdal, only the number of membrane tanks in operation will be constantly aerated, which means a lower
air consumption.

3.2 Process description sludge treatment

During 2017 the MBR-pilot was supplemented with a sludge treatment line proportional to the sludge
treatment of the future Henriksdal design. The sludge treatment pilot is visualized in Figure 6.

Figure 6. Process set-up for the sludge treatment line.

3.2.1 Thickening
Primary sludge and waste activated sludge was intermittently pumped to the mixed sludge tank. In the
standard set-up, mixed sludge was then pumped to a rotating drum sieve thickener with addition of polymer
and from there into the digester. In 2020, however, the thickener was bypassed (see chapter 6.7 for
motivation).

3.2.2 Digestion
The digester is cylindrical with a base area of 2.54 m2 and a variable sludge level. A volume of 5.7 m3
corresponds to full digester capacity in the future Henriksdal WWTP (scale 1:6,700). During 2020 the volume
was kept between 3.4 and 5.5 m3 (instead of 5.7 m3) in order to reach lower HRT (down to 4 days). The sludge,
which is fed continuously into the recirculation circuit, is kept in suspension by a stirrer and by the
recirculation flow. The recirculation circuit consist of a pump which is operated at its minimum capacity,
approximately 3 m3/h, and a heat exchanger controlled by a temperature meter in the digester. Digested

19
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

sludge is pumped out of the digester, through a heat exchanger which can cool the sludge to a chosen
temperature, and thereafter into an equalization tank (digested sludge tank).

During 2020, the trial with decreasing retention time at thermophilic condition, which was initiated in 2019,
was finalized. A new trial with decreasing retention time at mesophilic conditions started. Special attention
was given the operation at short HRT, with a master thesis student performing additional monitoring and
evaluating the mesophilic period.

In the future Henriksdal WWTP design, fat from grease traps at restaurants and industrial by-products like
glycerol will be co-digested with sludge. However, no external organic material was fed to the pilot digester,
except during the last weeks of the trials with thermophilic and mesophilic digestion at 4 days HRT, when
grease and glycerol was added to simulate a high organic loading rate (OLR) until the process crashed.

3.2.3 Dewatering
Digested sludge was stored in the digested sludge tank and pumped into a pressurized, stirred mixing tank.
Polymer was dosed inline just before the inlet to the mixing tank. From the mixing tank digested sludge was
fed into a screw press. Dewatered sludge was collected in a vessel and weighted. The dewatering equipment
is shown in Figure 7.

Reject water from dewatering of digested sludge was collected in a tank for continuous pumping into the
RAS-Deox zone in the wastewater treatment line. Due to operational problems the reject water was not
continuously returned to the process.

Mid-November to mid-December, the dewatering unit was not in operation, but since one of the trials
required reject water addition, reject water was collected at Henriksdal WWTP and dosed into the water line.

Figure 7. Photo of the dewatering equipment in the pilot.

20
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

3.3 Flow rate and load

Mean values for flow rates and loads of the pilot wastewater and sludge treatment lines during 2020 are
shown in Table 2 and Table 3, respectively, together with the design values for the future Henriksdal WWTP.
The design data for the pilot are also given in the table for comparison. The pilot was in operation during the
entire year without any longer interruptions in operation.

The average incoming flowrate in 2020 was 4.37 m3/h which is higher than the design average flow rate 3.16
m3/h. This was done in accordance with the test plan for the pilot which included testing operational
strategies with high load.

Table 2. Operation and design data for the wastewater treatment line in the pilot plant and design data
(year 2040) for the future Henriksdal WWTP.
Value Design Value Pilot/
Pilot Design future Design H-
Parameter Unit 2020 Pilot H-dal dali
Flowrates
Average influent flowrate, Qin m3/h 4.37 3.16 20 880 0.021%
Design average flowrate, Qdim m3/h 3.32 21 960 0.015%
Max flowrate m3/h 5.5 5.44 36 000 0.015%
Min flowrate m3/h 1.8 1.8 11 600 0.016%
Nitrate recirculation flowrate m3/h 5.1-13.1 3.8-13.3 - -
Nitrate recirculation flowrate × Qin 2.5 1.2-4.2ii 0-4 -
RAS flowrate m3/h 4.1-23.3 3.6-19 - -
RAS flowrate × Qin 4.2 1.1-5.9ii 4 (3-5) 100%
Temperatures
Temperature influent °C 19.0 - - -
Temperature biology °C 17.8 - - -
Incoming load
BOD7 influent mg/L 195 206iii 216 90%
SS influent mg/L 221 201iii 280 79%
TN influent mg/L 39 44iii 37 105%
TP influent mg/L 4.7 5.7iii 4.9 96%
Primary settler (SED)
BOD7 reduction over SED % 21 46 50iv 42%
SS-reduction over SED % 33 60 60iv 55%
TN reduction over SED % 0 10 10iv -
TP reduction over SED % 7 40 40iv 17.5%
BOD7 PTW mg/L 159 112 108 147%
SS PTW mg/L 145 80 112 130%
TN PTW mg/L 39 40 33 118%
TP PTW mg/L 4,4 3.4 3.0 147%
SS removed over SED kg SS/d 7.6 13.3v 89 300 0.009%
Primary sludge production kg SS/d 21.7 17.2v 115 000 0.019%
VS-concentration PS % of TS 88% 77% 77% 114%
Biological treatment
BOD7-load PTW (at average flowrate) kg BOD7/d 16.0 8.6 57 500 0.028%
Specific WAS-production vi kg SS/kg BOD7 0.99 1.02 1.02 97%

21
 Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Value Design Value Pilot/

Pilot Design future Design H-
Parameter Unit 2020 Pilot H-dal dali
WAS production, average kg SS/d 15.9 8.8 58 600 0.027%
VSS-concentration WAS % of TSS 80% 64% 64% 125%
SS in biological tanks mg/L 7 600 8 000 8 000 95%
SS in membrane tanks mg/L 9 900 10 000 10 000 99%
Total sludge age d 17.5 32.0 31.2 56%
Membrane tanks
Installed membrane area (gross) m2 206 206 1 600 000 0.013%
Permeate recirculation m3/h 0.03-0.9 0.05-2 - -
Net flux average (at average T) l/m2,h 22.0 17.9 20.9 105%
Net flux max l/m2.h 25.3 30.8 30 84%
Permeate pumping max m3/h 7.0 12.4 62 250 0.011%
Permeate pumping min m3/h 0 0 0 -
Specific air demand at Leap-Lo vii Nm3/h, m2 0.136 0.136 0.098 139%
Specific air demand at Leap-Hi vii Nm3/h, m2 0.252 0.252 0.196 129%
i Value pilot divided by Design future H-dal. Value 100% for complete compliance.
ii Based on average flowrate 3.2 m3/h.
iii Design based on data from 2015.

iv Measured at Fe-dosage ca 10 g/m3 in FL/sand trap.

v Calculated based on incoming load/scaled from future H-dal design with factor 6,700.

vi Excluding external carbon source. Calculated from process data for Values Pilot 2020. Design values from German standard ATV

DVWK-A 131E (2000) based in incoming SS and BOD, and SRTtot.

vii Aeration of the membranes had two modes, one with lower (Leap-Lo) and one with higher air flowrate (Leap-Hi).

Table 3. Operation and design data for the sludge treatment line in the pilot plant and design data (year
2040) for the future Henriksdal WWTP. The data is presented as average from the thermophilic (w.1-17)
and the mesophilic trial (w. 25-53)
Value pilot Value pilot Design future
Parameter Unit w.1-17 w.25-53 H-dal
Sludge into digester (Thickener bypassed in the pilot. Thickened sludge in design Henriksdal)
Flow into digester (TMS/MS) L/h 31.4 27.0 118 000
TS-concentration TMS/MS % 2.4% 1.9% 6.0%
TS-load TMS/MS kg TS/d 18.0 12.4 172 000
VS-load TMS/MS kg VS/d 13.1 9.8 124 000
VS-load EOM kg VS/d 1.7a 0 44 000
Digestion
Digester temperature °C 55 37 55
Rector volume m3 4.1 4.9 38 000
Hydraulic Retention time, HRT d 5 8 13b
Specific VS-load (OLR) kg VS/m ,d
3 4.3 2.0 3.3b
Organic degradation rate (ODR) % of VSin 52% 40% 42%b
VFA/Alkalinity mg CH3COO- 0.12 0.07 -
eq/mg CaCO3
Out of digester
Flow Digested mixed sludge, DMS L/h 31.4 27.0 123 000
TS-concentration DMS % 1.2% 1.1% 3.9
VS-concentration DMS % of TS 77% 75% 60%
TS-load DMS kg TS/d 8.8 7.4 124 000
VS-load DMS kg VS/d 6.8 5.5 74 000

22
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Value pilot Value pilot Design future

Parameter Unit w.1-17 w.25-53 H-dal
Specific biogas production m3/kg VSdeg. 0.67 0.88 1.0
Specific biogas production m3/kg VSin. 0.35 0.35 0.42
Flow biogas m3/d 4.5 3.1 52 000b
Methane content biogas % 57% 59% 65%
TS-concentration DDMSc % 26% 21% 30%
SS-concentration reject RWD mg/L 1 810 652 900
NH4-N in reject water mg/L 298 150 1 500
Polymer consumption dewateringc g/kg TS 14.5 11.5 6-10
a) Only feed to digester at the end of the crash test during 3 weeks. Averaged out over the whole period.
b) Numbers excluding addition of external organic material in order to be comparable to data from the pilot.
c) Different methods of dewatering. Not comparable.

3.4 Chemicals
During 2020, glycerol and ethanol were used as external carbon source (only one at a time) in the post-
denitrification zone. Phosphorus was precipitated using ferrous(II)sulphate at one dosing point and
aluminium(III)chloride in another dosing point. For membrane cleaning sodium hypochlorite was used for
both MTs while one MT was cleaned using citric acid and the other one using oxalic acid.

3.4.1 External carbon source

Glycerol was tested in the pilot during 2020 since it is a potential temporary carbon source for the full-scale
process at Henriksdal WWTP. The full-scale process is designed to use methanol as external carbon source,
but the methanol storage facility will be constructer in a later phase of the project and therefore a temporary
installation will be needed at Henriksdal during a few years. Scandinavian Biogas, the company responsible
for upgrading and selling the produced biogas at Henriksdal WWTP, have storage tanks for glycerol on site
and use glycerol to increase the biogas production in Henriksdal digesters when the biogas demand is high.
The facts that tanks are already available, and that glycerol is not flammable, and thus doesn’t require ATEX
approved installations and special permits, makes it an attractive candidate for post-denitrification. However,
the efficiency of glycerol as a carbon source was not known and therefore it was tested in the pilot. Glycerol
was collected in 25 L containers from Scandinavian Biogas’s storage. The measured COD concentration was
about 850-900 g COD/L.

Ethanol was used during a couple of months for a special trial when a master student evaluated the effect of
carbon sources on the enhanced biological phosphorus removal (EBPR) activity. The ethanol was delivered as
96% solution in 25 L containers with a concentration of about 1 600 g COD/L. As the available pump for
dosing was too large, the ethanol solution was diluted 1:1 based on volume before added to the process.

The external carbon sources that have been tested previously in the pilot are:

 Sodium acetate (2014-2015)

 Brenntaplus (2015-2016)
 Methanol (2017-2019)
 Acetic acid (2019)
 Glycerol (2019)

23
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

The dosing point of external carbon source was between the BR5 deox-zone and BR6. This point was tested
out previously and provided longer residence time compared to dosing directly in BR6 (which also led to a
higher risk of carbon source leakage to the membrane tanks) while avoiding risk of recirculating carbon
source to BR1 via the nitrate recirculation from the BR5 deox-zone.

More about carbon source addition and treatment results can be found in section 6.2.2 Denitrification and
section 6.9 Resource consumption.

3.4.2 Precipitation chemicals

Phosphorus was removed in the aqueous phase by precipitation with iron(II)sulfate heptahydrate (termed
"hepta" in the report) and aluminium(III)chloride. Previous years, a combination of iron(II)sulfate and
iron(III)chloride (PIX 111) was used but during 2020 the iron(III)chloride was replaced with
aluminium(III)chloride.

Hepta was collected in diluted form from Henriksdal treatment plant in batches of about 500 L. The iron
content of the hepta solution varied during the experimental period between 36 and 70 g/L. For the batches
used in the experiment, the iron content was determined by density measurement for each batch.

Aluminium chloride (PAX-XL60 from Kemira) was delivered as a solution with at concentration of 7.5% Al by
weight as specified by the supplier. An aluminium concentration of 98.3 g Al/L has been used for dose
calculations.

3.4.3 Chemicals for membrane cleaning

The membranes have been cleaned regularly with sodium hypochlorite and either citric acid or oxalic acid.
For more information on how the cleanings were carried out, see section 6.5.3.

Sodium hypochlorite was delivered as a solution with a concentration of 10-20% by weight (150-185 g Cl2/L),
as specified by the supplier. The chlorine concentration in sodium hypochlorite decreases during storage. To
prevent fast degradation the sodium hypochlorite has been stored in a closed, dark container. According to
literature the rate of the degradation also decreases if the solution is diluted upon delivery (p.68. Svenskt
Vatten, 2010a). During 2020, both diluted and non-diluted sodium hypochlorite in the storage tank has been
tested, and pumping have been adjusted to provide the right concentration in the solution entering the
membranes during cleanings. Dilution was done with tap water to a concentration of about 60 g Cl2/L. The
concentration of sodium hypochlorite in the storage tank varied between 35 and 75 g Cl2/ L during the year.
Due to trials with reduced use of sodium hypochlorite in 2020, the storage time and potential degradation has
been greater than before.

Citric acid solution was delivered with a concentration of 51% by weight as specified by the supplier.

Oxalic acid was delivered as powder which was dissolved in batches to a saturated solution (8% by weight).

24
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

3.4.4 Polymers
For dewatering of digested sludge Flopam EM 640 HIB (SNF) was used during 2020 for both the mesophilic
and thermophilic sludge. Previous years other polymers have been tested.

Polymer was delivered in solution and prepared to desired concentrations in % by weight solution in
automated polymer make up units.

3.5 Control system

The pilot plant uses a control system consisting of a PLC (ABB AC800M) and a SCADA (UniView version
9.01). The control system is a standard system used at several treatment plants in Sweden. All equipment
connected to the pilot, including the membranes, is controlled via the control system, except for pumping of
reject water that was locally controlled. Implementation of the control has been carried out within the project,
which provides great flexibility to adapt and optimize control.

25
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

4 Experimental plan year 2020

An overview of the experimental plan year 2020 is presented in Table 4 and in more detail in later chapters of
the report.

The complete history of the pilot project includes initial pilot trials with flat sheet membranes and a standard
MBR process configuration in 2014-2015 with the purpose to get to know MBR technology in general. Early
2016 the pilot was reconstructed to resemble the design of the future Henriksdal WWTP that had been set by
then. During 2016-2017 the main goal of the project was to verify that the process design could meet the future
effluent requirements for nitrogen (6 mg/L), BOD7 (5 mg/L) and phosphorus (0.20 mg/L) and that the
membranes functioned as expected. In 2017 the performance was tested with inlet temperatures <10°C for four
weeks. With the first objectives reached, the overall goals for 2018 were to continue with stable operation at
different operational conditions, to minimize the resource consumption in the process, to test and evaluate
specific processes/functions within the MBR-line, and to achieve proper function of the sludge pilot. During
2019, the main theme was “how low can we go” – regarding use of membrane cleaning chemicals, membrane
air scouring, membrane relaxation, nitrogen and phosphorus in the effluent and the retention time in the
digester. In addition, a transition from mesophilic to thermophilic digestion was done.

In 2020 the “how low can we go” theme continued with low setpoints for effluent phosphorus and nitrogen,
low chemical consumption for membrane cleaning and low retention time in the digester.

Table 4. Experimental plan of year 2020.

Month of the year
Trial J F M A M J J A S O N D
Increased inflow
Imitation of first phase operation of full-scale
Cooling in the biology
Glycerol as carbon source
Ethanol as carbon source
Reject water recirculation to pre-aeration
Reject water recirculation to RAS-deox
Effluent phosphorous target 0.10 mg TP/L
Aluminium chloride as precipitation chemical
In-depth study of EBPR
Minimising membrane scouring air use
Oxalic acid and citric acid comparison
Minimising membrane cleaning chemical use
Recovery cleaning
Offgas measurement
Extended membrane operational cycle
How low can we go - thermophilic
Emptying and re-seeding the digester
How low can we go - mesophilic
Sludge dewatering in operation
Study on indicators of process failure

26
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Minimizing the resource consumption, especially for membrane cleaning was continuously in the spotlight
this year. Further reductions in acid cleaning chemical consumption, both oxalic and citric have been
evaluated. In addition, the membranes were operated without sodium hypochlorite cleaning from 24th of
March to 9th of September (169 days). Recovery cleaning of the membranes was carried out twice, first in
March to April and later in December. Chlorine gas emissions from the process was measured during soaking
of the membranes in sodium hypochlorite.

During most of the year, the production goal for phosphorus and nitrogen was lowered according to the
theme of the year. The new permit for future Henriksdal will include both a maximum amount of
phosphorous to be released with the effluent as well as the maximum effluent concentration. The amount
specified in the permit will in the future, with expected high flows, imply that the effluent phosphorous must
be even lower than the concentration limit of 0.20 mg TP/L. To test the capacity of the process, the target
concentration in the pilot operation was lowered to 0.10 mg TP/L.

At the same time, the inflow was increased in order to mimic the first years of operation of the full-scale MBR-
line at Henriksdal WWTP, with a high, fixed inflow (37% higher than design average and 26% higher than
2019). The difference from previous similar trials was that glycerol was used as carbon source, while the
previous version of this trial was carried out without external carbon source. The reason for this was that
glycerol will be temporarily utilized as carbon source in the first MBR-line at Henriksdal WWTP for a few
years until the methanol storage facilities have been constructed. It was therefore interesting to test its
denitrification potential and possible effects on the overall treatment and membrane performance. In addition
aluminum(III)chloride (PAX) was used in combination with ferrous sulphate (Hepta) for phosphorus
precipitation, something that was not tested before. A strategy with three precipitation dosing points (a flow
proportional dose of Hepta added at the inlet and two dosages controlled by phosphate feedback control as
simultaneous precipitation using Hepta and ferric chloride (PIX) in BR4 and BR6, respectively) has been
shown successful in maintaining stable low effluent phosphorous concentrations. But, as the PIX dosing
equipment will not be ready to take into operation during the start of the first full-scale treatment line, PAX
(which is currently used for storm water treatment at Henriksdal WWTP) will be used in the full scale
initially. To study the capacity and any potential differences, PIX was replaced with PAX in the pilot in March
2020. To get a higher dosage of PAX, the dosage of Hepta in the biology was stopped. In addition to
evaluating the phosphorus removal using PAX, the plan included monitoring potential effects of the acids
used for membrane cleaning and possible effects on the sludge treatment from the aluminium product.

Digestion was tested at very low HRT at both thermophilic and mesophilic conditions to find out at what
point the process starts to fail. The trial is important since the digestion retention time will be jeopardized in
near future when digesters will be reconstructed and in a more distant future with high load and high
dependency on mechanical thickeners to maintain the retention time.

Separate trials in 2020 included the evaluation of enhanced biological phosphorus removal within two master
thesis projects. As part of one of the projects, another external carbon source, ethanol, was used in March. In
the autumn, another master thesis project was performed focusing on indicators of digester failure while
lowering the retention time.

27
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

5 Method
5.1 Sampling and analyses
Eurofins Environment Sweden AB (Lidköping) conducted analyses of water samples from five different
sampling points: IN (influent wastewater), PTW (primary treated water), EFF (effluent water), activated
sludge from bioreactor BR4 (SLUDGE 1) and return sludge from RAS-DeOx (SLUDGE 2), and analyses of
sludge samples from three different sampling points: PS (primary sludge), WAS (waste activated sludge) and
DS (digested and dewatered sludge). The sampling points (except SLUDGE 1 and 2) are illustrated in Figure
8.

Figure 8. Sampling points in the pilot process marked as black circles (SLUDGE1 and SLUDGE2 sampling
points not included in the figure).

Three different sampling types were used: daily composite samples, weekly composite samples and grab
samples. Daily samples were taken with automatic samplers set for flow proportional sampling. Weekly
samples were mixed from the daily samples proportionally to the mean flow during the respective days. Grab
samples were an instantaneous sample taken from the respective tank. The weekly composite samples were
conserved with 1 part 4M sulfuric acid to 100 parts sample volume, except for the samples analysed for TOC
which were conserved with 2M hydrochloric acid in corresponding proportions.

Table 5 lists the parameters analysed at the accredited laboratory for the respective sampling point and
sample type. One portion of the grab sample of sludge from the RAS-DeOx which was sent to Eurofins, was
used to measure sludge volume (SVI) and time to filter (TTF) at IVL’s laboratory at Hammarby Sjöstadsverk.
The filtrate from the TTF analysis was also sent for analysis of TOC. This was done in order to calculate the
colloidal TOC (cTOC, see section 5.3.2 Sludge quality) which, according to the membrane supplier, could
relate to membrane performance.

In addition to the samples and analyses presented in Table 5 a monthly composite sample of dewatered
digested mixed sludge (DDMS), which was stored at -30°C during the sampling period, was sent to external
laboratory for analysis of TS, VS, pH, nitrogen, phosphorus, chlorine, and 15 different metals. In addition,
multiple organic parameters and three more metals were analysed each three months, including
Polybrominated diphenyl ethers (PBDE, 24), Triclosan, Polychlorinated biphenyls (PCB, 7), Polycyclic
aromatic hydrocarbons (PAH, 6), organotin compounds (10), Phenols (19), Perfluorooctanoic acid (PFOA),
Perfluorooctanesulfonic acid (PFOS) and Per- and polyfluoroalkyl substances (PFAS). Parameters

28
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Table 5. Sampling points, parameters and number of samples sent per week for external analyses.
Parameters

NO3-N + NO2-N

Al (digested)
Fe (digested)

P (digested)
Alkalinity
NH4-N
PO4-P

cTOC
BOD7
COD
TOC

VSS

TN
TP

SS
Sampling point
Daily composite samples
IN 1 1 1 1 1 1
PTW 1 1 1 1 1 1 1
EFF 1 1 1 1 1 1
Grab samples
RAS-DeOx 1 1 1 1 1 1
Reject water mixed sludge thickening1 1 1
Reject water digested sludge dewatering1 1 1 1 1 1 1 1 1
Mixed sludge (MS)2 1
Digested mixed sludge (DMS)2 1
Weekly composite samples
IN 1 1 1 1 1 1 1
PTW 1 1 1 1 1 1 1
EFF 1 1 1 1 1 1
Total number 6 4 3 7 4 6 3 1 4 3 4 2 8 1 1
1Sampleswere sent when thickener/dewatering unit was in operation.
2Sampling started September 2 (w. 36) 2020.

