Like many other large cities, Stockholm is facing increased urbanization with densification of infrastructure as a result. At the same time, implementation of the Baltic Sea Action Plan and the EU Water Framework Directive is expected to result in more stringent effluent quality demands. The current situation gives rise to new challenges for the municipal wastewater treatment plants (WWTPs). This paper describes how two of Sweden's largest municipal water organizations; Stockholm Vatten and Syvab, will face these challenges using ultrafiltration (UF) membrane bioreactor (MBR) technology. The effluent requirements for the rehabilitated plants are expected to be tightened to 6 mg/l and 0.2 mg/l for total nitrogen (TN) and total phosphorus (TP), respectively.

INTRODUCTION

Stockholm Vatten

Stockholm Vatten operates the Henriksdal (780,000 pe1) and Bromma (219,000 pe) wastewater treatment plants (WWTPs) in central Stockholm. The plants are largely underground in constructed rock caverns. They are configured similarly, with pre-, simultaneous- and post- precipitation in primary clarifiers, modified Ludzack-Ettinger (MLE) activated sludge processes, and, finally, dual-media sand filters, see Figure 1. Today's common effluent requirements for the two plants, 10 mg TN/l and 0.3 mg TP/l, are usually met but the annual population growth rate of 1.5% (15 to 20,000 persons/year) is making this more challenging for their operation. The plants are also facing relatively large, maintenance investment needs. Feasibility studies considering different strategic alternatives for future wastewater treatment in Stockholm were therefore begun in 2011.
Figure 1

The current configuration of the treatment process at Henriksdal WWTP.

Figure 1

The current configuration of the treatment process at Henriksdal WWTP.

The study reports showed two feasible alternatives, both meeting the future, more stringent, demands and increased loads. These were either (1) to extend and upgrade the Bromma and Henriksdal WWTPs; or, (2) to decommission Bromma and direct its influent wastewater to an extended version of Henriksdal WWTP. Although the second option was judged to be more complicated technically, as well as more expensive, it was the strategy selected.

One driver for decommissioning Bromma WWTP is that part of the plant is ‘outdoors’ in a popular residential area, which is highly valued for future development. In addition, by constructing a new, 15 km sewer tunnel from Bromma to Henriksdal, several combined sewer overflows can be removed, with benefits for the local environment. The decision to start decommissioning planning for Bromma, and upgrade and extend Henriksdal WWTP was taken in May 2014. Thereby, two large WWTPs will be merged into one, centrally located WWTP serving 1.6 million persons.

Syvab

Syvab operates Himmerfjärden WWTP (250,000 pe), 30 km south of Stockholm. The plant has been remodeled several times; new processes have been added over time to meet new demands. The result is a complex plant with many process steps, using several kinds of chemicals and with high energy consumption. The present layout consists of pre-precipitation, nitrification in activated sludge, particle polishing in tertiary clarification, followed by methanol dosing to fluidized beds for denitrification, and particle separation in disc-filters, see Figure 2. Himmerfjärden WWTP has suffered from severe sludge bulking and foaming problems, leading to the installation of ozone for filament control.
Figure 2

The current configuration of the treatment process at Himmerfjärden WWTP.

Figure 2

The current configuration of the treatment process at Himmerfjärden WWTP.

The planning for an extension to meet expected future effluent requirements and the increased load (335,000 pe in 2040) started in 2010 with a conceptual design study challenge in which several consultancy firms participated. The goal was a flexible design with good redundancy and few treatment steps. A proposal for activated sludge with step-feed pre-denitrification followed by polishing filters was crowned the winner. A feasibility study (between 2010 and 2012), including pilot-trials on disc-filters and membrane bioreactor (MBR)-technology, further investigated different process options in more detail.

CHOICE OF TECHNOLOGY

One striking feature of an MBR-process is that the sludge concentration in the bioreactor can be increased significantly; up to 14 g MLSS/l (Itokawa et al. 2008), providing a higher organic treatment capacity per m3 reactor volume than conventional activated sludge (CAS). In addition, the space occupied by the membrane installation is usually smaller than that required for secondary clarification.

