You are on page 1of 18

i An update to this article is included at the end

Science of the Total Environment 727 (2020) 138618

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Review

Constructed wetlands for combined sewer overflow treatment:


A state-of-the-art review
A. Rizzo a, K. Tondera b,c,⁎, T.G. Pálfy c,j, U. Dittmer d, D. Meyer e, C. Schreiber f, N. Zacharias f, J.P. Ruppelt g,
D. Esser h, P. Molle c, S. Troesch i, F. Masi a
a
Iridra Srl, Via La Marmora 51, 50121 Florence, Italy
b
IMT Atlantique Bretagne—Pays de Loire, Department of Energy Systems and Environment, 44307 Nantes, France
c
INRAE, REVERSAAL, F-69625, Villeurbanne, France
d
Institute for Water, Infrastructure and Resources, Department for Urban Water Management, TU Kaiserslautern, Paul-Ehrlich-Straße 14, 67663 Kaiserslautern, Germany
e
Municipal government City of Mayen, Department 3.1 - City Planning and Construction Supervision, Town Hall Rosengasse, D-56727 Mayen, Germany
f
Institute for Hygiene and Public Health, University Hospital Bonn, GeoHealth Centre, Venusberg-Campus 1, 53127 Bonn, Germany
g
Institute of Environmental Engineering (ISA), RWTH Aachen University, 52056 Aachen, Germany
h
SINT, Société d'Ingénierie Nature & Technique, Chef-Lieu, F-73370 La Chapelle du Mont du Chat, France
i
Eco Bird, 3 route du Dôme, 69630 Chaponost, France
j
University of Sopron, Institute of Geomatics and Civil Engineering, H-9400 Sopron, Hungary

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• Combined sewer overflows (CSOs) con-


vey significant pollutant loads to water
bodies.
• Constructed wetlands (CWs) are
Nature-Based Solutions for CSO pollu-
tion control.
• Review gives state-of-the-art of CSO-
CWs based on research and year-long
practice.
• CSO-CWs show effective removal of
conventional, emerging and pathogenic
pollutants.
• CSO-CWs can provide other ecosystem
services (e.g. flood protection,
biodiversity).

a r t i c l e i n f o a b s t r a c t

Article history: Combined sewer overflows (CSOs) are a major source of surface water pollution and degradation. This is partic-
Received 21 November 2019 ularly visible where sewage collection with combined sewer and centralized treatment are well established, such
Received in revised form 10 March 2020 as in Europe and North America: an overwhelming number of surface water bodies are in insufficient status of
Accepted 8 April 2020
ecology, hydrology and physico-chemical parameters. Therefore, several countries have started implementing
Available online 16 April 2020
constructed wetlands (CWs) as mainstream on-spot treatment. This paper summarizes the main design ap-
Editor: Jan Vymazal proaches that can be adopted. We identified eight different schemes for the implementation of CSO-CWs,
based on our international experience and documented by a literature analysis. The performance review includes
Keywords: conventional water quality parameters, as well as pathogen and emergent contaminant removal. Furthermore,
Stormwater management modelling tools for advanced design and for understanding a wide applicability of these green infrastructures
Constructed wetland are presented. This paper also provides a review on other side benefits offered by the adoption of Nature-

⁎ Corresponding author at: IMT Atlantique Bretagne—Pays de Loire, Department of Energy Systems and Environment, 44307 Nantes, France.
E-mail address: katharina.tondera@inrae.fr (K. Tondera).

https://doi.org/10.1016/j.scitotenv.2020.138618
0048-9697/© 2020 Elsevier B.V. All rights reserved.
2 A. Rizzo et al. / Science of the Total Environment 727 (2020) 138618