In addition to the external analyses, analyses were also performed internally at IVL’s laboratory at Hammarby
Sjöstadsverk for daily operation. Water samples were analysed by means of colorimetric methods using a
spectrophotometer (WTW photolab 6600) and standard cuvette tests. The daily composite samples were
analysed according to Table 6. Additional analyses of daily composite samples or grab samples were also
done in order to further observe the process (for example measurements of NO2-N during disturbances) and
to calibrate process instruments.

Table 6. Internal analyses on daily composite samples from effluent water samples.
Weekday
Analysis Monday Wednesday Friday
EFF NH4-N ×
EFF NO3-N × × ×
EFF TN ×
EFF PO4-P × × ×
EFF TP ×

Sludge phase samples were analysed regarding total solids (TS (%)) and volatile solids (VS (%)) between 2-3
times per week. This regards to all different sludges; primary sludge, waste activated sludge, mixed sludge,
thickened mixed sludge, digested sludge and dewatered and digested sludge. The reject water from sludge
thickening and sludge dewatering was analysed onsite with the same approximate frequency regarding total
suspended solids (mg/L). To monitor the digestion process, a sample from the digester was taken at least once
per week and pH, VFA (mg CH3COO-eq/L), alkalinity (mg CaCO3/L) and ammonium (mg NH4-N/L) were
analysed. Measurements of methane, carbon dioxide and hydrogen sulphide in the produced biogas was
conducted several times per week with a hand-held gas meter (Sewerin Multitec 54).

29
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

5.2 Online measurements

The process was controlled and/or monitored with several online sensors installed in the treatment line.
Dynamic values from online measurements supplemented information from the analysis results and were
used for continuous follow-up and control of the process. A summary of the most important online
measurements is shown in Table 7 and Table 8. In addition to online sensors, there was also an online analyser
for PO4-P sampling from the effluent.

Table 7. Placement of online sensors in the MBR pilot.

Placement Parameter Function
General Flowrate (water) Measure all recirculation flows
IN Temperature Measure the incoming wastewater temperature
IN Flowrate (water) Measure the influent water flow
IN SS Monitor influent suspended solids concentration
PTW NH4-N Measure incoming ammonium concentration
BR1 DO Monitor Dissolved Oxygen
BR2 DO Monitor Dissolved Oxygen
BR2 NH4-N Measure ammonium concentration into aerated part of biological
treatment. Sometimes used for control of aeration.
BR3 DO Controlling Dissolved Oxygen
BR3 Flowrate (air) Measure air flow
BR4 DO Controlling Dissolved Oxygen
BR4 Flowrate (air) Measure air flow
BR4 SS Measure suspended solids concentration in activated sludge
BR5ox DO Controlling Dissolved Oxygen
BR5ox Flowrate (air) Measure air flow
BR5deox NO3-N Measure nitrate concentration, monitor function of post-denitrification
BR5deox NH4-N Measure ammonium after aerated part of biological treatment, monitor
function of nitrification. Sometimes used for control of aeration.
BR5deox DO Monitor Dissolved Oxygen
BR6 NO3-N Measure nitrate concentration, control dosage of external carbon
BR6 pH Measure pH in the biological treatment
MT1/MT2 Temperature Measure temperature in membrane tank (x2)
MT1/MT2 DO Measure Dissolved Oxygen in membrane tank (x2)
MT1/MT2 Pressure Level and pressure measurements for calculation of TMP (4 sensors)
MT1/MT2 Flowrate (water) Effluent flow of permeate from membrane 1 and 2 (x2)
MT1/MT2 Flowrate (air) Measure air flow (x2)
MT1/MT2 pH Monitor pH, especially during membrane cleaning
RAS-deox SS Measure suspended solids concentration
RAS-deox DO Monitoring Dissolved Oxygen
RAS-deox NH4-N Measure ammonium concentration (after addition of reject water)
EFF PO4-P Measure effluent phosphate concentration and control dosage of
precipitation chemicals
EFF NO3-N Measure effluent nitrate concentration
EFF NH4-N Measure effluent ammonium concentration

30
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Table 8. Placement of online sensors in sludge treatment line.

Placement Parameter Function
PS Flowrate Measure flowrate of primary sludge
PS TS Measure total solids in primary sludge
WAS Flowrate Measure flowrate of waste activated sludge
MS TS Measure total solids in mixed sludge, used to control dosage of polymer to
sludge entering the thickener
TMS TS Measure total solids in thickened mixed sludge
AD Temperature Monitor temperature in anaerobic digester, used to control heating of sludge
AD Level Measure the level in the anaerobic digester, used to test variable volumes
AD Pressure Measure the pressure of the gas
AD pH Monitor pH in the anaerobic digester
AD Radar Monitor foam in anaerobic digester
DMS TS Measure total solids in digested mixed sludge, used to control dosage of
polymer to sludge entering the dewatering

5.3 Evaluation parameters

5.3.1 Membrane performance
The membranes were evaluated using several parameters described in this section. The membranes are
operated in cycles with 15 minutes of permeate withdrawal and 1-minute relaxation (extended operational
cycle, standard operational cycle is 10+1 minutes). The membrane performance parameters can be calculated
as gross values (using only data from the 15 minutes of actual permeate withdrawal) or as net values (using
average data from the full operation cycle, permeation and relaxation = 16 minutes). The gross values are
higher than the net, however the net corresponds better to the average operation. All values for the
parameters described below are given as net values in this report.

Flux: Flowrate per membrane area, unit L/(m2·h). The flux is describing the load on the membranes. Flux is
calculated as permeate flow divided by membrane area.

TMP: Transmembrane pressure, unit mbar. The difference in pressure before and after the membranes, this
can be compared to filter resistance if TMP can vary. TMP is the driving force for transportation through the
membrane and measured using online pressure transmitters in the membrane tank and on the permeate pipe.

Permeability: Flux per TMP, unit L/(m2·h·bar). Permeability is a measurement of how well a certain flux is
withdrawn through the membranes. The permeability is gradually decreasing with time due to fouling.

The permeability is affected by the temperature. Because of this, temperature compensated permeability
(normalised to a standard temperature of 20°C) was used for evaluation. The normalisation equation is shown
below and was provided by the membrane supplier.

𝐿𝐿
𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 𝑝𝑝𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 � 𝑎𝑎𝑎𝑎 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 20°𝐶𝐶 � = 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 ∙ 𝜃𝜃 (20−𝑇𝑇)
𝑚𝑚2 ∙ ℎ ∙ 𝑏𝑏𝑏𝑏𝑏𝑏

where 𝑇𝑇 = 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇; 𝜃𝜃 = 1.025 𝑖𝑖𝑖𝑖 𝑇𝑇 ≥ 20 °𝐶𝐶; 𝑎𝑎𝑎𝑎𝑎𝑎 𝜃𝜃 = 1.033 𝑖𝑖𝑖𝑖 𝑇𝑇 < 20 °𝐶𝐶

31
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

5.3.2 Sludge quality

In addition to the parameters analysed at the external and internal laboratory as listed in Table 5 and Table 6
above, a number of additional analyses were performed on the sludge from the RAS-deox. These included
sludge volume index (SVI), Time To Filter (TTF), colloidal TOC (cTOC) and trash content.

Sludge volume index (SVI)

Sludge volume index was analysed according to APHA's standard method (2005) with dilution of the sludge
as described by Svenskt Vatten (2010b).

Time To Filter (TTF) and colloidal TOC (cTOC)

TTF was analysed according to instructions from the membrane supplier. 25, 50 and 100 mL (TTF-25, TTF-50
and TTF-100 respectively) of the sludge was filtered through 1.5-micron filter (particle retention 1.5 µm) and
the filtration time was noted. The amount of colloidal TOC (cTOC) in permeate, was defined as the difference
of TOC between the permeate and the filtrate after TTF-test. The filtrate was sent to the external laboratory for
analysis with respect to TOC (mg/L) and compared to TOC analysis from daily composite sample in effluent
(permeate). The concentration of colloidal TOC (cTOC) should be less than 10 mg/L (membrane supplier).

Trash content
The method for defining the trash content is described in detail in a previous report (Andersson et al., 2017).
In short, the sludge is filtered through screens with different slot widths and the amount of trash captured in
the screens is measured. This analysis was carried out in order to assure that particles larger than 2 mm,
which could harm the membranes, would not accumulate in the treatment line. For a well-functioning
process, the amount of trash content in the sludge, at a screen size of 2 mm, should not exceed 2 mg/L
(according to the membrane supplier).

5.3.3 Anaerobic digestion

Organic degradation rate (ODR)
The organic degradation rate (% of VSin) was calculated based on the mass flow of VS (kg VS/d) into the
digester (mixed sludge, MS) and out of the digester (digested sludge, DMS) using the equation below:

(𝑉𝑉𝑉𝑉𝑀𝑀𝑀𝑀 −𝑉𝑉𝑉𝑉𝐷𝐷𝐷𝐷𝐷𝐷 )
𝑂𝑂𝑂𝑂𝑂𝑂 =
𝑉𝑉𝑉𝑉𝑀𝑀𝑀𝑀

Specific biogas and methane production

The specific biogas production was calculated in two ways, either based on the amount of VS that was fed to
the digester, VSin, or the amount that was degraded in the digester, VSdeg. The values that are primarily
evaluated in this report is m3/kg VSin, which was calculated using the daily biogas production (m3/d) divided
by the amount of VS fed to the digester per day (VSMS). The specific methane production was calculated in the
same way, but the daily biogas production was multiplied with the percentage of methane in the gas.

Theoretical biogas production

Periodically there has been problems with the accuracy of the gas flow meters. Therefore, the theoretical
biogas production was calculated as a reference to control and sometimes replace the measured values.
Theoretical biogas production was calculated by multiplying the amount of degraded VS per day with a
factor. For biogas the factor was 0.8 m3/kg VSdeg. The factor was based on results from previous trials and
corresponded well with values foud in literature.
32
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

6 Results and discussion

6.1 Primary treatment
Trial J F M A M J J A S O N D
37% increased inflow
Imitation of first phase operation of full-scale

6.1.1 Inlet screen

Different punch hole screens, from 2 mm to 6 mm, have been used for the initial screening since the start of
the project. Since 2017 a 6 mm size was been used to increase the primary sludge production which was
required to meet the design load for the sludge pilot. As Figure 9 indicates, inlet suspended solids
concentrations corresponded, after that change, well to data from Henriksdal WWTP, especially during 2017
and 2018. During 2019, there were frequent maintenance work done in the Henriksdal digesters, which
include that the sludge is emptied into the inlet channel where the inlet pumps to Hammarby Sjöstadsvek are
located. This resulted in higher suspended solids concentrations in the pilot inlet compared to Henriksdal
(and previous years). Figure 9 shows that the inlet SS-concentration to the pilot decreased in 2020 although the
6 mm screen was used. The main contributing factor was considered to be the corona pandemic which
resulted in less people in the central parts of Stockholm, from where the wastewater to the pilot originated
(Danviken inlet).

SS in incoming wastewater
500

400

300
mg SS/L

200

100

0
2014 2015 2016 2017 2018 2019 2020

Pilot 3 mm Pilot 2 mm Pilot 6 mm Henriksdal

Figure 9. Incoming SS-concentration to the pilot after passing through 2-, 3- or 6-mm punch hole screens
and to Henriksdal (3-mm step-screen).

33
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

6.1.2 Efficiency of primary settler

The primary settling volume in the pilot (3.3 m3) and its capacity, is smaller than it should be (design value 4.5
m3). This results in a poor reduction of solids and organic material compared to Henriksdal WWTP, and
insufficient primary sludge production. The reduction rate over the primary settler is showed Table 9 and
Figure 10 below. During 2020, two additional aspects affected the efficiency of the primary settler. First, the
inflow to the pilot was high, 4.4 m3/h compared to the design average flow of 3.2 m3/h. This resulted in a
shorter reaction time for phosphorous coagulation as well as shorter settling times and thus even lower
reduction compared to previous years, but higher total primary sludge production. Second, the corona
pandemic affected the composition of the influent wastewater resulting in lower inlet concentrations.

Table 9. Reduction over the primary settler and primary sludge production in the pilot compared to
Henriksdal WWTP 2020 and the future design of Henriksdal WWTP.
Henriksdal Design future
Parameter 2020 2019 2018 2017 WWTP 2020a Henriksdal
Inflow (m3/h) 4.37 3.56 3.50 3.40 12 400 22 000
PE (1 PE = 70g BOD7/d) 292 352 318 314 850 000 1 600 000
Reduction of SS 32% 36% 35% 37% 56% 60%
Reduction of BOD7 19% 27% 25% 30% 54% 50%
Reduction of TP 8% 12% 10% 14% 33% 40%
Reduction of TN 0% 4% 1% 4% 2%b 10%
Reduction of TOC 10% 20% 18% 17% 37% -
PS-production (kg TS/d) 21.7 18.5 16.1 13.1 61 000c 115 000
PS-production (g TS/PE/d) 74 53 51 42 73 72
a) At high flows Enhanced Pre-Precipitation using Al3+ is used for half of the primary clarifiers.
b) New sampling point for the Sickla inlet (SIN), no longer including the reject water from digested sludge dewatering.
c) Based on uncertain TS-measurements.

Higher concentrations of solids, organic material and nitrogen after the primary settler than in incoming
wastewater have been observed at certain times, mainly after w45 (Figure 10). No obvious cause could be
identified, but the installation of new samplers during this period may partly be responsible. This aspect will
be further investigated.

Flow and SS reduction over primary settler

400 120
% reduction, inflow m3/d

350 105
300 90
250 75
mg SS/L

200 60
150 45
100 30
50 15
0 0
w1
w3
w5
w7
w9
w11
w13
w15
w17
w19
w21
w23
w25
w27
w29
w31
w33
w35
w37
w39
w41
w43
w45
w47
w49
w51
w53

inflow SS reduction SS in SS out

Figure 10. Reduction of SS over the primary settler after a 6 mm screen.

34
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

6.1.3 Screen and sieve – effect on trash content

During 2020 the pilot line was operated with a 6 mm punch hole inlet screen (see 6.1.1 Inlet screen) and at first
a 0.6 mm, and later a 2 mm, mesh fine sieve before the biological treatment. Over the years, different screen-
sieve configurations have been tested. In order to monitor the amount of particles, fibres and hair that
accumulate in the activated sludge with potential to cause problems in the membrane tanks, analysis of trash
content (see 5.3.2 Sludge quality) was made. Results are presented in Table 10.

The results from previous years have been consistent when the 6 mm + 0.6 mm screen/sieve configuration was
implemented. In addition, visual inspection of the membrane cassettes in 2018 showed very little build-up of
trash, indicating that the measured values are accurate. On June 24th, 2020, the 0.6 mm fine sieve before the
biological treatment was replaced with a 2 mm fine sieve. The 0.6 mm sieve was initially selected (instead of
the recommended 2 mm) to provide further protection of the membranes and to evaluate the amount of
screenings removed by the fine sieve. Not much material was removed by the sieve and the trash-content in
the sludge was low which led to the conclusion that larger hole size, which is preferable from a hydraulic
point of view, can be used. In July the 0.6 mm drum was changed to one with 2 mm holes, which is the size
that will be used in the full-scale plant. Surprisingly, the measurements from 2020 indicate less amount of
trash content in the sludge although both screen (6 mm) and sieve (2 mm) are the largest sizes tested.

Table 10. Trash content in waste activated sludge (WAS) with various sieve-configurations. According to
the membrane supplier, the trash content with a 2 mm sieve should not exceed 2 mg/L.
No of 1 mm sieve 2 mm sieve
Sieves and hole size Dates analyses mg/L mg/L
3 mm screen at inlet pump Dec 2013 2 11.6 ± 5.4 1.0 ± 0.7
2 mm fine screen at inlet pump Nov 2016 – Feb 2017 4 6.4 ± 2.4 1.1 ± 0.7
6 mm screen at inlet pump and
Mar 2017 → Dec 2017 8 4.1 ± 3.8 0.6 ± 0.3
0.6 mm fine sieve before biology
6 mm screen at inlet pump and
May 2017 → Dec 2017 6 2.2 ± 1.6 0.6 ± 0.3
0.6 mm fine sieve before biology
6 mm screen at inlet pump and
Jan 2018 → Nov 2018 6 2.0 ± 1.1 0.6 ± 0.5
0.6 mm fine sieve before biology
3 mm screen at inlet pump and
Dec 2018 – Mar 2019 1 2.3 2.0
0.6 mm fine sieve before biology
6 mm screen at inlet pump and
Apr 2019 – Dec 2019 2 5.1 ± 5.4 1.5 ± 2.0
0.6 mm fine sieve before biology
6 mm screen at inlet pump and
Jan 2020 – June 2020 1 0.4 0.2
0.6 mm fine sieve before biology
6 mm screen at inlet pump and
July 2020 – Dec 2020 3 1.4 ± 0.5 0.3 ± 0.2
2 mm fine sieve before biology

6.1.4 Pre-treated wastewater

The quality of the pre-treated wastewater (PTW) is presented in Table 11. The concentrations measured in the
pilot were significantly lower compared to previous years, however still higher than the corresponding
concentrations measured in Henriksdal WWTP for SS, BOD7 and TP. This is mainly due to the poor
performance of the primary settler and was also observed previous years. The concentrations of SS, BOD7 and
TP in pre-treated water were also higher in the pilot than the design values given for the future plant, which
though are relatively low since they are determined a weighted average between Henriksdal WWTP and
Bromma WWTP, and Bromma WWTP has a diluted inflow.

35
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

The concentrations of nutrients and organic matter in pre-treated wastewater will affect the biological
treatment, including WAS production (and thereby SRT and the amount of phosphorus assimilated in
sludge), pre-denitrification capacity and the need for simultaneous precipitation. The difference in
concentrations between the pilot and the future Henriksdal design thereby will affect comparison of the
evaluation parameters related to these aspects. It can be noted that the iron dosage in the primary settler was
low in the pilot compared to the full-scale and the future Henriksdal design. This is due to enhanced
biological phosphorus removal (EBPR), which is described in chapter 6.3.

Table 11. Data on PTW from the pilot compared to data from Henriksdal 2020 and the design data for the
future Henriksdal WWTP.
Value Henriksdal Design future Value pilot/
Parameter Value Pilot 2020 2020 Henriksdal Design future H-dal
Pre-treated wastewater (PTW) – into biological treatment
SS (mg/L) 145 ± 50 135 113 128%
BOD7 (mg/L) 159 ± 35 99 108 147%
TN (mg/L) 39 ± 5 39 33 118%
TP (mg/L) 4.2 ± 0.8 3.3 3.0 140%
Fe (mg/L) 5.8 ± 1.9 10.4 12 48%
Alkalinity (mg/L) 251 250 250 100%
BOD7/TN (mg/mg) 4.1 2.5 3.3 124%

6.2 Nitrogen removal

Trial J F M A M J J A S O N D
Cooling in the biology
Glycerol as carbon source
Ethanol as carbon source
Imitation of first phase operation of full-scale
Reject water recirculation to pre-aeration
Reject water recirculation to RAS-deox

Nitrogen concentrations in the incoming water to the biological treatment (PTW, primary treated water) and
in the effluent are presented in Table 12. On average the effluent total nitrogen concentration was 3.9 mg/L.
Four out of 53 weekly composite samples were above the target concentration of 6 mg N/L. The reduction of
total nitrogen over the biological treatment step was 90%.

Table 12. Nitrogen concentrations in primary treated water (PTW) and effluent during 2020.
Parameter Limit Average Min Max No. of weekly samples
TN PTW (mg/L) - 39 28 52 53
TN EFF (mg/L) 6 3.9 1.7 11.0 53

Effluent nitrogen concentrations as weekly composite samples are presented in Figure 11. During most of the
year, nitrogen removal has been satisfactory with stable effluent concentrations below 6 mg/L, as also shown
previous years. In December (week 50-51), problems with the nitrification occurred after recovery cleaning of
the membranes with sodium hypochlorite. The resulted in that the effluent TN reached 11 mg N/L. This has
never been observed before. However, reduced nitrification capacity around that time of the year has been
observed every year. Nitrification though quickly recovered and by the end of the year effluent ammonium
was below 0.2 mg N/L.
36
 Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Incomming and effluent nitrogen

60 18
50 15
40 12

mg N/L eff
mg N/L in

30 9
20 6
10 3
0 0
w1
w3
w5
w7
w9
w11
w13
w15
w17
w19
w21
w23
w25
w27
w29
w31
w33
w35
w37
w39
w41
w43
w45
w47
w49
w51
w53
TN in TN eff TN limit NH4-N eff NO3-N+NO2-N eff

Figure 11. Incoming and effluent nitrogen concentrations from analysis of weekly composite samples
during 2020. Limit for effluent total nitrogen was set to 6 mg N/L.

In Table 13, key values for the nitrogen removal are presented for the MBR pilot and compared to the future
Henriksdal design. The amount of removed total nitrogen during 2020 (3.7 kg N/d) was the slightly higher
compared to 2018 and 2019 (3.6 kg N/d) and somewhat higher compared to the full-scale design. Different
external carbon sources have been tested during this year. Glycerol dosage was used for the majority of the
year and the average consumption per day was 1.57 kg COD/d. Details about the other external carbon
sources can be found in section 6.2.2 Denitrification. The nitrogen removal rate presented in Table 13 was
similar in the pilot and the future Henriksdal design.

The airflow to the pilot was much larger than the estimated for the future Henriksdal WWTP (Table 13). One
reason for the big differences in airflow is that the depth in the basins of the pilot were one fourth of the depth
in Henriksdal. The aeration of the biology in the pilot was also much higher in 2020 (72 m3/h), compared to
previous years, when average aeration has been around 50-54 m3/h (more about this in chapter 6.2.1).

Consumption of external carbon source was more than double this year using glycerol compared to last year
with methanol, which can partly be explained by the high inflow and thereby higher nitrogen load.