Considering the design figures (Table 1) and effluent requirements, the capacity of the Henriksdal WWTP needs to be more than doubled. The plant is built in a cavern, and extension for larger treatment volumes is difficult and expensive. By introducing MBR, enough organic and hydraulic capacity can be maintained within the existing biological treatment plant volume. Based on the feasibility study, a multi-disciplinary conceptual design for extension of the whole WWTP was finalized in January 2015. Pilot-scale testing – 1:1,000 – is currently in hand to verify the biological process design, and gain experience of membrane operation and performance. The membrane supplier was procured in 2015 and the first of a total of seven MBR-lines is planned to be in operation in 2017.

Table 1

Design data for the future Henriksdal and Himmerfjärden WWTPs

Parameter Unit Henriksdal WWTP Himmerfjärden WWTP 
Load in year 2014a pe 800,000 250,000 
Design load3 pe 1,600,000 335,000 
Average daily flow m3/d 521,000 134,000 
Design flow, Qdesign m3/s 6.1 1.6 
Max flow to biological treatment m3/s 10 2.8 
Peak flow m3/s 19 3.7 
BOD5 kg/d 98,700 20,400 
Tot-N kg/d 19,500 5,000 
Tot-P kg/d 2,600 540 
Sludge concentration bioreactor g MLSS/l 7,5 
Sludge conc. membrane tanks g MLSS/l 10 10 
Volume–bioreactor (total) m3 204,000 67,000 
Volume–membrane tanks (total) m3 40,000 Depends on supplier 
Aerated sludge age 6 to 12 9 to 13 
Total sludge age 28 29 
Design temperature, min/average °C 10/15 10/14 
Design fluxb l/m2, h 30 Depends on supplier 
Installed membrane area m2 1,600,000 Depends on supplier 
Return sludge flow ×Qdesign 3–4 
Nitrate recirculation flow ×Qdesign 1–4 
Parameter Unit Henriksdal WWTP Himmerfjärden WWTP 
Load in year 2014a pe 800,000 250,000 
Design load3 pe 1,600,000 335,000 
Average daily flow m3/d 521,000 134,000 
Design flow, Qdesign m3/s 6.1 1.6 
Max flow to biological treatment m3/s 10 2.8 
Peak flow m3/s 19 3.7 
BOD5 kg/d 98,700 20,400 
Tot-N kg/d 19,500 5,000 
Tot-P kg/d 2,600 540 
Sludge concentration bioreactor g MLSS/l 7,5 
Sludge conc. membrane tanks g MLSS/l 10 10 
Volume–bioreactor (total) m3 204,000 67,000 
Volume–membrane tanks (total) m3 40,000 Depends on supplier 
Aerated sludge age 6 to 12 9 to 13 
Total sludge age 28 29 
Design temperature, min/average °C 10/15 10/14 
Design fluxb l/m2, h 30 Depends on supplier 
Installed membrane area m2 1,600,000 Depends on supplier 
Return sludge flow ×Qdesign 3–4 
Nitrate recirculation flow ×Qdesign 1–4 

a70 g BOD7/pe, d (BOD7 is the Swedish standard BOD-analysis in Sweden).

bDesign at peak weekly flow and lowest average monthly temperature.

Himmerfjärden WWTP does not have the same spatial limitations as Henriksdal. The driving forces for selection of MBR were the stringent, future, effluent quality demands, and the possibility of using existing infrastructure, as well as the need to have fewer process steps, and to avoid the impact of severe sludge bulking and foaming on the treatment results. The final investment decision will not be taken until the new treatment demands have been established by the local authorities (expected in autumn 2015).

It is recognized at both Henriksdal and Himmerfjärden that the almost complete absence of solids in the MBR effluent will be beneficial, in view of the low effluent TP requirement of 0.2 mg/l. The low effluent TSS also means that the process is well suited to meeting expected future demand on subsequent treatment steps to remove pharmaceutical residues.

PROCESS DESIGN

The design figures for both WWTPs are shown in Table 1.

Treatment strategy

Important parameters for membrane design include projected flow (average, and peak hourly, daily, weekly and monthly), water temperature and the bioreactor sludge concentration required (Judd 2011). The locations of the two plants and the connected combined sewer systems lead to lengthy periods with high flows and cold wastewater, due to snow-melt in spring.