Table 13. Comparison of parameters related to the nitrogen removal between operational data from the
pilot and the design for future Henriksdal (2040).
Value pilot/
Value Future H-dal scaled future
Parameter Unit Pilot design H-dal designi
Removed nitrogen (incl. reject water) kg N/d 3.7 21 000 118%
Nitrogen removal rate g N/kg VSS, d 17.65 17.6 100%
Aerated sludge age (incl. MTs) d 6.5 9.4ii 69%
Aeration biology (activated sludge) Nm /h
3 72.3 iii 68 000 712%
Specific oxygen demand (SOTR) kg O2/d 83 iv 240 000 232%
Consumption of external carbon kg COD/d 1.7v 12 000 97%
i The value of future H-dal design divided by the scale factor 6 700
ii Assumed that ¾ of all membrane tanks are in operation as a yearly average.
iii m3/h not Nm3/h

iv SOTR was calculated from the measured airflow and a water depth (aerator surface to water surface) of 3.19 m and a specific
oxygenation capacity of 0.015 kg O2/Nm3, m.
v Only including methanol (week 1 to week 34).

37
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

6.2.1 Nitrification
Nitrification worked satisfactory most of the year with low effluent concentrations of ammonium except for
shorter periods and three peaks in effluent ammonium (w.11, w.36 and w.50-51). When comparing the
aeration of the biology with aeration of the membrane tanks (Figure 12) it can be observed that the airflow to
the biology was more than double the airflow to the membranes 47 out of 53 weeks. The membranes were
operated at the lower aeration level (Leap-Lo corresponding to 14 m3/h each, 28 m3/h in total) most of the time,
with only a few peaks above 28 m3/h. The airflow to the biology varied as weekly average between 31.5 and
127 m3/h and was 72 m3/h as yearly average.

A
Aeration
150 12
125 10
100 8

mg/L
m3/h

75 6
50 4
25 2
0 0
w1
w3
w5
w7
w9
w11
w13
w15
w17
w19
w21
w23
w25
w27
w29
w31
w33
w35
w37
w39
w41
w43
w45
w47
w49
w51
w53
Aeration Bio (m3/h) Aeration MT (m3/h) NH4-N eff (mg/L)

B 150 12

125 10

100 8
mg/L
m3/h

75 6

50 4

25 2

0 0
A N J O M

2016

Inflow Aeration Bio (m3/h) Aeration MT (m3/h) NH4-N eff (mg/L)

Figure 12. A) Aeration need in biology and membrane tanks (MT) together with effluent NH4-N during
2020. B) Aeration, influent flowrate and effluent NH4-N from 2016.

Table 14 shows the annular average airflows and DO levels in the aerated tanks year 2016-2020 together with
the calculated theoretical airflow demand. As can be seen, the airflow year 2020 was higher than previous
years. To some extent this can be explained by the high inflow resulting in a higher load of nitrogen and BOD
on the biology. The inflow 2020 was 37% higher than design average flow and 26% higher than 2019.
Nevertheless, the load of TN to the biology only increased by 12% and the BOD load was the same in 2020 as

38
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

in 2019. The airflow, on the other hand, was 49% higher in 2020 than in 2019. When calculating the airflow
demand, using standardized design calculations, it can be seen in that the calculated value is within 10%
difference from the measured value year 2016-2019 (Table 14). In 2020, however, the deviation is 38%, which
clearly indicates some type of problems with the aerators, such as fouling or broken membranes, problems
with the airflow meters or, less likely, some other type of unknown interference with aeration. The influence
from the increased dose of external carbon source compared to previous years did not affect the airflow
demand since it was consumed in the post denitrification zone, or in case of overdosing, in the membrane
tanks. The possible effect of the BOD load from external carbon on the airflow demand was also investigated
by adding the external carbon source to the incoming BOD load to the biology and recalculate the airflow
demand. The result show that the airflow demand increased with 4% if all the external carbon went to the
aerated zones, which would still not explain the 38% deviation. Because of the exceptionally high measured
airflow, it was decided to exchange the aerator membranes in the beginning of 2021.

Table 14. Airflow and DO in the different zones year 2016-2020. Calculated airflow demand is based on the
load of BOD and TN in PTW, as well as DO and temperature in the biological reactors for each year.
Calc.
Airflow DO Airflow DO Airflow DO Airflow Airflow
Inflow BR3 BR3 BR4 BR4 BR5 BR5 Tot Tot
Year (m3/h) (m3/h) (mg/L) (m3/h) (mg/L) (m3/h) (mg/L) (m3/h) (m3/h)
2020 4.38f 23.2 1.42 38.7 2.71 10.5 1.67 72.5 52.6
2019 3.48d 18.3 1.45 16.2 2.64 13.9 2.00 48.5 51.8
2018 3.52d/f 17.3 1.75 19.1 1.98 15.9 1.60 52.3 47.6
2017 3.47d/f 20.9 1.91 20.4 2.89 10.0 2.19 51.3 51.4
2016 2.84d 6.0 0.78 20.6 2.74 11.2 3.70 37.9 42.2
f - fixed flow
d -dynamic flow

The first period with incomplete nitrification in week 11 was related to low DO setpoints (< 1 mg/L) in the
treatment line, especially BR3. This was done as a measure to reduce aeration to limit the overflow of sludge
due to foaming. The same problems with foaming sludge occurred in week 35 and again the aeration was
reduced, and effluent ammonium increased.

In week 50, directly after startup from recovery cleaning (RC) with sodium hypochlorite, nitrification was
significantly affected and the ammonium concentration in the treatment line and effluent increased (Figure
13). After a couple of weeks, the process had recovered, and effluent ammonium was low again. This was the
12th recovery cleaning since 2016 and never before has the nitrogen removal been affected. No clear
differences in the cleaning or neutralization procedure were observed and no other explanation has been
found.

39
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Nitrification inhibition during startup from hypo-RC

20 5

4
15
mg NH4-N/L

m3/h
10
2
5
1

0 0
2020-11-30 2020-12-05 2020-12-10 2020-12-15 2020-12-20 2020-12-25 2020-12-30

Effluent NH4-N (online) Effluent NH4-N (daily composite sample)

Permeate flow rate MT1 Permeate flow rate MT2

Figure 13. Nitrification inhibition after startup from recovery cleaning (RC) with sodium hypochlorite.

On average, the aeration of the membranes accounted for 28% of the total aeration, previous years (2017 to
2019) this number was about 36%. However, the lower value for 2020 is mainly caused by an increase in
aeration of the biology. Membrane aeration has been significantly reduced compared to the first year (2016)
when 54% of the total aeration was used for the membranes and the average membrane aeration was 42 m3/h.

As previous years, problems with foaming and unreliable DO sensor readings in the foamy sludge have made
aeration control difficult at times. BR4 and BR5ox was aerated to keep a manually selected DO setpoint
between 1 and 4 mg/L throughout the year. Aeration of BR3 was using a flow-based control, where aeration
was on (with DO setpoint 2.5 mg/L) only when flowrates to the treatment was above 2.5 m3/h to prevent
excessive aeration at night when inflow was low. Due to problems with foaming, aeration of BR3 was
manually operated from week 11 to week 23. From week 23 aeration was again automatically controlled, this
time based on effluent ammonium concentration. Aeration of BR3 to DO setpoint (varying between 1.0 - 2.5
mg/L) was triggered if effluent ammonium was above target concentration (varying between 0.2 to 1.0 mg
NH4-N/L).

The initial aim for the pilot was to operate at a total sludge age of 25 days, to evaluate the performance at
design conditions. However, as the membrane supplier terms states that the membranes should not be
operated in sludge concentration above 10 000 mg SS/L for longer periods, the WAS outtake was determined
based on the MLSS concentration in the line and the sludge age was not actively controlled and has only been
monitored.

Waste activated sludge (WAS) flowrate has been automatically controlled using a feedback controller to keep
the suspended solids concentration in BR4 at setpoint 8 000 mg SS/L. The resulting total and aerated
(including membrane tanks) sludge age is presented in Figure 14 together with the WAS flowrate. The peaks
in sludge age around week 12 and week 28 are the result of lowering the WAS-pumping to maintain 8 000 mg
SS/L in the biological treatment. The calculated total sludge age was on average 17.5 days, which is lower than
future Henriksdal design (25 days). This is partially due to the higher load of SS and BOD to the biology
compared to the Henriksdal design. However, the calculations of sludge age in the pilot are uncertain because
of foaming in the aerated bioreactors leading to overflow and loss of sludge not accounted for in the
calculations. The calculated aerated sludge age (including membrane tanks) was on average 6.5 days.

40
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Sludge age
30 100

25 80
20 60

L/h
15
d

40
10
5 20

0 0
w1
w3
w5
w7
w9
w11
w13
w15
w17
w19
w21
w23
w25
w27
w29
w31
w33
w35
w37
w39
w41
w43
w45
w47
w49
w51
w53
Qwas (L/h) Aerated SRT incl. MT (d) Total SRT (d)

Figure 14. Total and aerated (incl. MTs) sludge age (moving average one month back in time) together with
WAS flowrate.

6.2.2 Denitrification
Both pre- and post-denitrification is utilized in the pilot. For pre-denitrification (BR1, BR2 and sometimes BR3)
the nitrate recirculation flow rate (from BR5deox to BR1) has been flow proportional to the inflow using a
factor of 3 x Qin, except for a short period in January/February (6 days) when the pump was out of operation
and a month-long trial using 1 x Qin (September). A maximum flow of about 13 m3/h have been used as this
corresponds to the maximal flow in the full-scale. The concentration of NO3-N in the recirculation flow was
1.4 – 6.2 mg/L (average 4.2 mg/L) according to the online sensor. In addition, nitrate was recirculated with the
RAS (from RAS-deox to BR1) with a flow corresponding to 4 x Qin. The nitrate concentration in the RAS was
0.8 - 4.9 mg NO3-N/L with 2 mg/l as yearly average value (based on effluent online sensor).

For the post denitrification (BR6) different external carbon sources have previously been used. This year
glycerol was used, except for a 10-week trial (w.14 to w.23), when ethanol was tested. The reason for testing
ethanol was mainly for evaluation of the enhanced biological phosphorus removal (EBPR), as ethanol is
considered a suitable carbon source for EBPR. The external carbon source has been added to a point in the
piping between BR5 and BR6 and the dosage has been controlled by the online nitrate concentration.

Due to sensor problems, the signal used for dosage control has been alternating between the nitrate sensor in
BR6 and the nitrate sensor in the effluent. The sensor readings from BR6 were preferred as these provide a
faster control strategy. However, due to foaming and sensor failure, the effluent nitrate sensor was used from
time to time. The nitrate setpoint was 2 mg NO3-N/L until June when glycerol was again used and the setpoint
was lowered to 1 NO3-N/L. From w.41 onward the setpoint was increased to 3 mg NO3-N/L.

An overview of the external carbon source addition is given in Figure 15. The effluent nitrate concentration
was possible to control and keep below 3 mg NO3-N/L with both glycerol and ethanol. The 10-week long test
with ethanol showed a similar consumption of COD as for glycerol. A more detailed evaluation of the
denitrification potential of ethanol is presented in the next section (6.2.3 Comparison of different external
carbon sources). During week 36 the nitrate recirculation was decreased from 3 x Qin to 1x Qin in order to
induce an increased carbon source dosage. The objective with this test was to evaluate any relation between
glycerol dosage and cTOC. It is clear from Figure 15 that the lower nitrate recirculation resulted in an
increased carbon dosage but not lower effluent nitrate, as expected. In week 41 the nitrate recirculation was
readjusted to 3 x Qin and carbon source dosage was stopped until w.48 due to specific trials.
41
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Carbon source addition

6 120

5 100
mg NO3-N/L

4 80

mg COD/L
3 60

2 40

1 20

0 0
w1
w3
w5
w7
w9
w11
w13
w15
w17
w19
w21
w23
w25
w27
w29
w31
w33
w35
w37
w39
w41
w43
w45
w47
w49
w51
w53
Dosage Glycerol (mg COD/L) Dosage Ethanol (mg COD/L)
Nitrate Effluent mg/L

Figure 15. Dosage of glycerol and ethanol (mg COD/L incoming wastewater) as daily average values and
effluent nitrate analysed in weekly composite samples.

The online sensors for nitrate in BR5deox and BR6 were helpful in the daily operation to monitor the
denitrification. However, it was difficult to use the data for estimation of how much nitrogen was denitrified
in the post-denitrification as the sensors tended to drift and the error in the reading was close to the difference
in measured concentration. The weekly average concentration of nitrate, as measured by the sensors, in
BR5deox, BR6 and in the effluent is presented in Figure 16. The concentrations in BR5deox varied between 1.4
and 6.2 mg NO3-N/L while the concentration in BR6 varied between 0.6 and 3.7 mg NO3-N/L. The NO3-N
concentration in BR6 was similar to the effluent reading, sometimes higher and sometimes lower. Although
effluent nitrate can be higher than in BR6 due to nitrification in the membranes, the difference is more likely
due to sensor inaccuracy in BR6. Note that week 42 to 49, when there was no external carbon source addition,
effluent nitrate was higher than normal, and higher than nitrate readings in BR6. This is also explained by
faulty sensor readings in BR6.

42
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

No external
Nitrate concentrations carbon source
Glycerol Ethanol Glycerol Glycerol
7
6
5
mg NO3-N/L

4
3
2
1
0
w1
w3
w5
w7
w9
w11
w13
w15
w17
w19
w21
w23
w25
w27
w29
w31
w33
w35
w37
w39
w41
w43
w45
w47
w49
w51
w53
Nitrate BR5 (mg/L) Nitrate BR6 (mg/L) Nitrate eff online (mg/L)

Figure 16. Online nitrate concentrations as weekly average.

6.2.3 Comparison of different external carbon sources

The nitrogen treatment process design for future Henriksdal WWTP is based on methanol as external carbon
source for post-denitrification. Methanol was chosen since it is a well-known, well-functioning carbon source
that is available in large quantities at low cost. The carbon source storage facility for the full-scale treatment in
Henriksdal will not be in finished during the first phases of operation of the full-scale plant and therefore a
temporary installation will be needed at Henriksdal during a few years. With this background, glycerol has
been tested as an alternative external carbon source in the pilot during 2020. A shorter test (10 weeks) was also
done using ethanol as carbon source as part of evaluation of enhanced biological phosphorus removal.

Pilot results
Although the operational conditions in the treatment line have varied over time, an attempt to compare the
specific consumption of carbon source for post-denitrification has been made. Based on the online data of
COD dosage and nitrogen removal (NO3-N removal over the post-denitrification zone is labelled “g N-red”,
and TN-removal over the whole biological treatment is labelled “g N”), the specific consumption of carbon
source as g COD/g N or g COD/g N-red was calculated. Results for glycerol and ethanol from 2020 are shown
in Figure 17 and Table 15, previous results using other external carbon sources can be found in Andersson et
al. (2021).

The theoretical COD-consumption for denitrification is 2.86 g COD/g N-red (Metcalf & Eddy, 2014). Normally
the meassured specific COD-consumption is higher than the theoretical value since some of the added carbon
is utilized for microbial cell growth (sludge production) and in cases when all oxygen is not removed in the
deox zone some of the COD might also be oxidized. In the case of the pilot trials, the calculated specific COD
consumption for the different carbon sources also include denitrification utilizing internally produced carbon
from hydrolysis, something that occures all the time, also during periods with no addition of external carbon.
The numbers presented should therefore not be compared to litterature values but can still be used to evaluate
the relative performance from the different carbon sources.

43
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

As can be seen in Figure 17, the specific COD consumption was higher for ethanol compared to glycerol
when only considering the nitrgen removed in the post-denitrification (2.91 compared to 1.84 g COD/g N-
red). When relating the COD consumption to the total nitrogen removal the consumptions were similar for
the two carbon sources.

Specific carbon source consumption

3.0
2.5
2.0
1.5
1.0
0.5
0.0
Ethanol Glycerol

2020 2020

Total N-red (g COD/g N) N-red in post DN (g COD/g N-red)

Figure 17. Comparison of COD consumption for ethanol and glycerol used in the pilot during 2020.

When looking at some of the operating conditions during the trials, both effluent nitrate and total nitrogen
removal was lower when using glycerol (see Table 15). This is explained by a higher load (3.6 kg TN/d)
entering the biological treatment in the period when ethanol was used compared to the period with glycerol
(when the load was 3.2 kg TN/d).

A comparison between glycerol and ethanol in the pilot during February to April 2020 was done by Roberts
(2020) and this evaluation also found the consumption of the two carbon sources to be comparable.

Table 15. Carbon source dosage and nitrogen removal for trials using ethanol and glycerol during 2020.

kg NO3-N Effluent
kg red in post Total kg Temperature NO3-N
Carbon source Weeks COD/d DN/d N-red /d biology (°C) (mg/L)
2020 Ethanol 9 1.70 0.6 3.3 15.6 2.1
2020 Glycerol 22* 1.57 0.9 2.9 19.6 1.5
* w.41 to 48 excluded (dosage was paused).

Lab tests
In order to further compare and follow any adaption of the sludge to a new carbon source denitrification
batch tests have been carried out. The methodology used for the denitrification batch tests is described by van
Loosdrecht et al. (2016) and is the same as used previously in the project. Temperature compensated
denitrification rates (normalized to 20°C) for ethanol and glycerol are presented in Figure 18.

For glycerol, the denitrification rate in March 2020 was 2.4 mg NO3-N/ g VSS, h. This is similar as in
September 2019 when glycerol was tested and showed a denitrification rate of 2.1-2.4 mg NO3-N/g VSS, h
(Andersson et al., 2021).

44
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

For ethanol, the denitrification rate increased from 3.1 to 4.7 mg NO3-N/ g VSS, h as the sludge adapted to
ethanol. The first denitrification test (March 2020) was carried out before dosage of ethanol had started in the
pilot and the second test (May 2020) was carried out after two months using ethanol in the pilot.

Denitrification rate (temp.compensated)

5 4.74
mg NO3-N/g VSS, h

4
3.11
3 2.40
2
0.96 0.96 0.90
1

0
Mar 19 2020 Mar 27 2020 May 29 2020

Ref Glycerol Ethanol

Figure 18. Denitrification rate during laboratory denitrification batch tests using glycerol and ethanol as
carbon source.

The specific COD consumption during denitrification tests varied (Figure 19), especially for the reference test
without external carbon source, and showed correlation to how denitrification was affected by pH. For
methanol the COD-consumption was between 2.1 and 2.6 g COD/g NO3-N denitrified. The COD consumption
for glycerol was much higher, 6.6 g COD/g NO3-N denitrified which is slightly higher than previous test in
September 2019 when COD consumption was 6.2 g COD/g NO3-N denitrified, which in turn is similar to
reference values for glycerol (USEPA 2013). For ethanol, the COD consumption increased from first to second
test when the sludge had adapted to ethanol. Similar results have been obtained for methanol in previous
batch tests.

The specific COD-consumption in the lab test was significantly higher than in the pilot. This is most likely due
to the intermittent dosing of external carbon source in the pilot that is controlled by the effluent nitrate
concentration. The incoming nitrogen peaks in the morning and afternoon but goes down during nighttime.
Thus, dosing of external carbon source is only active at daytime. However, at nighttime when no external
carbon source is dosed, some post denitrification still takes place using organic carbon produced from
hydrolysis of activated sludge. So, when calculating the specific COD-consumption by summarizing the dose
of external organic carbon and comparing it to the amount of nitrogen denitrified in the post-denitrification
zone, the ratio will be lower than the actual value.

45
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Specific COD consumption during denitrification

10
g COD/g NO3-N denitrified

8 6.56
6 5.00 4.40
4 2.35 2.12 2.59
1.70
2 0.12
0
-2
-4
-3.33
Mar 19 2020 Mar 27 2020 May 29 2020

Ref Methanol Glycerol Ethanol

Figure 19. Specific COD consumption during laboratory denitrification batch test in laboratory tests using
different carbon sources.

6.3 Phosphorus removal

Trial J F M A M J J A S O N D
Effluent phosphorus target 0.10 mg TP/L
Aluminum chloride as precipitation chemical
In-depth study of EBPR

The goal of reaching a stable effluent phosphorus concentration below 0.15 mg P/L as monthly and annular
averages was achieved previous years by using a control strategy with dosage of ferrous sulphate and ferric
chloride in three points in the process; one flow proportional dose prior to pre-sedimentation and two dosage
points in the biology (ferrous sulphate to BR4 and ferric chloride to BR6) controlled using feedback control
from online effluent phosphate measurements. During 2019 the effluent target was lowered to 0.10 mg TP/L
which was achieved using the same precipitation strategy but using lower phosphate setpoints. This year
(2020) the goal of 0.10 mg TP/L was met using aluminum chloride (PAX, in operation from week 7) instead of
ferric chloride (PIX) in the third dosing point. This was tested because the first MBR-line in Henriksdal will
use precipitation chemicals from the existing storm water treatment system during the first years of operation
as the PIX-tanks will not be built until a later stage of the project. The second dosage point of ferrous sulphate
to the biology was taken out of operation to increase the aluminum chloride dosage and enable a better
evaluation of potential effects on membranes and sludge quality of using an aluminum product in the process.

The phosphorous concentrations in and out from the biological treatment is presented in Table 16 and Figure
20 below. The yearly average effluent total phosphorus was 0.05 mg TP/L, the lowest since the project started,
and well below the target of 0.10 mg TP/L. It should be noted that the inlet total phosphorus was also around
18% lower than normal. Total phosphorus in primary treated water has previous years varied between 5.3
and 5.5 mg TP/L as yearly average. In Henriksdal WWTP, the TP in the inflow was around 16% lower than
previous years.

Table 16. Phosphorus concentrations in primary treated water (PTW) and effluent during 2020.
Parameter Limit Average Min Max Nr of weekly samples
TP PTW (mg/L) - 4.4 3.0 5.8 53
TP EFF (mg/L) 0.20 0.05 0.023 0.28 52

46
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Incoming and effluent phosphorus

6 0.12

EFF, Target [mg TP/L]

IN, PTW [mg TP/L]

5 0.10
4 0.08
3 0.06
2 0.04
1 0.02
0 0.00
w1
w3
w5
w7
w9
w11
w13
w15
w17
w19
w21
w23
w25
w27
w29
w31
w33
w35
w37
w39
w41
w43
w45
w47
w49
w51
w53
TP IN TP PTW TP EFF TP target

Figure 20. Influent and effluent total phosphorus analysed in weekly composite samples during 2020.
Effluent data from week 17 is considered a measurement error and was excluded from calculations.

Key parameters for the phosphorus removal, both for the pilot and for the future Henriksdal design, are
presented in Table 17. The phosphorus load in the pilot was much higher than the phosphorus load which the
future Henriksdal design is based on. This resulted in higher phosphorus removal in the biology of the pilot
in comparison to the future Henriksdal design. Despite the high phosphorus removal, the relative metal
consumption in the pilot was lower in the pilot compared to the future Henriksdal design. Although the
consumption of iron for precipitation of phosphorus was one of the most uncertain parameters in the future
Henriksdal design, the difference is supported by the observed EBPR in the pilot. In the future Henriksdal
design a yearly average dosage of 20 mg Fe/m3 treated water was assumed (sum of the three dosing points). In
the pilot, the average dosage 2020 was 7.1 mg Me/L (6.6 mg Fe/L + 0.5 mg Al/L).

Table 17. Comparison of operational data from the pilot with data for the future Henriksdal design, yearly
average values.
Value pilot Future H-dal Value pilot/scaled
Parameter Unit 2020 design Future H-dal designa
Phosphorous load influent kg P/d 0.49 2 594 127%
Phosphorous load PTW kg P/d 0.46 1 580 194%
Phosphorus load reject water kg P/d 0.01 480 16%
Total phosphorous load biology kg P/d 0.47 2 060 153%
Phosphorus load effluent kg P/d 0.0055 80 46%
Phosphorus removed in biology kg P/d 0.46 1 980 157%
Iron consumption kg Fe/d 0.7 10 000 47%
Aluminium consumption kg Al/d 0.05 0 -
Metal consumption (Fe+Al) kg Me/d 0.75 10 000 50%
Metal consumption per removed mole Me/ 0.95 2.80 34%
phosphorus mole P
Phosphorus in sludge % of SS 3.3 5.4b -
Iron in sludge % of SS 4.1 - -
Aluminium in sludge % of SS 0.5 -
VSS in sludge % of SS 80 66 -
a) Future H-dal design divided by 6 700.
b) No EBPR. Mainly chemically bound phosphorous.

47
 Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

As can be seen in Table 17, the phosphorus fraction of the activated sludge (% of SS) was lower in the pilot
than in the future Henriksdal design although the relative amount of phosphorus that was removed is larger.
This can be explained by the high WAS production (high microbial cell growth) in the pilot, caused by high
incoming load of SS and BOD. A high sludge production means that more phosphorus is bound in the sludge
which gives a high removal of phosphorus per day even though the percentage in sludge is low.

The total amount of iron dosed is presented as daily average values in Figure 21. The base dose of Fe2+ was
added to the pre-aeration which was controlled flow proportionally to a dose of 5-10 mg Fe/L (manually
adjusted based on effluent phosphate). A supplementary dosage of Al3+ was added in BR6 (just prior to the
membrane tanks). This dose was controlled by effluent phosphate feedback control and only added during
shorter peaks in effluent phosphate. Although a higher aluminium dosage was intended, it was not needed
due to very low effluent phosphate concentrations. An even lower dosage of Fe2+ to the pre-aeration would
have risked the primary sludge production, and indirectly the ongoing trials in the sludge pilot, and was
therefore not tested as an alternative to increase the aluminium dosage. The yearly average dosage of
aluminium chloride was only 0.5 mg Al/L, and the highest weekly dosage was 3.8 mg Al/L.

Dosage of precipitation chemicals

20 0.40

15 0.30
mg Me/L (Fe+Al)

mg PO4-P/L
10 0.20

5 0.10

0 0.00
w1
w3
w5
w7
w9
w11
w13
w15
w17
w19
w21
w23
w25
w27
w29
w31
w33
w35
w37
w39
w41
w43
w45
w47
w49
w51
w53

Al3+ to BR6 Fe2+ to pre aeration Effluent phosphate Effluent TP

Figure 21. Metal dosage (the sum of Fe2+ to pre-aeration and Al3+ to BR6) as daily average, effluent
phosphate analysed in daily composite samples and effluent total phosphorus in weekly composite
samples. Note that the iron and aluminium are presented as a sum of added metals (g Me/d).

The iron, aluminium and phosphorus content of the sludge was monitored during 2020. The aluminium
content in the sludge was mostly below 0.5% of TSS but increased to a maximum of 1.7% of TSS in week 48
after some weeks with slightly higher dosage of aluminium chloride (Figure 22). The iron content of the
sludge, which varied between 2% and 7% of TSS, did not clearly correlate to iron dosage. The reason for this is
most likely because ferrous creates an iron hydroxide buffer in the sludge (which has a sludge age of around
20 days) that delays its effect since phosphate can bind to the iron hydroxide at a later time point. Al3+ and Fe3+
rapidly forms stable molecules with phosphate (AlPO4, FePO4). This difference complicates the evaluation.
The phosphorus content in the sludge was even throughout the year, around 3-4% of TSS, which is similar as
previous years.

48
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Metals and phosphorus in sludge

8 1600

7 1400

6 1200

5 1000
% of TSS

g Me/d
4 800

3 600

2 400

1 200

0 0
w1
w3
w5
w7
w9
w11
w13
w15
w17
w19
w21
w23
w25
w27
w29
w31
w33
w35
w37
w39
w41
w43
w45
w47
w49
w51
w53
Al3+ to BR6 Fe2+ to pre-aeration P in sludge (% of TSS)

Fe in sludge (% of TSS) Al in sludge (% of TSS)

Figure 22. Iron, aluminium and phosphorous in sludge (weekly data). Note that the iron and aluminium
are presented as a sum of added metals (g Me/d). The highest dose was in week 48 when 1334 g Me/d was
added. Out of this, the majority was iron (1039 g Fe/d) and the remainder was aluminium (295 g Al/d).

Enhanced biological phosphorus removal (EBPR)

The treatment process in the future Henriksdal WWTP, and the MBR-pilot, was designed for chemical
phosphorus removal and therefore, an anaerobic zone required for enhanced biological phosphorus removal
(EBPR) was not included in the design. Still, operational data from the pilot regarding Fe-dosing and
phosphorus removal, in combination with the phosphate release during acid cleaning of the membranes,
indicated EBPR-activity within the biological process. This was also confirmed by monthly phosphate release
tests (as described in Tykesson & la Cour Jansen, 2005) initiated in 2018. Further, the data on the molar ratio of
iron to removed phosphorus from 2020, presented in Table 17, confirms that EBPR occurred in the pilot.

Results from the P-release tests from 2018-2020 are shown in Figure 23.

49
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

P-release rate
12 30

10 25

8 20
g P/kg VSS,h

°C
6 15

4 10

2 5

0 0

Temp bio MBR-pilot Henriksdal WWTP High Low

Figure 23. EBPR activity in the MBR-pilot and Henriksdal WWTP (CAS). Dotted lines show limits for
EBPR-activity (Janssen et al., 2002), below the lower line = poor EBPR, above the higher line = high EBPR.
Temp bio is the temperature in the treatment process.

During 2020, the EBPR activity was higher than previous years, with values on or above 7 g P/kg VSS, h from
January to August. The P- release rates reached top values above 10 g P/kg VSS, h, which is almost in line with
the well-established Swedish EBPR-plant Öresundsverket, 12,5 g P/kg VSS,h (Tykesson et al., 2005), although
still significantly lower than Duvbacken WWTP, 20 g P/kg VSS,h (Salmonsson et al., 2017). In winter, the
phosphate release rate tends to decrease to almost as low levels as in the reference sludge from Henriksdal
WWTP (CAS with no EBPR).

According to design guidelines for EBPR (Metcalf & Eddy, 2014) the following is required to achieve EBPR:

(1) a strictly anaerobic zone with access to VFA (>8 g/g P) and 0.5-1.0 h
hydraulic retention time followed by an aerated zone,
(2) a COD:P ratio in the inflow of >60:1,
(3) low load of TP from digested sludge reject water.

The MBR-pilot does not contain an anaerobic zone in the water line. However, the DO and nitrate levels are
often low with <0.2 mg O2/L and <3 mg NO3-N/L, respectively, in the anoxic pre- and post-denitrification
zones. These zones are both followed by aerated zones, but the hydraulic retention time in the anoxic zones is
shorter than 30 min. The COD:TP ratio in the inlet to the biological treatment (PTW) averaged at 81:1 in 2020,
which should be enough for EBPR. In addition, external carbon source is added to the post-denitrification
zone. The reject water from sludge dewatering contained 99 mg TP/L and 60 mg PO4-P/L (yearly average),
which is significantly higher than at Henriksdal WWTP (15 mg TP/L and 0.7 mg PO4-P/L), indicating P-release
in the digesters.

In Figure 24 a and b the influence of iron and aluminum dosage on the bio-P and P-removal is visualised.
Combinations of EBPR and chemical precipitation have been shown to work well at Swedish WWTPs (Jansen
et al., 2009). The phosphorus content in sludge was rather constant over time, between 3 and 4% of SS in WAS.
The iron content in sludge was low during 2020 (2-5% of SS) and the modest dosing of PAX due to EBPR
resulted in low levels of aluminum in the sludge (0-2% of SS). When EBPR activity was high, the dose of
coagulant decreased, since it is controlled by the PO4-analyser in the effluent. Therefore, a correlation between

50
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Fe/Al in sludge, coagulant dosing and EBPR activity could be seen. It is, however, hard to determine the exact
reason for the drop in EBPR activity from September to December, due to the combined effect of various
parameters.

A Fe and Al dose vs TP removal

30 3.0

25 2.5

20 2.0

[mole/mole], [mg/L]
[g Me/m3], [g P/h]

15 1.5

10 1.0

5 0.5

0 0.0

Fe dose (g Fe/m3) Al dose [g Al/m3] TP removed (g P/h)

Me:TP (mole/mole) TP effluet (mg/l)

B Fe and TP in activated sludge

14
12
[% of SS], [g P/kg VSS, h]

10
8
6
4
2
0

Fe in WAS (% of SS) TP in WAS (% of SS)

P-release rate (g P/kg VSS,h) Al in WAS (% of SS)

Figure 24. A) Fe- and Al-dose in activated sludge process and TP removed in the process. Me = Metal. B) Fe,
Al and TP in waste activated sludge (WAS) together with P-release rate.

During spring 2020, two MSc thesis projects performed in cooperation with KTH, by Ross Roberts and
Benjamin Fridh, studied the EBPR process in depth (Roberts, 2020; Fridh, 2020). It was concluded, based on
phosphate profiling and Neisser staining (polyphosphate granules in poly-P accumulating organisms are
stained blue and visualization in microscope according to method described in Fridh (2020), that the P-release
mainly took place in the pre-denitrification zone (Figure 25 and Figure 26). The Neisser staining also indicated
some release in the post-denitrification zone while this could not be seen in the PO4-profilings. In a previous
PO4-profiling performed on one single occasion 7/6 2018, the largest P-release was seen in the post-
denitrification zone (Andersson et al., 2020). The external carbon source used at the time was methanol and
the nitrate concentration in and out of the post-denitrification zone was low at the time of sampling, 2.8 mg/L

51
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

and 0.8 mg/L respectively. Corresponding concentrations during the sampling period 2020 were around 4
mg/L and 2 mg/L while the nitrate concentration in the pre-denitrification zone was around 0.5 mg/L.
Although the studies by Roberts and Fridh both pointed to the pre-denitrification zone for P-release, it is
possible that both the pre- and post-denitrification zones act as the “anaerobic” zone in the EBPR-process
depending on other factors such as nitrate concentration, available VFA, recirculation of oxygen from RAS-
deox and deox zone etc.

Figure 25. Phosphate profile in the water line. Y-axis show the mass flow of phosphate as grams PO4-P per
hour. Since the total flow is different in different zones the concentrations cannot be compared. From
Roberts 2020.

Figure 26. Left, a micrograph of sludge from ox zone where poly-P granules are stained with Neisser´s blue
and microorganisms are counterstained with Neutral red. Right, the number of poly-P clusters in different
zones. From Fridh (2020).

P-release tests in the presence of nitrate showed P-release rates in the same range as the reference without
nitrate when acetate was added in excess, dose of 400 mg/L acetate as soluble COD. However, if the addition
of acetate was limited (20-40 mg/L sCOD) in the presence of nitrate, no P-release was measured. Results are
shown in Table 18. The conclusion was that P-release in the MBR-pilot is achieved in the presence of nitrate

52
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

and that sufficiently high VFA-concentrations is provided by the incoming wastewater or by hydrolysis in the
pre-denitrification tanks.

Table 18. P-release rates in presence of nitrate, with different acetate addition (Roberts 2020).
1st P-release
Carbon Dose Temp. Start sCOD Start VFA Start NO3-N VSSg Max PO4-P rate
source mg COD/L ºC pH mgCOD/L mg H3COO·/L mg/L /L mg/L mg P/g VSS·h
Acetate 400 19.9 7.8 459 326 26.8 5.9 56 8.0 (anox)
Acetate 400 20.0 7.4 450 324 <0.5 5.9 60 9.0
Acetate 400 20.2 7.9 384 333 39.0 6.2 54 7.8 (anox)
Acetate 400 20.3 7.3 380 332 0.7 6.2 60 8.7
Acetate 20 20.4 7.1 84 <50 29.4 5.8 0.9 0.1(anox)
Acetate 20 20.3 7.4 84 <50 3.9 5.8 7.6 0.1
Acetate 40 20.3 7.3 96 <50 27.3 5.8 3.2 0.3 (anox)
- 0 20.4 7.2 86 <50 4.1 5.7 7.8 0.1

The effect of using different external carbon sources as substrate for the P-release test and as a carbon source
for the post-denitrification in the MBR-pilot was also studied. Acetate gave the highest P-release rates with an
average of 8.6 g P/kg VSS, h. A VFA-slurry, produced by fermentation of primary sludge and food waste,
followed at 6.3 g P/kg VSS,h (Owusu-Agyeman et al. 2020). Brenntaplus, a mixture of proteins, sugars and
alcohols, gave a P-release rate of 5.0 g P/kg VSS, h. Ethanol, methanol and glycerol resulted in P-release rates
insignificantly higher than the negative reference (no carbon addition), 0.1-1.3 g P/kg VSS, h compared to 0.1 g
P/kg VSS, h. The external carbon source added to the post-denitrification zone was changed from glycerol to
ethanol for a few months to see if it affected the EBPR. However, no effect could be detected in the pilot line or
when performing P-release tests in the lab. To improve the EBPR-activity in the line, a VFA-based carbon
source could be used instead of glycerol or methanol, either in the pre- or post-denitrification zone.

In addition, the influence of temperature on the P-release rates was investigated in the range of 15-25°C with
the conclusion that the initial P-release rate was higher at higher temperatures whereas the highest overall P-
release was achieved at lower temperatures.

During spring 2020 samples were also collected regularly for DNA-analysis of the PAO community. DNA-
extraction and PCR were carried out but because of the COVID-19 situation, the samples could not be
sequenced during this study. This is planned in the future.

Details about the two studies and additional results are presented in the MSc-thesis reports (Roberts 2020;
Fridh 2020).

6.4 BOD reduction

Analysis on BOD7 from daily composite samples have, since the start-up of the MBR pilot in 2013, shown
values of <2 mg O2/L, except for one sample where the analysed concentration was 3 mg O2/L. Since the
expected effluent requirement of BOD7 in year 2040 is 6 mg O2/L as an annual average, there is no reason to
assume that the effluent requirement will not be met. Analysis of BOD was not carried out in 2020 and no
specific measures have been taken to achieve a higher BOD reduction.

53
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

6.5 Membrane performance

Trial J F M A M J J A S O N D
Minimising membrane scouring air use
Oxalic acid and citric acid comparison
Minimising membrane cleaning chemical use
Recovery cleaning
Offgas measurement
Extended membrane operational cycle

During 2020 the membranes were operated in extended cycles with 15 minutes of permeation, followed by 1
minute of relaxation (standard is 10 + 1 minute). Both feed (pumping from BR6) and aeration was on during
the normal operation cycle. In order to manage the varying flowrate with only two membrane tanks, the
pumping of permeate was proportional to the feed, which in turn was proportional to the level in BR6. This
control strategy means that the membrane flux will be controlled (but varying with inflow), and the
transmembrane pressure (TMP) will be allowed to vary.

As it is inefficient to operate membranes at too low fluxes, the membrane tank longest in operation went into
standby mode at low influent flowrates. In standby mode the membranes were aerated intermittently 5
minutes every half hour.

During 2020 the inflow to the pilot was higher than average and permeate recycling, which normally is used
as the membranes area is slightly too large, was therefore not applied.

Table 18 list the main events and experiments regarding the membrane operation during 2020.

Table 19. Main events and experiments regarding the membrane operation.
Event Time period (2020)
Recovery cleaning (RC) with sodium hypochlorite Week 11 and 50
Recovery cleaning (RC) with acids (oxalic and citric) Week 15 and 51
Maintenance cleaning (MC): Hypo replaced with water Week 13 to 23
No hypo MC Week 24 to 37
Hypo MC initiated based on fouling control algorithm Week 42 continuing
Operation with prolonged permeation cycle (15 min instead of 10 min, Throughout 2020
capacity increased by 3%)
Operation with forced Leap-Lo (low membrane aeration) Week 1 to 43

6.5.1 Permeability
Generally, a permeability above 200 L/(m2·h·bar) is considered as good according to the supplier. As can be
seen in Figure 27, the permeability was above 200 L/(m2·h·bar) throughout most of the year 2020 for both
membranes. The permeability was greatly affected by the different cleaning strategies tested this year, see
details in section 6.5.3 Membrane cleaning.

Recovery cleaning was carried out twice during 2020 with a clear increase in permeability after cleaning. Prior
to the first RCs, permeability was higher for MT1 (cleaned with oxalic acid) compared to MT2 (cleaned with
citric acid). After the first RCs, both membranes had similar permeability. From week 15 a decreasing trend
can be seen for both membranes and the lowest permeability was obtained in week 37 after 13 weeks of
54
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

operation without use of sodium hypochlorite for maintenance cleaning. To restore permeability, the hypo
cleanings were started again in week 37. The maintenance cleanings using acid did have a small positive effect
on permeability, which can be seen as a biweekly increase in permeability (Figure 27).

Permeability RC acids
RC hypo
MT 1 & 2 RC acids RC hypo MT 1 & 2
500
MT 1 & 2 MT 1 & 2
450
400
350
L/(m2·h·bar)

300
250
200
150
100
50
0
w1
w3
w5
w7
w9
w11
w13
w15
w17
w19
w21
w23
w25
w27
w29
w31
w33
w35
w37
w39
w41
w43
w45
w47
w49
w51
w53
MT1 MT2

Figure 27. Permeability (temperature compensated) for membrane 1 (MT1) and 2 (MT2) during project year
7 (2020). Recovery cleanings (RCs) were carried out with hypo in week 11 and week 50 and with acids in
week 15 and 51 for both membranes. MT1 was cleaned with oxalic acid, MT2 was cleaned with citric acid.

During 2020, the membrane aeration was forced to the lower air flow rate (see section 6.5.4 Membrane
aeration). No negative effects on membrane permeability were observed. The decrease in permeability could
be related to aeration, but the reduced cleanings are considered the major factor affecting membrane
permeability.

There was a rapid decrease in permeability for MT1 from week 39 to week 43, but no clear explanation to this
has been found. Previously, the general trend has been that MT1 has performed better than MT2 although it
has been cleaned using less chemicals. In week 43 the oxalic acid settings were changed to increase the
amount of acid used during each MC (from 1040 mg/L to 1300 mg/L which is the design concentration).

During 2020, aluminium chloride was dosed in BR6. The highest dosage was from week 43 to 48 when 2-4 mg
Al/L inflow was added. No negative effect on the permeability was observed when adding aluminium
chloride.

6.5.2 Flux and TMP

Fluxes for the two membranes are presented in Figure 28. Normally the membranes were operated with net
flux around 21 to 25 L/(m2·h). Design net flux for future Henriksdal is 20.9 L/(m2·h), and design max net flux is
30 L/(m2·h). If the flux was lower than design, one of the membranes would enter standby mode. During 2020
the flux was controlled proportional to the inflow. This year the inflow to the pilot was kept constant at 4.54
m3/h and was increased to 5.5 m3/h (design max flow) during rain events. The even flux with shorter peaks
reflects the inflow flowrates. As the inflow was constantly higher than design, membranes rarely entered
standby mode (with exception for downtime).
55
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Flux
30
25
20
L/(m2·h)

15
10
5
0
w1
w3
w5
w7
w9
w11
w13
w15
w17
w19
w21
w23
w25
w27
w29
w31
w33
w35
w37
w39
w41
w43
w45
w47
w49
w51
w53
MT1 MT2

Figure 28. Net flux during 2020.

The transmembrane pressure (TMP) is presented in Figure 29. Based on daily average data, net TMP varied
between 51 and 177 mbar for MT1 and between 56 and 128 mbar for MT2 during 2020. TMP is reduced after
recovery cleaning (RC) with hypo (w11 and w50). A smaller decrease in TMP can be observed for MT2 in
week 15 when RC with citric acid was carried out. The peaks in TMP around week 37 is related to the trial
with only acid MCs which was ended in week 37 and TMP decreased after MC with hypo. MT1 show higher
TMP than MT2 from week 39 with a peak week 43. During this period the membranes were operated with
different amount of acids for MCs and membrane aeration was forced to the lower air flow (Leap-Lo) until
week 43. The higher TMP for MT1 resulted in lower permeability and extra MCs were manually initiated in
week 43. Other abrupt changes in TMP are often related to the normal maintenance cleaning events.

RC hypo TMP RC acids

MT 1 & 2 MT 1 & 2
200
RC acids RC hypo
175
MT1 & 2 MT 1 & 2
150
125
mbar

100
75
50
25
0
w1
w3
w5
w7
w9
w11
w13
w15
w17
w19
w21
w23
w25
w27
w29
w31
w33
w35
w37
w39
w41
w43
w45
w47
w49
w51
w53

MT1 MT2

Figure 29. Net TMP during 2020. Recovery cleanings (RCs) were carried out with hypo in week 11 and
week 50 and with acids in week 15 and 51 for both membranes. MT1 was cleaned with oxalic acid, MT2
was cleaned with citric acid.