By allowing pre-treated and cold wastewater to bypass the MBR at high flows, the membrane design can be made more cost- and resource- efficient. The projected peak flows are 19 m3/s and 3.7 m3/s for Henriksdal and Himmerfjärden, respectively, but the MBRs will be designed with hydraulic capacities of 10 m3/s and 2.8 m3/s. At Henriksdal the hydraulic capacity in the existing infrastructure is the main limiting factor. By-passed wastewater is accounted for in the effluent discharge of both plants and, to ensure that effluent quality requirements are met, enhanced pre-precipitation and polishing in existing sand filters will be employed for by-pass wastewater at Henriksdal. At Himmerfjärden WWTP, precipitation, flocculation and disc filters have been evaluated at pilot-scale with good results, and will be used for stormwater treatment of the by-pass flows.

Primary treatment

The majority of MBRs in operation lack primary clarifiers (Itokawa et al. 2008; Judd 2011). Both Henriksdal and Himmerfjärden WWTP will, however, continue with primary clarification, following step screen and grit chamber stages. Primary clarification reduces the organic load on the biological treatment process and enhances biogas production. Both plants deliver biogas for vehicle fuel.

The last stage of primary treatment is fine sieves to remove hair and fibers, which might damage the membranes (Itokawa et al. 2008; Brepols 2011). At Himmerfjärden there will be eight, fine sieves in a building beside the biological treatment stage, see Figure 3. At Henriksdal two, rotary drum, punch-hole (2 mm) screens will be installed in each of seven parallel treatment lines. This configuration gives less redundancy than that at Himmerfjärden but, due to the existing cavern's configuration, no other options were available. The choice of sieve type was in accordance and linked with the process warranty demands of the membrane supplier.
Figure 3

Layout and flow diagram of the biological treatment stage at Himmerfjärden WWTP after reconstruction.

Figure 3

Layout and flow diagram of the biological treatment stage at Himmerfjärden WWTP after reconstruction.

MBR treatment

At both WWTPs, the membranes will be placed in separate tanks. This strategy is easier and more resource efficient to operate (Itokawa et al. 2008) and, in the case of Henriksdal, the existing secondary clarifiers can be used as membrane tanks (Figure 4). New volumes are planned for Himmerfjärden WWTP (Figure 3). To distribute the sludge evenly over the MBR and avoid sludge cake formation in the membrane tanks, the return activated sludge (RAS) should be 2 to 5 times the influent flow rate (Q) (Brepols 2011), i.e. higher than in CAS. Also, in contrast to CAS, the oxygen content in the RAS is high, because of membrane air scouring to avoid clogging and reduce fouling.
Figure 4

Henriksdal WWTP: the current layout (a) and flow-scheme of one of 7 biological treatment lines after reconstruction (b).

Figure 4

Henriksdal WWTP: the current layout (a) and flow-scheme of one of 7 biological treatment lines after reconstruction (b).

Nitrogen removal: High oxygen concentrations in the RAS are fatal for pre-denitrification in the MLE configuration (Figure 1) used at Henriksdal today. To optimize carbon use in the influent and the oxygen in RAS, the first approach for the new Henriksdal MBR was to pump RAS to the oxic zone. Nitrate (NO3) was then to be recirculated to the pre-anoxic zone with deoxygenized sludge, as in Crawford et al. (2006), see Figure 5. Two drawbacks were identified for implementation of this configuration at Henriksdal, however. First, the MLSS in the pre-anoxic zones will not be maximized, thus reducing capacity, and, secondly, it will yield very high internal flows. For the design of Henriksdal (Table 1), it is assumed that at least the same effluent TN concentration (10 mg N/l) can be reached by pre-denitrification in the future plant. The decrease to 6 mg N/l is, according to the design, made possible by post-denitrification in the final, post-anoxic, zone. The large RAS-flow (3 to 4 × Qin) with low NO3- concentration due to the low effluent TN concentration, results in a maximum pre-denitrification capacity of <65% of the NO3 produced, if the NO3- recirculation flow is kept at 4 × Q. Therefore, with this configuration, 10 mg N/l cannot be achieved by pre-denitrification. This motivated the choice of another configuration in which RAS is recirculated to the first RAS-deox zone (described below), as in the configuration shown in Figure 6.
Figure 5

A common MBR-process configuration with high oxygen content in the RAS (Crawford et al. 2006).