56
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

6.5.3 Membrane cleaning

The membranes were cleaned with sodium hypochlorite and citric or oxalic acid. MT1 was cleaned with
sodium hypochlorite and oxalic acid and MT2 was cleaned with sodium hypochlorite and citric acid. Two
types of cleaning procedures were carried out; maintenance cleaning (MC) and recovery cleaning (RC).

Maintenance cleaning
The maintenance cleanings (MCs) were automatically carried out on a weekly basis. In order to keep the
treatment line in operation, each membrane was cleaned separately, and the cleanings were scheduled at
night when the influent flow rate was low. In order to assure that the influent flow rate was not too high for
the membrane tank still in operation, the influent flow set-point was set to half of the current value, although
never lower than 1.8 m3/h, during MC.

The MC takes about one hour and according to the cleaning schedule provided by the supplier these cleanings
should be carried out with acid about once per week (after 345 m3 of permeate were produced by that
membrane) and with sodium hypochlorite about twice per week (after 173 m3 of permeate was produced).
The cleaning chemical was mixed with permeate and back pumped in pulses through the membranes.
Standard cleaning procedure included nine back pulses, the first one a bit longer (2-5 minutes) followed by
eight shorter with relaxation in-between (30 seconds followed by 4.5 minutes of relaxation). The chemical
solution was pumped with a back flux of 20 L/(m2·h) and the target concentrations of the solution entering the
membranes (after dilution with permeate) were 200 mg Cl2/L for sodium hypochlorite, 2000 mg/L for citric
acid and 1300 mg/L for oxalic acid.

In 2017 attempts of reducing the chemicals used for maintenance cleaning started. The time of the initial
backpulse was reduced from 5 minutes to 2 minutes and later the number of backpulses were reduced from 9
(incl. the first longer one) to 7 in total. In 2018, further reduction of oxalic acid usage was done by increasing
the interval in-between cleaning events. In 2019 both oxalic and citric acid usage was optimised and a
synchronization with the hypo MC was tested. During 2020 main focus has been on reducing the amount of
hypo used for MC.

The operational settings have been divided into separate trial periods. An overview of the trials regarding the
MCs are presented in Table 20.

Table 20. Overview of trials regarding the maintenance cleaning (MC) of the membranes.
Trial Start Description
T1 Sept 2017 Citric vs Oxalic - 7 BP (both MT)
T2 June 2018 Recovery Period (short switch between chemicals)
T3 July 2018 Trial reduced nr of BP oxalic acid, standard citric acid
T4 Aug 2018 Trial reduced nr of BP and 20% longer time in-between oxalic acid cleanings
T5 Oct 2018 No oxalic acid cleanings
T5 Dec 2018 One oxalic acid cleaning
T5 Dec 2018 No oxalic acid cleanings
T6 Feb 2019 Trial reduced nr of BP and 100% longer time in-between oxalic acid cleanings
T7 May 2019 Acid MC is carried out same night as Hypo MC, every 4th hypo for MT1 and every 2nd
hypo for MT2
T8 Aug 2019 Oxalic acid pumping reduced to 80% chemical flow during pumping
T9 Oct 2019 Citric acid cleaning with reduced chemicals, oxalic acid as previous period
T10 Oct 2019 Oxalic acid 80% chemical flow, 7 BP, every 4th Hypo MC
Citric acid 100% chemical flow, 7 BP, every 4th Hypo MC

57
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

T11 Feb 2020 Oxalic acid 80% chemical flow, 7 BP, every 4th Hypo MC
Citric acid 100% chemical flow, 9 BP, every 4th Hypo MC
T12 Mar 2020 Hypo MC replaced with water MC
T13 June 2020 Hypo MC excluded until permeability is below a low limit of <150 L/(m2·h·bar).
Acid MCs carried out with half standard interval (once every other week)
T14 Sept 2020 Recovery period
T15 Oct 2020 Fouling-based trigger for hypo MC
Acid MCs carried out with half standard interval (once every other week)

The amount of chemicals used normalized to the initial settings (back pulse duration 2 minutes + 8 x 30
seconds carried out after 345 m3 permeate produced) are presented in Figure 30 together with the
permeability for the trials during 2020.

Trial 10 (T10) started in October 2019 where MC with acid was carried out after every 4th hypo MC for both
MTs. This trial continued until February 2020. As citric acid consumption was reduced compared to the
design, the permeability for MT2 decreased compared to MT1. For T10 the average permeability for MT1 was
345 L/(m2·h·bar) and 277 L/(m2·h·bar) for MT2. Based on these results, it was decided to return to design
settings for citric acid MC (but still performed at half design interval) from February 2020 (T11). During T11,
recovery cleaning (RC) of the membranes were carried out in preparation for the next trial.

In March 2020, the sodium hypochlorite was replaced with water, trial 12 (T12) to see the effect of the
backpulse alone without addition of any chemicals. As permeability remained stable for more than two
months with this approach it was tested to completely exclude the hypo MCs (T13) and monitor for how long
the membranes could operate with only acid MCs before the permeability decreased below 150 L/(m2·h·bar).
This trial started on June 9th (w24) and continued for three months (until September 9th, w37). The membranes
were operated without hypo MC for 92 days and permeability decreased from maximum 394 L/(m2·h·bar) to
minimum 121 L/(m2·h·bar) for MT and from 383 to 175 L/(m2·h·bar) for MT2. After a short recovery period
with standard cleaning schedule (T14) a new trial started in October 2020 (T15) where the hypo MCs were
initiated based in indication of membrane fouling. The same algorithm that controls the membrane aeration
(called fouling control, see section 6.5.4 Membrane aeration) was used.

58
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Permeability

T10 T11 T12 T13 T14 T15

700 100

600
80
500
L/(m2·h·bar)

400 60

%
300 40
200
20
100

0 0
w1
w3
w5
w7
w9
w11
w13
w15
w17
w19
w21
w23
w25
w27
w29
w31
w33
w35
w37
w39
w41
w43
w45
w47
w49
w51
w53
M1 MC Acid M2 MC Acid M1 Permeability

Average M1 M2 Permeability Average M2

Figure 30. Membrane permeability and amount of acid used for MC, normalized to back pulse duration of
2 minutes + 8 x 30 seconds carried out with interval of 345 m3 of permeate produced. M1 was cleaned with
oxalic acid, M2 was cleaned with citric acid. T10 – T15 are trial periods. T10 started in October 2019.

Maintenance cleaning with sodium hypochlorite has been carried out with interval according to supplier, but
with reduced backpulses; 2 min initial pumping followed by 6 x 30 seconds until week 13. From week 13 to
week 37 no hypo was used. In the five week long recovery period (T14), four MCs with hypo + acid were
carried out for each MT. During T15 (week 42 to 52), when hypo was initiated by the fouling control
algorithm, 4 hypo MCs were carried out for MT1 and 1 hypo MC for MT2 (all carried out in October). During
November and December no hypo MCs were needed, but recovery cleaning was carried out by the end of the
year increasing the permeability for both membranes. The trial with fouling-based control of hypo MCs will
continue during 2021. The results so far indicate that hypo MC is required to maintain good permeability, but
the intervals can be considerable reduced from standard interval of about 2 MCs with hypo per MT per week.
Trial 13 (88 days), with no hypo MCs and half interval of acid MCs showed a slow reduction of permeability
followed by more sudden drops in permeability (especially for MT1) but permeability was only below 200
L/(m2·h·bar) for 9 days (MT1) and 1 day (MT2) and easily recovered by initiating standard cleaning.

Recovery cleaning
During recovery cleaning (RC) the membrane tank was emptied, then filled with chemical solution and the
membranes were left to soak overnight.

According to the membrane supplier, RC should be carried out twice every year with both sodium
hypochlorite and acid. However, as the permeability has been good, RC has normally only been performed
once per year. Last year (2019), RCs were carried out in March and focus was on measuring the chlorine gas
and chloramines emitted to surrounding air when the membranes were soaked in sodium hypochlorite. This
year RC was performed in March/April and in December. RC with hypochlorite was done 1-4 weeks before
RC with acid on both occasions.

When comparing the effect of the recovery cleanings on permeability, the RCs resulted in similar final
permeability in both MTs both times (Figure 31). After the acid RCs the permeability was around 400-450

59
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

L/(m2·h·bar). Generally, oxalic acid RCs resulted in less improvement of permeability compared to the citric
acid RCs. This can be explained by a higher permeability (and thus cleaner membranes) before the acid RC for
MT1 compared to MT2.

For MT1, RC with hypo in March followed by MC with oxalic acid (carried out 15th of March), led to a total
increase in permeability from 315 to 453 L/(m2·h·bar). The following RC with oxalic acid did not improve
permeability more than an MC with oxalic acid (compare increase after oxalic MC on the 31st of March to
oxalic RC on 7th of April). For MT2 the hypo RC in March plus a citric acid MC (carried out on the 24th of
March) resulted in a total increase from 235 to 373 L/(m2·h·bar), however the citric acid MC contribution was
small.

The decrease in permeability between the hypo RCs and the acid RCs in March/April indicate that a higher
final permeability after acid RCs could be achieved if hypo and acid RCs were carried out closer in time. This
was observed after the RCs carried out in December when permeability after acid RC reached a maximum of
around 450 L/(m2·h·bar) during the week after acid RCs. However, this was not better compared to the result
after hypo RC followed by acid MCs in March.

60
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Permeability before and after RCs March/April 2020

500

400
L/(m2·h·bar)

300

200 RC hypo
RC Citric RC Oxalic
RC hypo MT2
MT2 MT1
100
MT1

0
2020-03-02 2020-03-09 2020-03-16 2020-03-23 2020-03-30 2020-04-06 2020-04-13
Permeability MT1 Permeability MT2

Permeability before and after RCs December 2020

500

400
L/(m2·h·bar)

300
RC Oxalic RC Citric
MT1 MT2
200
RC hypo
100 MT1
RC hypo
MT2
0
2020-11-30 2020-12-07 2020-12-14 2020-12-21
Permeability MT1 Permeability MT2

Figure 31. Permeability before and after recovery cleaning (RC) with sodium hypochlorite (hypo) and acids
(oxalic acid and citric acid for MT1 and MT2, respectively) in March/April 2020 (upper) and December 2020
(lower).

6.5.4 Membrane aeration

The membranes are aerated by a coarse bubble aeration system mounted at the bottom of the membrane
tanks. The membrane aeration flow is designed to be operated at one out of two levels, Leap-Hi (26 m3/h) or
Leap-Lo (14 m3/h). An algorithm, provided by the membrane supplier, is used to select which of them to be
used. The control strategy is called fouling control. According to the membrane supplier, the aeration system
operates very close to the minimum air flow rate at Leap-Lo. If the air flow rate would be reduced below the
Leap-Lo level, the mechanical aeration equipment would not function as intended.

By the end of 2019 and throughout most of 2020 the membrane aeration was forced to the Leap-Lo aeration
flowrate (Figure 32). This was tested to evaluate the effects of low aeration on membrane permeability.
Instead, the fouling control algorithm was used to trigger MC with hypo (see section 6.5.3 Membrane
cleaning). Leap-Hi aeration was manually selected for MT1 during 6 days in October (w43) to improve

61
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

permeability. After this period, the fouling control algorithm was again activated to control membrane
aeration. No automatic usage of Leap-Hi occurred during the remaining part of 2020.

Air flow rate

30
Leap-Hi
25

20
Leap-Lo
m3/h

15

10

0
w1
w3
w5
w7
w9
w11
w13
w15
w17
w19
w21
w23
w25
w27
w29
w31
w33
w35
w37
w39
w41
w43
w45
w47
w49
w51
w53
MT1 MT2

Figure 32. Membrane aeration as daily average during 2020. Leap-Hi corresponds to 26 m3/h and Leap-Lo
corresponds to 14 m3/h.

6.5.5 cTOC and membrane performance

Two separate experiments focusing on factors effecting the collodial Total Organic Carbon (cTOC) were
performed in 2020. One experiment investigated how carbon sources and reject water effect the cTOC. The
second experiment, the effect of cTOC on the membrane performance in terms of transmembrane pressure
(TMP). According to the membrane supplier SUEZ, the cTOC concentration should be kept <10 mg/L to not
affect the membrane performance. However, it is not exactly known how much the variation in cTOC affects
the membranes in terms of transmembrane pressure (TMP). To gain good knowledge of the cTOC and thus
potential adverse effects on the membrane performance, the effect of additions that presumably affect the
cTOC, such as external carbon source and reject water, needs to be understood.

Effects of carbon sources and reject water on cTOC

During 2019 and 2020, the cTOC concentration in the pilot increased up to 25 mg/L (Figure 33). It was
observed that the cTOC concentration was higher from September 2019 until April 2020. During this time, two
crucial changes in the operational parameters were made. First, the carbon source was changed from
methanol to glycerol, and second, reject water from the sludge dewatering was added to the pre-aeration tank
(Figure 33). When both addition of reject water was stopped and the carbon source was changed to ethanol
(March 2020), there was a clear decrease in the cTOC concentration.

62
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

cTOC and carbon sources

60

50

40
mg/L; L/h

30

20

10

0
2017-01-03
2017-02-22
2017-04-12
2017-05-31
2017-07-19
2017-09-06
2017-10-25
2017-12-06
2018-01-24
2018-03-14
2018-05-02
2018-06-20
2018-08-08
2018-09-26
2018-11-14
2019-01-02
2019-02-20
2019-04-10
2019-05-29
2019-07-17
2019-09-04
2019-10-23
2019-12-11
2020-01-29
2020-03-18
2020-05-06
2020-06-21
2020-08-12
2020-09-30
2020-11-18
Methanol (mg/L) Glycerol (mg/L) Ethanol (mg/L)

RW (L/h) TOC out (mg/L) cTOC (mg/l)

Figure 33. Summary of effluent TOC, dosing of external carbon sources and recycling of reject water from
digested sludge dewatering (RW) during 2020.

In order to investigate if the increase in cTOC was related to the addition of glycerol or reject water, a first trial
T1 without the addition of reject water (w24 to 41) and a second trial T2 without addition of external carbon
source (w41 to 48) were carried out. After these trials (week 48 onwards), the process was operated with both,
reject water and carbon source, with only two interruptions due to operational problems in the dewatering
unit (trial 3, T3). A limitation is that it is unknown if the cTOC concentrations might increase by accumulation
during longer periods of time, which may affect the outcomes of the trials. Furthermore, the values shown are
weekly averages, which could minimize the impact of high and low doses on the real cTOC values.

Following the historical data and the results from the tests performed (Figure 34), no clear correlation between
the addition of glycerol or reject water and the increase in the cTOC concentration could be found. During the
first trial, where glycerol was the only carbon source and the reject water recirculation was off, the cTOC
concentration exceeded 11 mg/L just one time, on the 9th of September. This, even at high glycerol addition (up
to 44 mg/L), which demonstrates that the addition of glycerol alone, does not contribute to the cTOC increase.

During the second trial period, the cTOC concentration decreased from an average of 8.7 mg/L, in T1, to an
average of 5.3 mg/L in trial 2 (Figure 34).

63
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

cTOC and carbon sources

50
T1 T2 T3 T3
45
40
35
30
25
20
cTOC limit
15
10
5
0

Glycerol (mg/L) RW (L/h) cTOC (mg/l)

Figure 34. cTOC variations during the trial periods T1, T2, T3, and interruptions

In week 48, both addition of glycerol and reject water started. However, due to operational problems, the
reject water could be recirculated only for 3 weeks. An increase in cTOC during this period can be observed in
Figure 34; values increased from 5.5 mg/L to 14.1 mg/L in three weeks. The following four weeks had a stop in
the reject water and a low but increasing glycerol dose with oscillating cTOC values. Finally, during the last
three weeks, with addition of glycerol and reject water, there was an increase followed by a decrease in cTOC.

There is no clear increase in the cTOC concentration when glycerol or reject water were added separately to
the process. However, it is suspected that the addition of both glycerol and reject water might affect the
concentration of cTOC. This needs to be investigated over a more extended period of time with the addition of
both and with shorter interruptions. Furthermore, if a new trial is performed, it is advised to take samples
more than once per week.

Effect of cTOC on membrane performance

Generally, the two membrane cassettes perform differently due to their operational parameters, including use
of cleaning chemicals and cleaning schedules, and therefore, the evaluation of the effect of cTOC on membrane
performance is not possible with the available data. In order to investigate if there is any correlation between
the membrane fouling (measured indirectly by permeability and TMP) in MT1 and MT2 respectively and cTOC,
the two membranes had to be studied separately. The membrane permeates from the two MTs was therefore
collected by two different samplers. By separating the permeate flows, the data collection allowed investigating
the difference in the effluent TOC concentration and cTOC from the two membranes (oxalic and citric acid) and
if the variations of cTOC affect the permeability. The colloidal particles measured by cTOC is thought to easily
get stuck in the membrane pores and cause fouling, but it is not known if different cleaning chemicals restore
this type of fouling to an equal extent. Since both MTs are fed with identical sludge with the same amount of
colloidal particles, a lower value of TOC in the effluent (and subsequently a lower cTOC) from one of them

64
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

mean that more cTOC is deposited on the membrane surface of that MT. Weekly sludge samples and permeate
samples were collected and analysed for TTF, TOC in, TOC out, and cTOC. The trial was performed fro 2.5
weeks between 2020-06-22 and 2020-07-10.

The results are presented in Table 21. Unfortunately, the collected data is insufficient to draw clear conclusions
and cTOC concentrations were too low during the trial period (<10 mg/L) to observed clear effects. More data
and perhaps a better analysis method is needed to evaluate the effect of cTOC on the performance of the
membranes. It is further suggested that future trials include a measuring campaign with daily sampling to
obtain results of TTF and cTOC. Furthermore, cTOC should be >10 mg/L during such a campaign if possible.

Table 21. Data from the trial where the effect of cTOC on membrane performance was investigated.
TTF-100 TOC TOC
SS TTF-100 Normalized Filtrate Permeate cTOC TMP Permeability
Date Membrane (mg/L) (s) (s) (mg/L) (mg/L) (mg/L) (mbar) (lmh)
2020-06-22 MT1 9855 312.5 317 15 8.1 6.9 66.1 347.8
2020-06-22 MT2 9855 321 326 15 7.7 7.3 66.6 349.8
2020-07-01 MT1 9459 318.5 337 14 6.2 7.8 79.0 293.9
2020-07-01 MT2 9459 326 345 14 7.4 6.6 78.2 298.0
2020-07-06 MT1 8615 395.5 459 12 6.6 5.4 77.4 305.1
2020-07-06 MT2 8615 418.5 486 14 6.8 7.2 72.0 331.9
2020-07-08 MT1 9164 344 375 14 5.3 8.7 83.8 299.8
2020-07-08 MT2 9164 363.5 397 13 5.6 7.4 76.7 331.0
2020-07-10 MT1 10186 355.5 349 16 7.4 8.6 77.6 299.8
2020-07-10 MT2 10186 367.5 361 14 7.6 6.4 69.6 334.3

6.6 Offgas measurements during NaOCl RC

It is known that sodium hypochlorite used for water treatment gives rise to the formation of chloramines,
which are emitted into the air (Thicket, 2002; Jacobs, 2007). Health effects ranges from increased problems for
people suffering from asthma, to irritated airways, to chronic problems. There is no occupational limit value
in Sweden, but the World Health Organisation (WHO) has suggested a reference value of 0.5 mg/m3 for
trichloramine (WHO, 2006). The Nordic Expert Group – a Nordic collaboration for production of criteria
documents on chemicals for occupational exposure limits – has suggested a threshold limit value (time-
weighted for 8-hour workday) of 0.1 mg/m3 for trichloramine (Wastensson, 2019).

In March 2020, a RC was performed (one year since last RC) although permeability was good (around 250-300
L/(m2·h·bar)) to get more results on the chlorine gas and chloramines emitted to air during RC with hypo.
From previous testing in March 2019, it had been determined that chlorine gas and chloramines were formed
in the pilot during RC with sodium hypochlorite. For the RC in March 2020, a test was set up for sampling in a
ventilation channel connected to the hood covering the membrane tank (Figure 35). In addition, the sampling
procedure was further developed based on last years’ results and supplemented with analysis of other volatile
chlorinated organic compounds. Sampling was started at the same time as addition of sodium hypochlorite to
the membrane tank was started. Filling of the tank took about 45 minutes and once filled the aeration was
turned on for 5 minutes to mix the solution.

The air flow in the exhaust was measured using a vane anemometer (Testo 417) and a funnel (Testovent 417),
before and after the sampling was done, and the flow was determined to be approximately 150 m3/h.

65
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Figure 35. Left: Gas-sampling hood and ventilation channel placed over a membrane tank. Right:
Ventilation channel and example of sampling setup for volatile chlorinated compounds.

Chloramine samples (samplers supplied by the Department of Public Health and Clinical Medicine, Umeå
University) and chlorine samples (SKC 225-9006 samplers) were collected in series during a 7-hour time
period after addition of hypochlorite was started (see Table 22). Sampling of volatile organic compounds
(VOC) using Tenax samplers was added to the testing, to screen for possible formation of chlorinated organic
compounds. The latter samples were only collected during the first 2 hours after addition of hypochlorite was
started (see Table 22). Sampleswere collected using SKC AirChek TOUCH air sampling pumps.

Preferably sampling would have been done for a longer period of time than 7 hours, but the sampling
technique had limitations due to the capacity of the samplers and consideration had to be taken to working
day limitations.

Table 22. Sampling setup of chlorinated compounds during RC with hypo in March 2020.
Analyte Sampling period Flow rate (l/min)
Chloramine 0–0.5 h 2
0.5–1 h 2
1–1.5 h 2
1.5–2 h 2
2–3 h 2
3–5 h 1
5–7 h 1
Chlorine gas 0–1 h 1
1–2 h 1
2–4 h 1
4–7 h 1
Chlorinated VOC 0–0.5 h 0.1
0.5–1 h 0.1
1–2 h 0.1

Chloramine samples were then analysed using ion chromatography, by Department of Public Health and
Clinical Medicine, Umeå University, chlorine samples were analysed using ion chromatography by the
Department of Occupational and Environmental Medicine, Örebro University Hospital and VOC samplers
were analysed by thermal desorption, gas chromatography and mass spectroscopy.

66
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

The main objective of the measurement campaign was to study if membrane cleaning with hypochlorite will
result in high levels of these gases emitted, which can be hazardous from a working environment perspective.
Since the ventilation from the membrane tanks in the full-scale plant will be connected to a technical tunnel
where different types of equipment will be installed, and since chlorine gas is very corrosive and can damage
this equipment, it is from this perspective as well, interesting to investigate emitted levels.

As the membranes were operated with low maintenance cleaning during 2020, which resulted in lower
permeability towards the end of the year, and to repeat the measurement campaign on gas emissions during
RC with hypo, it was decided to do an additional set of RCs in December 2020. Unfortunately, the gas
emission measurement campaign could not be conducted this time due to various practical reasons but is
planned to be repeated in 2021.

The RCs were carried out first with sodium hypochlorite and then with acids (oxalic acid for MT1 and citric
acid for MT2) some weeks later.

The schedule for cleanings can be seen in Table 23 together with the amount of chemicals used and conditions
at start and end of the soaking. Higher starting values during 2020-12-09 to 2020-12-10 may be due to pipes
being flushed from the previous RC and thus more recent sodium hypochlorite reaching the tank?

Table 23. Results from RC.

Measurements Measurements
Membrane in tank at the Soaking in tank at end
Date tank Chemical Amount start of soak time of soaking
2020-03-09 to MT1 Sodium 19.0 L pH 9.3 18.5 h pH 8.5
2020-03-10 hypochlorite Cl2 895 mg/L Cl2 490 mg/L
(66 g/L)
2020-03-11 to MT2 Sodium 16.8 L pH 9.2 21 h pH 8.0
2020-03-12 hypochlorite Cl2 730 mg/L Cl2 425 mg/L
(74.5 g/L)
2020-04-06 to MT2 Citric acid 4.3 L pH 2.8 20 h pH 2.9
2020-04-07 (51%)
2020-04-07 to MT1 Oxalic acid 18.9 L pH 2.3 19.5 h pH 2.3
2020-04-08 (8%)
2020-12-08 to MT1 Sodium 23.9 L pH 9.4 20.5 h pH 8.2
2020-12-09 hypochlorite Cl2 600 mg/L Cl2 430 mg/L
(53.5 g/L)
2020-12-09 to MT2 Sodium 22.9 L pH 9.4 20.5 h pH 8.0
2020-12-10 hypochlorite Cl2 755 mg/L Cl2 375 mg/L
(53.5 g/L)
2020-12-14 to MT1 Oxalic acid 20.0 L pH 2.2 21 h pH 2.1
2020-12-15 (8%)
2020-12-17 to MT2 Citric acid 4.5 L pH 2.8 19.5 h pH 2.8
2020-12-18 (51%)

Results from the measurement of volatile chlorinated compounds are presented in Table 24 and Figure 36 to
Figure 38. The VOC samples showed that the most abundant compound was by far chloroform. Other
identified compounds were 1,1-dichloroethen, methylene chloride, carbon tetrachloride,
dibromochloromethane and tetrachloroethene. The latter were detected only in trace amounts so only data for
chloroform is presented.

67
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Table 24. Total amounts and peak concentration of volatile chlorinated compounds emitted during
recovery cleaning.
Total Timespan Peak Peak Limit value
emission (h) concentration occurrence (mg/m3)
Analyte (mg) (mg/m3) (h)
Chloramine 6200 7 9.1 1-1.5 0.1(1)
Chlorine gas 900 7 1 1-4 1.5(2)
Chloroform 200 2 1 1-2 10(3)
(1) Suggested threshold limit value (time-weighted for 8-hour workday)
(2) Short-term limit value (15 minutes exposure)
(3) Threshold limit value (time-weighted for 8-hour workday)

Trichloramine
10
9
Trichloramine (mg/m3)

8
7
6
5
4
3
2
1
0
0-0.5 h 0.5-1 h 1-1.5 h 1.5-2 h 2-3 h 3-5 h 5-7 h
Time (h)

Figure 36. Emission of trichloramine during a 7-hour period, while cleaning wastewater treatment
membranes with sodium hypochlorite.

68
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Chlorine gas
1.2

1
Chlorine gas (mg/m3)

0.8

0.6

0.4

0.2

0
0-1 h 1-2 h 2-4 h 4-7 h
Time (h)

Figure 37. Emission of chlorine gas during a 7-hour period, while cleaning wastewater treatment
membranes with sodium hypochlorite.

Chloroform
1.2

1
Chloroform (mg/m3)

0.8

0.6

0.4

0.2

0
0-0.5 h 0.5-1 h 1-2 h
Time (h)

Figure 38. Emission of chloroform during a 2-hour period, while cleaning wastewater treatment
membranes with sodium hypochlorite.

As can be seen in Figure 36 to Figure 38, the emission process was slower than expected, and the total
amounts of compounds presented in Table 24 are an underestimation as the measuring period of 7 hours most
likely was too short. The data works well to give a rough estimate of the amounts emitted, and the results can
be used as a source for further calculations of occupational exposure levels that can occur, when scaling up to
a full-size wastewater treatment plant. Even though sampling methods do not follow standards for measuring
occupational exposure, the peak concentrations can be compared to the occupational limit values in Sweden
(Arbetsmiljöverket, 2018), or in the case of trichloramine, recommended exposure limit as a reference.
Chloramine peaked at 90 times the recommended limit, chlorine gas at 70% of the short-term exposure limit
(15 min exposure) and chloroform at 10% of the occupational exposure limit (8-hour workday average).

As the intended longer measurement campaign planned for the December RCs was cancelled, it is the
intention to do another measurement campaign of volatile chlorinated compounds during RC next year.
69
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

6.7 Sludge production and sludge properties

In Table 25, data from the pilot related to sludge production and sludge quality is compared to data from the
Henriksdal WWTP (annual average 2020) and design data for the future Henriksdal WWTP according to the
SFA project. The differences in sludge production between the pilot and Henriksdal WWTP are caused by the
configuration of the primary treatment, which was described in previous reports together with the effect on
the subsequent treatment steps (Andersson et al., 2019, 2020, 2021).

Table 25. Sludge data from the pilot year 2020 compared to data from the Henriksdal WWTP 2020 and
design data for the future Henriksdal WWTP.
Pilot Henriksdal WWTP Design future
Parameter data 2020 data 2020 Henriksdal
WAS production (kg SS/d) 15.9 22 100 59 000
Part of total sludge production (%) 42% 27% 34%
VSS in WAS (% of SS) 80 68 63
Fe in WAS (% of SS) 4.1% 11 -
Al in WAS (% of SS) 0.5% - -
PS-production (kg TS/d) 21.7 61 000 117 000
Part of total sludge production (%) 58% 73% 66%
VS in PS (% of TS) 88 79 80
Total sludge production (kg TS/d) 37.6 83 100 176 000
Total sludge age, SRTtot (d) 17.5 14.5 28
Aerated sludge age, SRTox (d)* 6.5** 5.7 7***
SVI jan-jun (mL/g) 330 141 -
SVI jul-dec (mL/g) 346 117 -
*yearly average, the aerated volume is adjusted based on water temperature using the flex-zones
**including membrane tanks, without membrane tanks SRTox = 4.8 d
***including membrane tanks, yearly average

A summary of sludge properties analysed in this project that might affect the membrane performance are
shown in Figure 39 and Figure 40 below, together with normalized permeability. The coagulant dose (Fe2+ and
Al3+) was rather constant over the year until week 43 when the dose went up significantly. Consequently, the
metal content in the sludge started to increase slowly. At the same time the permeability increases but that is
most likely due to membrane cleaning and not the metal content in the sludge. In theory, high values on
Time-to-filter, TTF (described in chapter 5.3.2), and colloidal TOC (cTOC) should correlate with a decrease in
the permeability (Gkotsis and Zouboulis, 2019). This could not be seen in the process data. The cTOC value
should be below 10 mg/L to ensure minimal fouling and values above 20 mg/l indicates poor sludge
filterability (Gkotsis and Zouboulis, 2019). In 2020, the cTOC was above 10 mg/L during 26 out of 51 weeks
and above 20 mg/L during 7 weeks, which is worse than previous years. TTF-100 should be lower than 100 s
for high filterability while values above 300 s indicates poor filterability (Gkotsis and Zouboulis, 2019). In 2020
the TTF-100, normalized or not, was never under 100 s. TTF-100 and normalized TTF was under 200 s on 1
and 2 timepoints respectively and between 200-300 s on 10 and 11 timepoints out of 51. The highest values
were 855 and 1015 s for TTF-100 and normalized TTF. In 2019 the values were below 100 s from week 1 to 32
when the TTF started to increase steadily. This years results indicate very poor filterability of the sludge. Still,
the permeability was good throughout the year regardless of high TTF and cTOC and the fact that the
membrane cleaning scheme was decreased (see chapter 6.5.3). In the beginning of the year, high values for
both TTF and cTOC were observed. From week 17, both TTF and cTOC dropped and from week 41 onwards
there was a drop in cTOC values but not in TTF. All in all, only a weak correlation between TTF and cTOC
was found (R2=0.38). Previous years a correlation between TTF-100 and Fe in sludge as well as Fe-dose could
70
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

be seen. This year no such correlation existed. This could be due to the high EBPR activity (chapter 6.3) and
subsequent low dose of Fe2+ and Al3+ in the pilot.

Weekly averages
500 25
450
400 20
350
300 15
250

°C; % of TSS; g/m3

lmh

200 10
150
100 5
50
0 0
w.1 w.5 w.9 w.13 w.17 w.21 w.25 w.29 w.33 w.37 w.41 w.45 w.49 w.53
Week no.
Temp (C) Permeability MT1 (lmh) Permeability MT2 (lmh)
Me in sludge (% of SS) Me-dose (g/m3)

Figure 39. Normalised permeability, temperature, Metal (Me)-dose and Metal (Me) content in sludge over
the year.

Weekly averages
900 30
800
25
700
600 20
500
15
°C; mg/L
lmh; s; ml/s

400
300 10
200
5
100
0 0
w.1 w.5 w.9 w.13 w.17 w.21 w.25 w.29 w.33 w.37 w.41 w.45 w.49 w.53
Week no.
Temp (grad.C) TTF normalised (s) SVI (ml/g)
Permeability MT1 (lmh) Permeability MT2 (lmh) cTOC (mg/l)

Figure 40. Normalized permeability, temperature, TTF-100 (normalized to a MLSS-concentration of 10 000

mg/L), SVI and cTOC over the year.

In Table 26 the TSS concentration in waste activated sludge (WAS) as well as the content of iron, aluminium,
phosphorus and VSS are listed as annular averages for the years the pilot plant has been in operation. A more

71
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

efficient precipitation strategy combined with the previously described enhanced biological phosphorus
removal, EBPR, caused the Me/P ratio to decrease over the first years. Stricter effluent goals for phosphorus in
the effluent then caused the ratio to increase 2019. During 2020 the EBPR was higher than previous years
which most likely is the reason for the low Me/P ratio in the sludge.

Table 26. WAS composition (annual average) in the pilot over 7 years of operation (Me = metal (Fe + Al), n
= number of samples)
TSS Fe in sludge Al in sludge P in sludge VSS Me/P in sludge
Year (mg/L) (% of TSS) (% of TSS) (% of TSS) (% of SS) (mole/mole)
2020 8 967 4.1 0.5 3.3 80 0.9
n 51 51 43 51 51 51
2019 9 932 7.6 - 3.5 75 1.5
n 50 50 - 50 50 50
2018 8 480 6.4 - 3.3 77 1.1
n 50 50 - 50 50 50
2017 9 632 10.3 - 3.0 71 1.9
n 50 47 - 47 47 47
2016 8126 8.3 - 3.4 74 1.3
n 31 31 31 31 31
2015 9910 10.1 3.3 71 1.7
n 44 44 42 44 42
2014 9263 11.9 3.1 69 2.3
n 38 38 27 38 27

Data on metals and some organic micro pollutants in dewatered digested sludge have been collected over the
years and was analysed and compared to data from Henriksdal WWTP in a separate report (Nähri et al.,
2021). A short summary of the results and conclusions is given below.

Contradictory to the expectations, the concentrations of both metals and organic micro pollutants were in
general lower in the sludge from the MBR-pilot than in sludge from Henriksdal WWTP. Since practically all
particulate matter is retained by the membranes and accumulated in the sludge, the total area of removed
particles is larger in an MBR and the removal of metals and organic materials that adsorb to those surfaces
should therefore be bigger in an MBR than in a conventional activated sludge process. On the other hand,
there is a dilution effect in the MBR-pilot sludge since the total amount of sludge produced per volume of
treated wastewater was larger and the organic degradation rate was lower than in the full-scale plant. In
Henriksdal an average of 0.16 kg TS/m3 incoming wastewater was produced while the value for the pilot was
0.25 kg TS/m3. This dilution effect seems to overshadow the effect of a larger particle surface. The fact that
Henriksdal WWTP receives wastewater from three different tunnels (uptake areas) and the pilot only receives
water from one of these can of curse also affect the results since the metal content in wastewater might differ
between the different tunnels.

Comparing the quality of the dewatered digested sludge from thermophilic and mesophilic digestion showed
that higher metal concentrations were found after thermophilic digestion but lower concentrations of organic
micro pollutants. This indicates a higher degradation of organic micro pollutants at the higher thermophilic
temperature.

Results from analysis of metals in dewatered digested sludge from the pilot and Henriksdal WWTP are found
in Figure 41. Figure 42 shows the metal to phosphorus ratios for some of the metals that are most important
for quality control before spreading sludge on farmland.

72
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

35

30

25
MBR HEN
mg ME/kg TS

20

15

10

0
Pb Co Cr Mo Ni Ag Sn W Bi

600 2.5
MBR HEN
MBR HEN
500
2.0

400
mg ME/kg TS
mg ME/kg TS

1.5
300
1.0
200

0.5
100

0 0.0
Cu Zn Sb Cd Hg
Figure 41. Average values of metal concentrations in dewatered digested sludge from Henriksdal WWTP
(HEN) and the MBR pilot. The unit is mg ME/kg TS. Error bars show the lowest and highest analysis
value.

73
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Metal/phosphorous ratio Metal/phosphorous ratio

25 500
MBR P-kvot HEN P-kvot
20 400

MBR P-kvot
15 300
HEN P-kvot

10 200

5 100

0 0
Cd Hg Pb Ag
Figure 42. Metal/phosphorus ratio (mg/kg P) for a few important metals. HEN = Henriksdal WWTP.

6.8 Sludge pilot

Trial J F M A M J J A S O N D
How low can we go - thermophilic
Emptying and inoculation of the reactor
How low can we go - mesophilic
Sludge dewatering in operation
Study on indicators of process failure

During 2020, the theme of the sludge pilot trials was “how low can we go”. Two interconnected trials were
performed, a thermophilic and a mesophilic crash test, where the goal was to see at how low HRT the
digestion process could be operated before it failed. The thermophilic trial started in 2019 and the mesophilic
trial was finalized in the beginning of 2021. Nevertheless, both trials are presented in this year’s report. Since
the operation of the thickener was very unreliable during previous years, it was decided to bypass the
thickener and feed mixed sludge directly into the pilot during these trials since it was crucial to control the
HRT. This resulted in an initially low organic loading rate (OLR) which increased as the HRT decreased.

6.8.1 Feed characteristics

As shown in Figure 6, the mixed sludge tank receives Primary Sludge (PS) and Waste Activated Sludge
(WAS) from the MBR-pilot line. The digester was continuously fed with mixed sludge (MS) from the MS tank
throughout the trials, bypassing the thickener. The proportion of PS to WAS, in terms of TS weight, was
adjusted to 60/40 by discharging a fraction of the WAS to be more similar to the mixed sludge at Henriksdal
WWTP. The total and volatile solids in the MS varied during the year which can be seen in Figure 43. Since
the sludge was not thickened, the TS-concentration in the feed was low, generally around 1.5-3.0%, and the
OLR was consequently also low, around 2 kg VS/m3,d; although increasing with reduced HTR. The OLR at
each retention time for the two trials can be seen in Figure 44. OLR was slightly lower during the mesophilic
trials due to lower TS in the primary sludge.

74
 Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Feed characteristics
8 13.0 100
9.5
7 90
80
6
70
[kg VS/m3,d], [%]

5 60

[% of TS]
4 50
3 40
30
2
20
1 10
0 0
2019-28
2019-31
2019-34
2019-37
2019-40
2019-43
2019-46
2019-49
2019-52
2020-03
2020-06
2020-09
2020-12
2020-15
2020-18
2020-21
2020-24
2020-27
2020-30
2020-33
2020-36
2020-39
2020-42
2020-45
2020-48
2020-51
2021-01
OLR [kg VS/m3,d] TS [%] VS [% of TS]

Figure 43. Feed characteristics in terms of TS and VS in mixed sludge and total organic loading rate (OLR)
for the trial periods w.28 2019 to w.1 2021.

The MS-pump takes sludge from the bottom of the MS-tank and on several occasions during the trials, sand
and grit from the primary clarifier accumulated in the bottom of the tank. Samples for TS and VS were
collected from the pipe after the MS-pump. This accumulation of inorganic material resulted in
unrepresentative values of TS and VS for some weeks. To compensate for this in Figure 44, a corrected value
for TS and VS was used for five of the 73 weekly averages shown. The corrected value was obtained by
multiplying the measured TS and VS and then dividing by a fictive VS-value of 75%, which is the average
value for the period (example 15/8 2019 w. 33; measured TS=8.8%, VS=19.5%, corrected VS=75%, corrected
TS=8.8x19.5/75=2.10%). For calculation of other key values reflecting digestion performance, such as VS%
degradation and OLR, the original data was used since the amount of VS [kg/d] is not affected by the
correction. In order to mitigate the problem of sand and grit, the mixed sludge tank was partially emptied
from the bottom regularly during the second half of the year.

OLR [kg VS/m3,d]

10
mesophilic trial thermophilic trial
8
kg VS/m3,d

0
10 9 8 7 6 5 4 4+EOM
HRT
Figure 44. Organic loading rate (OLR) at each retention time (HRT) during the mesophilic and
thermophilic trial.

75
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

From the end of September 2020, mixed sludge (MS) was analyzed for the content of carbohydrates, proteins
fats and COD. The result can be seen in Table 27. The average COD-value of MS was 20 827 mg/L (±15%)
which corresponded to 0.9 g COD/g TS or 1.2 g COD/g VS in the sludge.

Table 27. The average content of carbohydrates, proteins and fats in MS.
Component % of VS Standard deviation
Carbohydrates 45 ±11
Proteins 37 ±9
Fats 16 ±6

Due to limitations in the digester configuration and pump capacity, the HRT could not be decreased below 4
d. After operation at 4 d HRT for at least 3 HRTs the OLR was increased by addition of external organic
material (EOM). At Henriksdal WWTP fat separated from restaurant kitchen drains is added directly to the
digesters together with glycerol (a byproduct from biodiesel production) which is added according to biogas
demand and digester capacity by Scandinavian biogas Fuels. Therefore, kitchen oil and glycerol were added
to the pilot digester week 15-17 2020 (Table 28) and glycerol only week 2 2021 (Table 29). The COD-content
was different in the two different glycerol batches used, resulting in different organic loading at the same
pump flow. Since it was tricky to pump oil during the end of the thermophilic trial, it was decided to only use
glycerol at the end of the mesophilic trial.

Table 28. Addition of EOM during the thermophilic crash test in 2020.
Flow Glycerol Org. Loading Glycerol Flow Oil Org. Loading Oil [kg
Date [L/d] [kg VS/d] [L/d] VS/d]
06-apr 0 0 2.5 2.3
07-apr 2.52 2.70 2.5 2.3
08-apr 5.07 5.43 5 4.6
09-apr 7.61 8.15 5 4.6
10-apr 7.61 8.15 0 0
11-apr 7.61 8.15 0 0
12-apr 7.61 8.15 0 0
13-apr 3.80 4.07 0 0
14-apr 8.54 9.15 5 4.6
15-apr 9.98 10.69 0 0
16-apr 15.00 16.07 5 4.6
17-apr 19.92 21.33 5 4.6
18-apr 19.92 21.33 5 4.6
19-apr 19.92 21.33 0 0
20-apr 24.96 26.73 0 0
21-apr 30.00 32.13 0 0
22-apr 31.68 33.93 0 0
23-apr 31.68 33.93 0 0
24-apr 36.68 39.28 0 0

76
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Table 29. Addition of EOM during the mesophilic crash test in 2021.
Flow Glycerol Org. Loading Glycerol Flow Oil Org. Loading Oil [kg
Date [L/d] [kg VS/d] [L/d] VS/d]
11-jan 2.5 1.61 0 0
12-jan 5 3.21 0 0
13-jan 7.5 4.82 0 0
14-jan 10 6.43 0 0
15-jan 12.5 8.03 0 0

6.8.2 How low can we go – background and set-up

During the reconstruction of the digesters at Henriksdal WWTP, the operating volume will be reduced for a
few years, which will cause a reduction in the hydraulic retention time (HRT). In addition, the retention time
year 2040 is expected to be around 12 days at normal operation which includes thickening to >6% TS. If
thickening fail or some of the digesters need to be taken out of operation for longer periods, the HRT might be
reduced to <8 days. Low retention times increase the risk of washing out methanogens because of their long
generation time (slow growth), and the lack of time for hydrolysis of the substrates. This trial aimed to
decrease the retention time until a crash in the pilot’s digester performance occurred. Either by inhibiting the
methanogens activity leading to accumulation of VFA or by the inability of bacteria to perform hydrolysis.
The results would add information to the decision-making during the reconstruction of the sludge line at
Henriksdal WWTP where a stable operation, good quality gas production and good properties of the digested
sludge must be ensured.

In these trials, it was important to have a controlled and steady HRT. Because of the major problems related to
the previous operation of the thickener it was decided to bypass the unit during these trials. Thus, the digester
was operated with a low organic loading rate (OLR). Although this was not the optimal mode of operation it
was preferable to big fluctuations in HRT.

The pilot digester was operated at thermophilic conditions (55°C) during the first trial and mesophilic
conditions (37°C) during the second trial. Digestion of sludge at Henriksdal is expected to be performed at
thermophilic conditions in the future since thermophilic operation is supposed to better cope with shorter
retention times and higher organic loads (Metcalf & Eddy 2014; Henze et al. 2010). The thermophilic trial was
performed to determine the critical HRT (which is equal to sludge retention time, SRT, in a continuous stirred-
tank reactor, CSTR) for thermophilic digestion and assess the risks associated to the future full-scale
operation. In order to investigate if thermophilic operation really is required in the future or if energy can be
saved by continuing to operate mesophilic, it was decided to also perform a mesophilic trial at similar
conditions as the thermophilic.