Figure 5

A common MBR-process configuration with high oxygen content in the RAS (Crawford et al. 2006).

Figure 6

Configuration of the new treatment process at Henriksdal WWTP.

Figure 6

Configuration of the new treatment process at Henriksdal WWTP.

The aim in the Himmerfjärden project has been to reduce the total recirculation rate of the MBR, while maximizing organic capacity with high MLSS in the bioreactors. In addition, the increased strain on the plant's construction and mechanical equipment, generated by high internal flows, was taken into account. An optimized MBR configuration, combining the benefits of the original idea of step-feed pre-denitrification with MBR technology, was therefore selected. In step-feed pre-denitrification, influent wastewater is fed to the bioreactor in cascades, creating a sludge concentration gradient. The average sludge concentration in the reactor will thus be higher than that in the last cascade. As in the Henriksdal design, RAS will be recycled to an RAS-deox zone before the first cascade. The benefits of step-feed pre-denitrification include: (1) higher MLSS and thus higher organic capacity per m3; (2) additional NO3-recirculation for pre-denitrification is not required since the nitrate produced in the first cascade is denitrified in the next cascade; and, (3) the low MLSS in the last cascade induces reduced RAS-flow (2 × Q) while maintaining a high average sludge concentration in the bioreactor. Step-feed provides significantly reduced recirculation flows, which reduces energy consumption in the plant. Also, a low RAS-flow reduces the oxygen load on the RAS-deox zone, which can thus be smaller. However, in order to achieve 6 mg TN/l in the final effluent, post-denitrification with an external carbon source will also be required at Himmerfjärden.

The choice of a step-feed pre-denitrification process in three cascades fits well with the existing tank layout at Himmerfjärden WWTP (see Figure 7).
Figure 7

Configuration of the new treatment process at Himmerfjärden WWTP.

Figure 7

Configuration of the new treatment process at Himmerfjärden WWTP.

RAS-deox – reject water nitrification: As seen in Figures 6 and 7, both Henriksdal and Himmerfjärden will nitrify reject water from the dewatering of anaerobically digested sludge by mixing it with RAS in a separate zone referred to as RAS-deox. The benefits are threefold: (1) the oxygen added to the RAS for membrane scouring is used for a major objective of wastewater treatment (nitrogen removal); (2), the high ammonia content in the reject water does not need recirculation to be pre-denitrified; and, (3) all NO3 in the RAS is directly recirculated for pre-denitrification.

Phosphorus removal: Both treatment plants will remove phosphorus chemically. Removal of dissolved phosphorus to 0.2 mg P/l by precipitation requires both a sophisticated strategy and control of chemical dosing, as overdosing can cause phosphorus deficiency, disturbing all other biological processes. Moreover, the pilot trials at Henriksdal showed that Fe(II) use at low phosphate concentrations involves the risk of generating phosphate-hydroxide flocs, which might clog the membranes rapidly. Tests are currently in hand to determine whether Fe(III) can be used as an addition to abate this problem.

1

70 g BOD /(pe, d)

REFERENCES

REFERENCES
Brepols
C.
2011
Operating Large Scale Membrane Bioreactors for Municipal Wastewater Treatment
.
1st edn
.
IWA Publishing
,
London
,
UK
.
ISBN: 9781843393054
.
Crawford
G.
Daigger
G.
Erdal
Z.
2006
Enhanced biological phosphorus removal within membrane bioreactors
. In:
Proceedings of the Water Environment Federation
,
WEFTEC 2006
,
1856
1867
.
DOI: http://dx.doi.org/10.2175/193864706783750105.
Itokawa
H.
Thiemig
C.
Pinnekamp
J.
2008
Design and operating experiences of municipal MBRs in Europe
.
Wat. Sci. Tech.
58
(
12
),
2319
2327
.
Judd
S.
2011
The MBR book – Principles and Applications of Membrane Bioreactors for Water and Wastewater Treatment
,
2nd edn
.
Elsevier
,
Oxford
,
UK
.
ISBN: 9780080966823
.