During each trial, the digester’s HRT was decreased stepwise with one day HRT at the time (Table 30 and
Table 31). Each HRT was planned to be in operation for 3-4 retention times each. If the sludge pilot
experienced operational problems affecting the trial, the time of operation was extended. For the thermophilic
trial, an initial retention time of 9 days was selected, and for the mesophilic trial 10 days, based on previous
experiences from operation of the sludge pilot at stable conditions.

Table 30 and Table 31 summarize the implementation of the thermophilic and mesophilic trial respectively.

77
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Table 30. Experimental set up for the thermophilic trial.

Duration OLR [kg Flow MS
HRT [d] Date [year-week] [×HRT] VS/m3,d] Digester V [m3] [L/h]
9 2019-28 to 2019-33 4.2 1.91 5.0 22
8 2019-34 to 2019-41 6.2 2.09 5.0 24
7 2019-42 to 2019-49 8 2.18 5.0 29
6 2019-50 to 2020-6 10.5 2.89 5.0 34
5 2020-7 to 2020-10 5.6 3.06 4.2 34
4 2020-11 to 2020-14 7 4.11 3.4 35
4 + EOM 2020-15 to 2020-17 5 9.98 3.4 35

Table 31. Experimental set up for the mesophilic trial.

HRT Date Duration OLR [kg Digester V Flow MS
[d] [year-week] [×HRT] VS/m3,d] [m3] [L/h]
10 2020-25 to 2020-29 3.5 1.56 5.0 20
9 2020-30 to 2020-37 6.2 1.66 5.0 22
8 2020-38 to 2020-41 3.5 1.63 5.0 24
7 2020-42 to 2020-44 3.0 2.11 5.0 29
6 2020-45 to 2020-47 3.5 2.42 5.0 34
5 2020-48 to 2020-51 5.6 2.45 4.6 36
4 2020-52* to 2020-01 5.3 3.21 3.4 35
4 + EOM 2021-02 1.8 4.66 3.4 35
*Year 2020 had 53 weeks.

Sampling and analysis

Besides the regular sampling and analysis program described in chapter 5.1, intensive sampling and analysis
was done on a daily basis towards the end of each trial. In addition, a sludge sample from the digester was
collected and frozen every week for DNA-analysis, which was done in cooperation with SLU. The aim was to
determine how the microbial population changes during the experiments. At the time of writing, data from
the DNA-analysis was not yet provided and results will instead be published in next year’s report. During
the mesophilic trial, autumn 2020, analysis of fat, protein and carbohydrates were done by Eurofins for three
samples per HRT of mixed sludge (MS) and digested sludge (DS) respectively. At the end of each trial, when
VFA increased over 150 mg/L, samples were taken daily and sent to KTH for GC analysis according to
Owusu-Agyeman et al. (2020).

Previous studies and expectations

Methanogens grow slowly, with a generation time of 1-12 days depending on species and environmental
conditions (Jarvis and Schnürer, 2009). Therefore, it is common to choose a retention time that is at least 12
days when designing anaerobic digesters. At higher retention times, a higher diversity of methanogens is
most likely present. When the retention time decreases only the fastest growing methanogens, like the family
of Methanosaetaceaes, will survive (Lee at al. 2011), which will make the process more vulnerable to
disturbances. Studies using semi-continuous CSTR lab scale reactors have previously shown that process
imbalances (foaming, VFA accumulation etc.) started to show at retention times below 9 days and that
digestion could be performed down to 5 days retention time at both thermophilic and mesophilic conditions
but with significantly decreased VS destruction efficiency (Nges and Liu 2010). At 3 days retention time both
processes crashed due to accumulation of VFA. Lee et al. (2011) showed that mesophilic digestion could be
maintained in lab-scale at 4 days retention time, although less efficient in terms of specific gas production, and
that a shift in microbial population occurred at 5 and 4 days retention time. Both studies were performed at

78
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

higher OLRs than the study presented here. It is difficult to find relevant references on digestion of WWTP-
sludge at low retention times in pilot- or full-scale. Since pilot trials usually experience more process
disturbances than laboratory experiments, it was not expected that the conducted pilot test would be possible
to run at such low retention times. Based on literature and experiences the expectations were set on a crash at
6 days HRT at thermophilic digestion and 7 days at mesophilic.

During autumn 2020, Jessica Sellin performed her MSc-thesis project studying parameters for process
monitoring and the indicators for digestion process instability, both by going throw operational data from the
thermophilic crash test and by following up the results from the ongoing mesophilic crash test (Sellin 2021). It
was expected that changes in VFA and alkalinity, and especially the ratio between them (VFA/TA) would be
the earliest indicators of process instability, followed by a drop in pH, as can be seen in Figure 45.

Figure 45. Important process parameters and their expected behaviour when the process is approaching a
crash (Henze et al., 2010).

79
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

6.8.3 Results thermophilic trial

An overview of the operation of the digester during the thermophilic trial can be seen in Figure 46. As can be
seen, the thermophilic digestion process was operated rather stably down to 4 days retention time. After 28
days (7 x HRT) at 4 days retention time with an OLR of just above 4 kg VS/m3,d and no sign of instability; EOM
was added in an attempt to crash the digester (since the HRT could not be lowered further). The OLR was
increased over 20 days to 15.5 kg VS/m3,d on the 24th of April when it finally crashed.

It can also be seen in Figure 46 that the actual HRT increased slightly at the end of HRT 9, 6 and 5. This was
caused by short stops in the inflow to allow the water level in the digester to decrease and thereby decrease the
reactor volume. In week 44, the inlet pump feeding the digester broke on Friday afternoon and was not fixed
until Monday morning the week after, which caused the HRT to rise from 7 to 11 days as a weekly average.

14 60
pH, HRT [d], OLR [kg VS/m3,d]

12 50
10
40

Temp. [C]
8
30
6
20
4

2 10

0 0

year-Week
OLR (kg VS/m3,d) HRT target (d) HRT actual (d) pH Temp. [C]

Figure 46. Operation during the thermophilic trial. Weekly averages.

During the thermophilic trial, problems with the heating system occurred three times, resulting in temperature
drops (Figure 46). The largest temperature decrease occurred during 15 days at HRT 6 d, starting on Christmas
eve 2019 when no operational staff worked, and no mechanics were available until after the holidays (see Figure
47). The temperature went as low as 18°C. The feeding was stopped on December 30 and resumed on January
9. The VFA increased to a maximum of 767 mg/L on January 2. In order to stop acidification and save the trial,
1.7 kg of bicarbonate mixed with water was fed to the digester on December 30 and <1 kg on January 2, resulting
in normal VFA values from January 14 onward. Although the pH dropped to 6.11, the process recovered
surprisingly well and after three retention times the “How low can we go” trial was resumed.

Data from times with process disturbances was not included in the evaluations.

80
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Figure 47. Online data on temperature, feed flow and pH from December 22, 2019 to January 17, 2020 when
the heating system broke. Note that the online pH sensor shows a value that is about one unit higher than
the actual value (Screenshot from aCurve, resolution 1 min).

Parameters often used to monitor digestion processes and more specifically the buffering system in the sludge,
are shown in Figure 48 as an average value per retention time. pH decreased from 6.96 to 6.60 before addition
of EOM while VFA decreased somewhat from 138 to 102 mg/L. At the same time, the total alkalinity (TA) went
down from 2081 to 1458 mg/L and ammonium decreased from 374 to 268 mg/L, further lowering the buffer
capacity. Since both VFA and TA decreased, the VFA/TA ratio was rather constant around 0.08. The average
values for the period when EOM was added shows a steep increase in VFA and a steep decrease in ammonium,
whereas the total alkalinity did not change much. These changes resulted in a VFA/TA ratio of 0.26 and a pH of
6.1. In conclusion, neither VFA and TA nor pH were considered good early indicators of a process collapse. The
ammonium values on the other hand could possibly be used to foresee process instability.

12 400

350
10
300
OLR [kg VS/m3,d], pH

8
250

6 200
[mg/L]

150
4
100
2
50

0 0
9 8 7 6 5 4 4+EOM

OLR [kg VS/m3,d] pH VFA [mg/l] TA/10 [mg/l] NH4-N [mg/l]

Figure 48. Parameters related to the buffering system in the digester. Note that total alkalinity (TA) values
are divided by 10. Averages for each retention time.

81
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Figure 49 shows process parameters related to degradation and biogas production as average values for each
retention time. The organic degradation rate (ODR) showed steady values between 49 and 51% of VSin except
for the period with 5 days HRT when it was only 41%. When EOM was added, the ODR increased since
EOMs are easily available substrates that are almost entirely converted to biogas. The biogas production
increased with increasing OLR and the methane content of the biogas was stable at 59-62% until HRT 5 d
when it slowly started to decrease to a final value of 54%, which is not that bad. The specific biogas
production and the specific methane production decreased slightly with decreasing HRT. One of the
parameters that changed the most was H2S in the biogas, which increased over the trial.

0.8 12
[part of VSin/gas], [m3/kg VSin], [per mille]

0.7
10

[m3/d], [kg VS/m3,d]

0.6
8
0.5
0.4 6
0.3
4
0.2
2
0.1
0.0 0
9 8 7 6 5 4 4+EOM

Gas prd. [m3/d] ODR [part of Vsin] CH4 [part of gas]

OLR [kg VS/m3,d] Spec. gas prd. [m3/kg Vsin] Spec. CH4 prd. [m3/kg Vsin]
H2S [per mille]

Figure 49. Parameters related to degradation and gas production during the thermophilic trial. Note that
H2S is given in ‰ and not ppm. Averages for each retention time

There was an approximately 200 mm thick layer of foam in the reactor throughout the trial that did not
change much over time.

Dewatering
Dewatering of digested sludge was in operation from week 31 (HRT 9 days) to week 10 (HRT 5 days). Data
from dewatering is shown in Table 32. Dewatering in the screw press failed when the retention time was
lowered from 5 to 4 days. However, no big efforts to optimize the operation were made during the trial since
it was not the central focus of the study, and better results could probably have been obtained.

Table 32. Data from dewatering of thermophilic digested sludge. At HRT 4 the dewatering failed.
HRT [days] Polymer dose [g/kg TS] TS [%] SS in reject water [mg/L]
9 6.8 28.6 863
8 7.4 26.3 723
7 12.2 24.6 412
6 12.6 24.4 956
5 12.7 24.6 2146

Last three weeks

Table 33 shows process data from the last 3 weeks of thermophilic operation when EOM was added to the
digester. It can be seen that VFA did not increase to values higher than 100 mg/L and the alkalinity, which
decreased inconsistently, didn’t drop below 1000 mg/L until the OLR reached around 10 kg VS/m3,d. VFA
82
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

didn´t reach values above 500 mg/L until the last day. Interestingly, the alkalinity increased the last two days.
The gas production increased slightly with increasing OLR and the methane content was rather similar until
the time of the crash. The specific gas production, on the other hand, decreased day by day indicating
decreased activity and/or capacity of the process. Figure 50 shows a high-resolution graph of the changes in
pH-value during the last phase of operation (HRT 4 d + EOM).

Table 33. Data from the last three weeks of thermophilic operation at 4 d HRT and addition of EOM
according to Table 28. No sampling and analyses were done on weekends and public holidays (10/4).
OLR VFA ALK NH4-N Gas prd. CH4 Spec. CH4 prd.
Date [kg VS/m3,d] pH [mg/L] [mg/L] VFA/TA [mg/L] [m3/d] [%] [m3/kg VSin]
06-apr 3.3 6.45 76 1480 0.05 236 4.69 54.3 0.23
07-apr 4.0 6.45 123 2060 0.06 234 4.98 56.3 0.21
08-apr 5.5 6.49 101 2240 0.05 406 5.99 58.6 0.19
09-apr 8.0 6.44 101 1300 0.08 182 5.73 58.3 0.12
10-apr 6.3
11-apr 6.3
12-apr 6.3
13-apr 4.8 6.36 101 1220 0.08 128 5.74 51.9 0.18
14-apr 6.5 6.34 5.06 53.1 0.12
15-apr 9.8 6.22 220 1080 0.20 140 5.92 56.0 0.10
16-apr 8.6 6.32 162 1000 0.16 116 5.59 54.7 0.10
17-apr 11.5 6.25 97 740 0.13 116 5.33 54.8 0.07
18-apr 11.8
19-apr 11.7
20-apr 11.9 6.16 209 460 0.45 20 5.32 54.5 0.07
21-apr 13.5 6.09 215 560 0.38 5.54 54.8 0.06
22-apr 14.0 6.10 120 1220 0.10 24 5.58 54.2 0.06
23-apr 14.1 5.30 697 1800 0.39 23 5.73 50.3 0.06
24-apr 15.5 4.88 1083 1200 0.90 44 1.45 46.6 0.03

Figure 50. Online data on temperature, feed flow and pH from April 6 to 27 2020. The digester was
proclaimed dead on the 24th. Note that the online pH sensor shows a value that is about one unit higher
than the actual value (Screenshot from aCurve, resolution 1 min).

83
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Sludge samples were analysed for content of different VFAs by GC as described previously. Samples were
taken on the 8th (OLR 5.5 kg VS/m3,d), 11th (OLR 6.3 kg VS/m3,d), and 22nd (OLR 14.0 kg VS/m3,d) of April
when the digester was operated at 4 d HRT and with EOM addition. Results are shown in Table 34. The most
abundant VFA at both low and high VFA concentration was acetic acid followed by propionic acid.

Table 34. Composition of VFAs in the digester around the time of the thermophilic crash. Value from
spectrophotometric cuvette tests as reference.
Date GC Results (mg/L) Cuvettes (mg/L)
Acetic Propionic Isobutyric Butyric Isovaleric Total VFA Acetic eq.
8-apr 48 8 0 0 0 56 100
11-apr 50 5 0 0 0 55 220
22-apr 1047 321 53 7 48 1476 991

Summary of performance
A summary of the results is shown in Table 35. As mentioned before, the digestion process performed stably
down to 4 days HTR without being irreversibly acidified. Translated to full-scale, this means that we probably
could operate the digesters at considerably lower retention times than expected (though not as low as 4 days),
at least if the reduction in retention time is done slowly and the OLR is relatively low. It should be noted that
the mixing of the pilot digester is most likely better than in the full-scale digesters due to the relatively high
recirculation flow, this could affect the performance since no local shock loadings occur. This information is
important when planning for maintenance work or during risk assessment of for example the thickeners
efficiency and reliability.

However, the process performance in terms of biogas production (Spec. CH4 prd., CH4) and degradation of
organic material (ODR, NH4-N), two factors important for full-scale operation of the digestion process, was
unsatisfying below 6 d HRT. Thus, operation at very low retention times should not be a standardized mode
of operation. To balance the up to 25% variations in sludge production over the year and keep a bit of a
margin, an HRT of less than 8 d is not recommended in the full-scale digesters.

Table 35. Evaluation of process performance.

Henriksdal Termophilic trial
HRT 16 d,
Parameters mesophilic Green Yellow Red HRT9 HRT8 HRT7 HRT6 HRT5 HRT4
Org. degradation rate (% of
50% >50% >45% <45% 52% 49% 48% 49% 41% 48%
VSin)
Spec. CH4-prod. (m3/kg VSin) 0.31 >0.25 >0.20 <0.20 0.21 0.21 0.21 0.26 0.18 0.17
pH 7.2 >6.90 >6.70 <6.70 6.96 6.76 6.71 6.73 6.75 6.60
VFA/TA <0.03 <0.07 <0.3 >0.3 0.08 0.06 0.07 0.12 0.08 0.07
- VFA (mg/L) <100 <150 <300 >300 138 114 113 116 108 102
- TA (mg/L) 3400 >1500 >1000 <1000 2081 2012 1554 1875 1271 1459
CH4 (% of biogas) 66 >60% >55% <55% 62% 60% 59% 59% 57% 54%
H2S (mg/L) - <50 <100 >100 38 113 218 139 176 214
NH4+-N (mg/L) - >500 >300 <300 374 314 318 355 349 268
TS after dewatering (%) 28% >26% >23% <23% 28.6 26.3 24.6 24.4 24.6 x

84
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

6.8.4 Results mesophilic trial

An overview of the operation of the digester during the mesophilic trial can be seen in Figure 51. As can be
seen, the mesophilic digestion process, just like the thermophilic one, was operated rather stably down to 4 days
retention time. After 21 days (5.3 x HRT) at 4 days retention time with an OLR of 3.2 kg VS/m3,d, EOM was
added in an attempt to crash the digester (since the HRT could not be lowered further). The OLR was increased
over 5 days to 5.3 kg VS/m3,d on the 15th of January 2021 when the process crashed.

Just like in the thermophilic trial it can be seen in Figure 51 that the actual HRT increased slightly at the end of
HRT 10, 9 (a lot) and 5. This was caused by short stops in the inflow to allow the water level in the digester to
decrease and thereby decrease the reactor volume. At the end of HRT 9, the inlet pipe got clogged during the
lowering of the volume a Friday afternoon and the pipe was not cleared until the following Monday resulting
in a high HRT peak. There was also a small failure in the heating system on week 35 with a temperature low at
33°C.

14 40

12 35
pH, HRT [d], OLR [kg VS/m3,d]

30
10
25

Temp. [C]
8
20
6
15
4
10
2 5

0 0

year-Week
OLR (kg VS/m3,d) HRT target (d) HRT actual (d) pH Temp. [C]

Figure 51. Operation during the mesophilic trial. Weekly averages.

Parameters often used to monitor digestion processes and more specifically the buffering system in the sludge,
are shown in Figure 52 as average values per retention time. pH decreased from 6.68 to 6.60 before addition of
EOM while VFA was rather constant at 65±10 mg/L until HRT 4 d when it increased to over 100 mg/L. The total
alkalinity (TA) went down from 1760 to 650 mg/L from HRT 10 to 5 d and then went up a bit to values around
1000 mg/L. Ammonium decreased almost linearly from 232 to 130 mg/L with exception of a small increase at
HRT 7 d. The VFA/TA ratio was below 0.06 until HRT 5 d when it increased to 0.13. The average values for the
period when EOM was added shows a steep increase in VFA while alkalinity remained the same. These changes
resulted in a VFA/TA ratio of 0.47 and a pH of 6.4. None of the parameters in Figure 52, not even ammonium,
provided a clear early warning of a collapse, although some of them clearly indicated irreversible instability at
HRT 4 with EOM dosage.

85
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

8 500
7 450
400
6
OLR [kg VS/m3,d], pH

350
5 300
4 250

[mg/l]
3 200
150
2
100
1 50
0 0
10 9 8 7 6 5 4 4+EOM

OLR [kg VS/m3,d] pH VFA [mg/l] TA/10 [mg/l] NH4-N [mg/l]

Figure 52. Parameters related to the buffering system in the digester. Note that total alkalinity (TA) values
are divided by 10. Averages for each retention time.

Figure 53 shows process parameters related to degradation and biogas production as average values for each
retention time. The organic degradation rate (ODR) decreased from 49% to around 30% of VSin. When EOM
was added, the ODR increased since EOMs are easily available substrates that are almost entirely converted to
biogas. The biogas production increased with increasing OLR until HRT 6 d. After that, it decreased even
though the OLR increased. The methane content of the biogas was stable at 58-60% until EOM was added
when it decreased slightly to 56%. The specific biogas production and the specific methane production
increased initially, from HRT 10 to 8 d where it peaked. Thereafter it was rather similar until EOM was added
when it dropped to 0.22 and 0.12 m3/kg VSin respectively. H2S in the biogas increased initially but then
decreased to low levels again.

0.7 5.0

0.6
4.0
0.5
[m3/d], [kg VS/m3,d]
[%], [m3/kg Vsin], [per mille]

0.4 3.0

0.3 2.0
0.2
1.0
0.1

0.0 0.0
10 9 8 7 6 5 4 4+EOM
Gas prd. [m3/d] ODR [part of Vsin] CH4 [part of gas]
OLR [kg VS/m3,d] Spec. gas prd. [m3/kg Vsin] Spec. CH4 prd. [m3/kg Vsin]

Figure 53. Parameters related to degradation and gas production during the mesophilic trial. Note that H2S
is given in ‰ and not ppm. Averages for each retention time.

The degradation of different components of the sludge at different HRT:s is found in Table 36. The analyses
were initiated in September, which means that the first data presented are from 8 days retention time. The

86
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

results did not show distinct trends, which could be due to difficulty to accurately analyse sludge
(measurement uncertainty 20-30%). The fat analyses gave very varied results in general which could depend
on the fact that fat often aggregates making it hard to get a representative sample for the analysis. The results
for fat in Table 36 are random making it difficult to draw any conclusions about degradation of fat.
Carbohydrates were degraded to slightly higher extent at higher HRTs, which corresponded to the results for
VS. For proteins the degradation decreased with decreasing HRT with exception for HRT 7. This correlates
with the results from ammonium in the digester presented in Figure 53, showing decreasing concentration
with decreasing HRT. The degradation COD was 38 ± 5 % with no correlation to the HRT.

Table 36. The degradation (% of incoming) of sludge components at different retention times. Averages of
3-4 analyses.
HRT Carbohydrates Proteins Fats VS COD
8 63% 27% 25% 38% 33%
7 73% 15% -13% 43% 43%
6 48% 20% 14% 32% 37%
5 47% 18% 57% 35% 41%
4 53% 15% 21% 34% 32%

There was an approximately 280 mm thick layer of foam in the reactor throughout the trial that did not
change much over time.

Dewatering
During the period with a retention time of 10 days the dewatering was not in operation. From week 35 (HRT 9
days) to week 46 (HRT 6 days) the digested sludge was dewatered. Data from dewatering is shown in Table
37. Dewatering in the screw press failed after two weeks at HRT 6 days. However, no big efforts to optimize
the operation were made during the trial since it was not the central focus of the study, and better results
could probably have been obtained. It should also be noted that the polymer used was the one that was tested
out for the thermophilic sludge. No test of different polymers was done for the mesophilic sludge. This might
affect the results and the comparison with the thermophilic results.

Table 37. Data from dewatering of mesophilic digested sludge. At HRT 5 the dewatering failed.
HRT [days] Polymer dose [g/kg TS] TS [%] SS in reject water [mg/L]
10 – no operation
9 7.7 23.0 210
8 12.0 21.2 283
7 15.5 20.1 953
6 14.4 18.7 1300

Last week
Table 38 shows process data from the last week of mesophilic operation when EOM was added to the
digester. It can be seen that VFA and the VFA/TA ratio increased while total alkalinity (TA) and NH4-N
decreased day by day, indicating that the process was on the verge of giving up already before EOM was
added. The gas production initially increased slightly with increasing OLR while the methane content
decreased a bit every day. The specific methane production did not change much until the last day when it
decreased to 0.09 m3/kg VSin. Figure 54 shows a high-resolution graph of the changes in pH value during the
last phase of operation (HRT 4 d + EOM).

87
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Table 38. Data from the last week of mesophilic operation at 4 d HRT + EOM addition according to Table
29.
NH4-
OLR VFA ALK N Gas prd. CH4 Spec. CH4 prd.
Date [kg VS/m3,d] pH [mg/L] [mg/L] VFA/TA [mg/L] [m3/d] [%] [m3/kg VSin]
11-jan 3.9 6.67 149 1480 0.10 156 3.67 59.4 0.16
12-jan 4.9 6.66 179 1030 0.17 152 4.04 58.4 0.13
13-jan 4.3 6.51 329 840 0.39 134 4.10 57.1 0.15
14-jan 4.0 6.32 366 870 0.42 128 3.92 55.5 0.15
15-jan 5.3 6.04 847 580 1.46 104 3.34 53.5 0.09

Figure 54. Online data on temperature, feed flow and pH from January 10 to 15 2021. The digester was
proclaimed dead on the 15th. Note that the online pH sensor shows a value that is about one unit higher
than the actual value (Screenshot from aCurve, resolution 1 min).

In addition to the daily measurements of total-VFA using spectrophotometric methods, GC was used to
measure the VFA the day before the crash, January 14 (OLR 4.5 kg VS/m3,d), and the day of the crash January
15 (OLR 5.3 kg VS/m3,d) and 3 days after the crash, on January 18 (OLR not measured). Results can be seen in
Table 39. Propionic acid was the most prevailing VFA followed by acetic acid and isobutyric acid.

Table 39. Composition of VFAs in the digester around the time of the mesophilic crash. Value from
spectrophotometric cuvette tests as reference.
GC Results (mg/L) Cuvettes (mg/L)
Date Acetic Propionic Isobutyric Butyric Isovaleric Total VFA Acetic eq
14-jan 108 503 17 0 8 636 366
15-jan mor. 100 786 63 0 17 966 711
15-jan aft. 121 950 96 3 18 1188 847
18-jan 92 298 47 7 27 470 393

Summary of performance
A summary of the results is shown in Table 40. As mentioned before, the digestion process performed stably
down to 4 days HTR also at mesophilic operation without being irreversibly acidified. However, the process
performance in terms of biogas production (Spec. CH4 prd., CH4) and degradation of organic material (ODR,
NH4-N), two factors important for full-scale operation of the digestion process, was unsatisfying below 9 d

88
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

HRT. With the same reasoning as for the thermophilic results, operation at an HRT of less than 11.5 d is not
recommended for the full-scale digester.

Table 40. Evaluation of process performance.

Henriksdal Mesophilic trial
HRT 16 d,
Parameters mesophilic Green Yellow Red HRT9 HRT8 HRT7 HRT6 HRT5 HRT4
Org. degradation rate (% of VSin) 50% >50% >45% <45% 47% 38% 43% 32% 34% 34%
Spec. CH4-prod. (m3/kg VSin) 0.31 >0.25 >0.20 <0.20 0,19 0,23 0,20 0,21 0,20 0,19
pH 7.2 >6.90 >6.70 <6.70 6,62 6,53 6,48 6,52 6,56 6,60
VFA/TA <0.03 <0.07 <0.3 >0.3 0,06 0,06 0,07 0,07 0,13 0,11
- VFA (mg/L) <100 <150 <300 >300 68 73 60 55 83 112
- TA (mg/L) 3400 >1500 >1000 <1000 1241 1271 825 890 648 1031
CH4 (% of biogas) 66 >60% >55% <55% 59% 58% 58% 59% 59% 59%
H2S (mg/L) - <50 <100 >100 86,7 160,9 179,2 133,0 96,2 64,9
NH4+-N (mg/L) - >500 >300 <300 185 189 231 172 164 144
TS after dewatering (%) 28% >26% >23% <23% 23,0 21,2 20,1 18,7 x x

6.8.5 Comparison
Figure 55 below shows comparative graphs from the thermophilic and mesophilic trials related to
degradation of substrate and production and quality of biogas. Values from the thermophilic trial are
generally higher than in the mesophilic trial except for the methane concentration that was quite similar in the
two trials. The higher biogas production could be due to the slightly higher OLR in the thermophilic trial (see
Figure 44). The higher degradation rate (ODR) during thermophilic digestion was also confirmed by the
higher ammonium concentrations (Figure 56) in the reactor, derived from degradation of proteins. Figure 56
shows a comparison of VFA, TA, VFA/TA, NH4-N and pH during the thermophilic and mesophilic trials.
Both VFA and alkalinity was significantly higher during thermophilic digestion compared to mesophilic
although the difference decreased with decreasing HRT. VFA/TA did not differ much but was a bit higher at
thermophilic digestion initially, but from 5 d HRT the ratio was higher during mesophilic digestion.
Ammonium was around twice as high in thermophilic digestion except for the last phase (4 d HRT + EOM)
when both temperatures gave equal values. pH was below 7 in both trials but was about 0.3 units higher
during thermophilic digestion until HRT reached 4 days.

Overall, thermophilic digestion showed better process values at low HRTs than mesophilic digestion.

89
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

ODR [% of kg VSin] Gas production [m3/d]

80% 6
70% 5
60%
50% 4
40% 3
30%
2
20%
10% 1
0% 0

HRT HRT
mesophilic trial thermophilic trial mesophilic trial thermophilic trial

Specific Gas and CH4 production [m3/kg VSin]

0.5
0.4
0.3
0.2
0.1
0.0
10 9 8 7 6 5 4 4+EOM
HRT
Gas - mesophilic trial CH4 - mesophilic trial Gas - thermomphilic trial CH4 - thermophilic trial

H2S [ppm] CH4 [% of biogas]

70%
600
60%
500
50%
400
40%
300
30%
200 20%
100 10%
0 0%

HRT HRT
mesophilic trial thermophilic trial mesophilic trial thermophilic trial
Figure 55. Graphs on degradation and biogas comparing the thermophilic and mesophilic trials.

90
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

TA [mg/L] VFA [mg/L]

2 000 450
400
350
1 500 300
250
1 000 200
150
500 100
50
0 0

HRT HRT
mesophilic trial thermophilic trial mesophilic trial thermophilic trial

VFA/TA [mg/mg] NH4-N [mg/L]

0.50 400
350
0.40
300
0.30 250
200
0.20
150
0.10 100
50
0.00 0

HRT HRT
mesophilic trial thermophilic trial mesophilic trial thermophilic trial

pH
7.0

6.8

6.6

6.4

6.2

6.0
10 9 8 7 6 5 4 4+EOM
HRT
mesophilic trial thermophilic trial
Figure 56. Graphs on parameters related to pH and the sludge buffer system, comparing the thermophilic
and mesophilic trials.

91
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

6.8.6 Method comparison for VFA and CH4 production

Jessica Sellin (2021) compared different methods for analysing VFA in the digesters and for estimating the
methane production as part of her Master thesis. The results show that the spectrophotometric cuvette tests
are best at low concentrations, <100 mg/L, while both HPLC, titration and GC give more reliable results at
high concentrations, >200 mg/L (Figure 57). Cuvette analysis of pure acetate and propionate at concentrations
between 80 and 1600 mg/L showed a 100% accuracy for acetate but only 66% for propionate. Since the VFA in
the digester was low during normal operation and acetate was the dominating VFA, even at low HRTs, it was
concluded that the cuvette tests are best suited for monitoring of the digestion process.

Figure 57. Comparison of VFA analysis methods. SP = spectrophotometric cuvette test. From Sellin (2021).

During a short period of the trials, there were problems with the gas flow meter. Therefore, Jessica also
investigated how well a theoretically calculated biogas and methane production based on either VFA, COD or
protein/fat/carbohydrate degradation corresponds to the measured values (Sellin 2021). Results, shown in
Figure 58, show that calculations based on protein/fat/carbohydrate degradation gave the most accurate
results with a 98% (±9%) accuracy. Calculations based on VS underestimated the methane production with
around 13% while COD overestimated the production with 22% as an average.

Figure 58. Theoretical methane production for mesophilic trial at HRT 8, 7 and 5 based on degradation of
VS, PFC (protein, fat and carbohydrate) and COD as percentage of measured methane production. From
Sellin (2021).

92
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

6.9 Resource consumption

Resource consumption in the pilot for 2020 is summarized in Table 41. A comparison with the future
Henriksdal design was made where possible (where design values are available). Pilot values contain
uncertainties due to problems with pumps, air in the tubes, difficulties in manually reading levels and
degradation of some chemicals, to mention a few sources for errors. Also, the design values for the future
Henriksdal WWTP are uncertain since they are theoretical values derived process calculations based on
design manuals and experiences from other WWTPs.

This year, two different external carbon sources were tested. Glycerol, which have been tested previous years
was used in the pilot for about 42 weeks in 2020. A shorter 10 weeks trial with ethanol was conducted before
switching back to glycerol as external carbon source for the remainder of the year. A pause in dosage of
external carbon source week 41 to 48 has been excluded from resource calculations. Comparing the two
carbon sources, ethanol and glycerol were similar in consumption per day and per kg N removed. The
consumption of external carbon source, expressed as COD, was the same as for the future Henriksdal design,
even though the nitrogen load on the biology was 18% higher and set-point for nitrate in the effluent was
>25% lower in the pilot.

The daily iron/metal consumption was about half compared to the future Henriksdal design, although the
phosphorus load to the pilot was about 50% higher compared to design values. This is reflected in the mole Fe
to mole P consumption, which was lower in the pilot compared to the design. The effluent phosphate
concentration was below the target concentration and the measured values were about half of those set in the
design calculations. The low metal consumption is explained by the EBPR activity in the pilot. This year, both
consumption of precipitation chemicals and effluent phosphate were the lowest measured in the history of the
pilot.

The consumption of cleaning chemicals was considerably lower than the pilot design values and the future
Henriksdal design. The effort to reduce the cleaning chemicals is described in chapter 6.5.3. In total, the
amount of chemicals used for RC was around the same as in the design, since RC was performed twice in
2020. Chemical consumption for MC, however, was only 22% of pilot design values for oxalic acid, 31% for
citric acid and 24% for hypochlorite. Design values are based on the design inflow of 3.2 m3/h while the inflow
in 2020 was 4.4 m3/h. Since MC frequency is based on the volume of water passing the membranes this 38%
increase in inflow should also render a 38% increase in chemical consumption. Based on the actual inflow, the
chemical consumption for MC was reduced by 84% for oxalic acid, 78% for citric acid and 82% for
hypochlorite. The results indicate that costs and environmental impact can be significantly reduced in the full-
scale plant by reducing the consumption of cleaning chemicals.

The air demand for the biological treatment cannot be compared in a reasonable way since the configuration
of the reactors are completely different with a water depth of 12 m in the full scale, compared to 3.2 m in the
pilot. The airflow to the membranes was 145% larger than the design value for future Henriksdal, which could
mainly be explained by the fact that the design values given in the table are based on design inflow and
membrane aeration at LEAP-lo.

93
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Table 41. Chemical consumption during 2020.

Value pilot/
Value Future H-dal scaled future
Resource Unit pilot design H-dal design
External carbon source (Methanol, not used kg COD/d 12 000 -
in pilot) g COD/g N - -
External carbon source (Glycerol, 253 days) kg COD/d 1.75 - -
g COD/g N* 0.60 - -
External carbon source (Ethanol, 68 days) kg COD/d 1.80 - -
g COD/g N* 0.55 - -
External carbon source (COD) kg COD/d 1.80 12 000 100%
Iron (FV) kg Fe/d 0.7 10 000 47%
mole Fe/mole P 2.8 -
Aluminium (BR6) kg Al/d - -
mole Al/mole P 0.05 - -
Metals (Iron + Aluminium) kg Me/d 0.75 10 000 50%
mole Me/mole P 0.95 2.8
Citric Acid (51%) L/d 0.066** 1 100 40%
Sodium hypochlorite (12%) L/d 0.150*** 1 600 63%
Oxalic acid (8%) L/d 0.27** 5 000 36%
Aeration biology m3/d 1 734 1 600 000 726%
Aeration MT m3/d 648 3 000 000 145%
* N removed in total, from inlet to effluent.
** Number of MCs with each acid, multiplied with time settings and number of backpulses using settings of flowrate of
chemical. Measured consumption for two RC preformed with each acid.
*** Number of MCs with hypo was multiplied with time settings and number of backpulses using design flowrate of
chemical. Measured consumption for four RC preformed in total (2 per MT).

94
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

7 Related publications
Several activities such as world water congress were canceled in 2020 due to CIVID-19. Further, the general
cooperation with different partners at the R&D-facility had to be restricted. As such, fewer publications were
generated from the project during 2020.

The project had one contribution at a digital international conference held in 2020:

 Karlsson J., Carranza Muñoz A. (2020) How low can you go – retention time reduction until
microbial inhibition in a thermophilic digester treating wastewater sludge. IWA Latin American
Meetings on Anaerobic Digestion, 22 October – 12 November 2020.

The project hosted three master students and following master theses were published:

 Fridh, B. (2020) Mapping of spontaneous biological phosphorus removal in MBR-process. KTH Royal
Institute of Technology. TRITA-CBH-GRU; 2020:116
 Roberts, R.A. (2020) Mapping spontaneous biological phosphorus removal observed in a membrane
bioreactor process without the anaerobic condition. KTH Royal Institute of Technology, TRITA-ABE-
MBT 20734.
 Sellin, J. (2021) Comparison of mesophilic and thermophilic anaerobic sludge digestion at
Hammarby Sjöstadsverk MBR pilot plant for wastewater treatment. Master thesis in Environmental
Science 2021:09, SLU Swedish University of Agricultural Sciences, Uppsala.

A related project, financially supported by Svenskt Vatten Utveckling, investigated what happens to different
types of micro pollutants in a WWTP with membrane technology was published early 2021:

 Närhi, K., Westling, K., Andersson, S., Baresel, C. & Wahlberg, C. (2021) Micro pollutants in wastewater
treatment plant with MBR-process. Comparison with conventional treatment plant and calculation of impact
on recipient. SVU-report 2021-2.

95
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

8 Bibliography
Andersson, S. L., Westling, K., Andersson, S., Laurell C., Baresel, C., Narongin, M., Royen, H. & Bornold, N.
(2017) Pilotförsök med membranreaktor för avloppsvattenrening, Delrapport 3 - försöksår 3. IVL B2285.

Andersson, S. L. Westling K., Andersson S. & Lindblom E. (2019) Long term trials with membrane bioreactor for
enhanced wastewater treatment -pilot Henriksdal 2040. IVL B2334.

Andersson, S. L., Westling K., Andersson S., Karlsson J., Narongin M. & Persson G. (2020) Long term trials with
membrane bioreactor for enhanced wastewater treatment coupled with compact sludge treatment -pilot Henriksdal 2040,
results from 2018. IVL B2388.

Andersson, S. L., Westling K., Andersson S., Karlsson J., Narongin M., Carranza Munoz, A. & Persson G.
(2021) Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact sludge
treatment -pilot Henriksdal 2040, results from 2019. IVL B2409.

Arbetsmiljöverket 2018. AFS 2018:1 Hygieniska gränsvärden. ISBN 978-91-7930-649-6.

Brepols, C. (2010) Operating Large Scale Membrane Bioreactors for Municipal Wastewater Treatment. IWA
Publishing ISBN: 9781843393054.

Fridh, B. (2020) Mapping of spontaneous biological phosphorus removal in MBR-process. KTH Royal Institute
of Technology. TRITA-CBH-GRU; 2020:116

Gkotsis, P. & Zouboulis, A. (2019) Biomass Characteristics and Their Effect on Membrane Bioreactor Fouling, review.
Molecules, 24, 2867; doi:10.3390/molecules24162867.

Henze, M., Petersen, G., Holm Kristensen, G., Höök, B. (2010) Drift av renseanlaeg – Teknik. 3rd Ed. Center
for Offentlig Kompetenceudvikling. ISBN: 9778-87-7848-977-7-pdf

Jacobs J.H., Spaan S., van Rooy G.B.G.J., Meliefste C., Zaat V.A.C., Rooyackers J.M., Heederik D. Exposure to
trichloramine and respiratory symptoms in indoor swimming pool workers. Eur Respir J 2007; 29: 690–698.

Jansen J. la Cour, Särner E., Tykesson E., Jönsson K. & Jönsson L-E. (2009) Biologisk fosforavskiljning i Sverige –
Uppstart och drift. SVU-rapport 2009-08. Svenskt Vatten AB.

Janssen, P.M.J., Meinema, K. & van der Roest, H.F. (2002) Biological phosphorus removal - Manual for design and
operation. London: IWA Publishing.

Jarvis, Å., Schnürer, A. (2009) Mikrobiologisk handbok för biogasanläggningar. Svenskt Gastekniskt Center
Rapport 207 1102-7371 (ISRN SGC-R-207-SE)

Judd, S. (2010) The MBR book. Butterworth-Heinemann, 2nd Ed. ISBN:9780080966823.

Judd & Judd Limited, (2017) The MBR site. [Online]

Available at: www.thembrsite.com

96
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

Karlsson J., Carranza Muñoz A. (2020) How low can you go – retention time reduction until microbial
inhibition in a thermophilic digester treating wastewater sludge. IWA Latin American Meetings on Anaerobic
Digestion, 22 October – 12 November 2020.

Lee, I.S., Parameswaran, P., E. Rittman, B. (2011) Effects of solids retention time on methanogenesis in
anaerobic digestion of thickened mixed sludge. Bioresource Technology 102, pp. 10266-10272.

Metcalf & Eddy (2014) Wastewater Engineering, Treatment and Resource Recovery, 5th Ed., McGraw-Hill
International Edition. ISBN 978-1-259-01079-8.

Nges, I.A., Liu, J. (2010) Effect of solids retention time on anaerobic digestion of dewatered sewage sludge in
mesophilic and thermophilic conditions. Renewable Energy 35 p. 220-2206.

Närhi, K., Westling, K., Andersson, S., Baresel, C. & Wahlberg, C. (2021) Mikroföroreningar i reningsverk med
MBR-process - Jämförelse med konventionellt reningsverk och recipientpåverkan. SVU-Rapport nr 2021-2.

Owusu-Agyeman, I., Plaza, E., Cetecioglu, Z. (2020) Production of volatile fatty acids through co-digestion of
sewage sludge and external organic waste: Effect of substrate proportions and long-term operation. Waste
Management 112 pp. 20-39.

Pellegrin, M.-L. & Neethling J. B. (2015) Application of Membrane Bioreactor Design Processes for Achieving Low
Effluent Nutrient Concentrations. Water Environment Research Foundation (WERF), IWAP ISBN: 978-1-78040-
675-6/1-78040-675-4.

Roberts, R.A. (2020) Mapping spontaneous biological phosphorus removal observed in a membrane
bioreactor process without the anaerobic condition. KTH Royal Institute of Technology, TRITA-ABE-MBT
20734.

Salmonsson T., Jönsson K., Andersson S., Bergslilja E. & Erikstam S. (2017) Sidoströmshydrolys och biologisk
fosforavskiljning på svenska avloppsreningsverk. SVU-rapport 2017-06. Svenskt Vatten AB.

Samuelsson, O., Royen, H., Ottosson, E., Baresel, C., Westling, K., Bergström, R., Bengtsson, L., Yang, J.J.,
Andersson, S.L., Björk, A., Dahlén N, Laurell, C., Lindblom, E. & Grundestam, J. (2014). Pilotförsök med
membranbioreaktor för avloppsvattenrening, Delrapport 1 - Försöksår 1. IVL B2215.

Sellin, J. 2021. Comparison of mesophilic and thermophilic anaerobic sludge digestion at Hammarby
Sjöstadsverk MBR pilot plant for wastewater treatment. Master thesis in Environmental Science 2021:09, SLU
Swedish University of Agricultural Sciences, Uppsala.

Svenskt Vatten (2010a) Avloppsteknik 2 - Reningsprocessen. Stockholm: Svenskt Vatten AB.

Svenskt Vatten (2010b) Avloppsteknik 3 - Slamhantering. Stockholm: Svenskt Vatten AB.

Thickett K.M., McCoach J.S., Gerber J.M., Sadhra S., Burge P.S. Occupational asthma caused by chloramines in
indoor swimming-pool air. Eur Respir J 2002; 19: 827–832.

Tykesson E., Jönsson L.-E. & la Cour Jansen J. (2005) Experience from 10 years of full-scale operation with enhanced
biological phosphorus removal at Öresundsverket. Water Science & Technology 52:12, pp 151–159.

UKWIR (2015) National Screen Evaluation Facility - Inlet Screen Evaluation Comparative Report (1999–2015).
Report no 15/WW/06/10.
97
Report B 2425 ­ Long term trials with membrane bioreactor for enhanced wastewater treatment coupled with compact
sludge treatment – -pilot Henriksdal 2040, results from 2020

USEPA, United States Environmental Protection Agency (2013) Wastewater Treatment Fact Sheet: External
Carbon Sources for Nitrogen Removal, EPA 832-F-13-016, Office of Wastewater Management, Washington, DC,
USA, 2013.

van Loosdrecht M.C.M., Nielsen P.H., Lopez-Vazquez C.M. & Brdjanovic D. (2016) Experimental Methods in
Wastewater Treatment, IWA Publishing, 2016.

Wastensson G., Eriksson K. The Nordic Expert Group for Criteria Documentation of Health Risks from
Chemicals. 152. Inorganic chloramines. Arbete och Hälsa (Work and Health) No 2019; 53(2).

Westling, K., Andersson, S.L., Baresel, C., Royen, H., Ottosson, E., Bergström, R., Björn, A., Andersson, S.,
Dahlén, N., Lindblom, E. & Laurell, C. (2016). Pilotförsök med membranreaktor för avloppsvattenrening, Delrapport
2 - Försöksår 2. IVL B2271.

WHO 2006. Guidelines for safe recreational water environments, volume 2: Swimming pools and similar
environments. ISBN 92-4-154680-8.

98
IVL Swedish Environmental Research Institute Ltd.
P.O. Box 210 60 // S-100 31 Stockholm // Sweden
Phone +46-(0)10-7886500 // www.ivl.se