Treatment wetland Based Solutions for CSO treatment, such as ecosystem services, and the most common issues related to their op-
Combined sewer overflow eration and maintenance. Our analysis has produced a list of key factors for design and operation, all derived from
Nature-Based Solutions full-scale installations in operation up to more than ten years.
Retention soil filter
© 2020 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2. Literature analysis and treatment schemes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1. German approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2. French approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3. Italian approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.4. American approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.5. Combination of CSO tanks and conventional CW solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.6. Aerated systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3. Treatment performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.1. Conventional pollutants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.2. Micropollutants and heavy metals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.3. Microbiological parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
4. Modelling tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4.1. Process-based model experiences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
4.2. Models for CSO-CW design support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
5. Ecosystem services additional to water quality improvement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
6. Design and operational experiences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
6.1. German experience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
6.2. French experience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
6.3. Design recommendation for the reviewed treatment schemes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
7. Discussion and conclusions: gap of knowledge and future research needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Appendix A. Supplementary data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

1. Introduction Tondera et al., 2019; Schreiber et al., 2016). Consequently, water moni-
toring shows a clear increase in microorganism concentrations in natu-
Combined sewers convey both domestic sewage and stormwater ral waters subjected to CSO discharge events (Rechenburg et al., 2006;
runoff in a single pipe, differently from separated sewers, in which sep- Rechenburg and Kistemann, 2009) as well as an increase in
arate pipes convey the two streams. Although several countries are micropollutant loads (Launay et al., 2015). Taking as an example the
pushing for a large implementation of separate sewers, many cities European Union, the current legislation misses to address CSOs directly
are still served by combined systems, e.g. in the European Union coun- in its directives. According to Pistocchi et al. (2019), CSO pollutant loads
tries such as Italy, many regions in Germany and France, Belgium, have been recognised as one of the most relevant loads remained un-
Greece and Poland (Pistocchi et al., 2019). treated according to the Water Framework Directive (2006/7/EC), and
In order reduce the frequency of flooding in urban areas, combined CSOs are considered as one the main reasons of the failure in achieving
sewers were designed for heavy storm events, with the assumed return good status of water bodies at European scale (European Commission,
period usually varying from 2 to 10 years. Since the treatment capacity 2019). Moreover, CSOs are one of the most common causes of losing
of centralized wastewater treatment plants (WWTPs) is limited, waste- bathing water status set by the EU Bathing Water Directive (2006/7/
water flow within the combined sewer is controlled with a series of EC), as pointed out by Al Aukidy and Verlicchi (2017).
overflow structures within the served catchment. The results is that, Another challenging point regarding the CSO environmental pollu-
during heavy rain events, combined sewer overflow (CSO) structures tion control is the large diffusion of combined sewer structures in the
discharge an untreated mixture of domestic sewage and stormwater drained catchment. Many examples can be given to this regard. The
into surface water bodies. “Grand Lyon” sewer system in the Yzeron catchment serves 950,000
Pollutant and hydraulic shock from CSOs are a serious environmen- PE (the 2nd most densely populated area in France) and has about
tal threat for inland and coastal waters (Marsalek, 1998). The term 200 CSO discharge points. A feasibility study on the potential impact
“urban stream syndrome”, for example, describes the general degrada- of CSOs on water quality status of water bodies in Lambro-Seveso-
tion of streams receiving urban runoff (Chocat et al., 1994). Various Olona catchment (Lombardia Region), commissioned by the Po River
studies show that CSO pollutant loads have magnitudes similar to the Basin, identified up to 47 critical CSO discharging points (among the
emissions from WWTPs for many substances on annual basis, or even hundreds present in the territory) distributed over the whole
higher (e.g. Gasperi et al., 2012a; Launay et al., 2015; Schreiber et al., catchment, i.e. across four provinces and 33 different cities. These few
2016). However, in contrast to continuous WWTP effluents, CSOs hap- examples show how control of CSOs is mainly a catchment-scale issue,
pen at maximum a few dozen times per year, and have a much lower which cannot be simply addressed with conventional end-of-pipe
water quality comparable to mechanically treated domestic wastewa- WWTPs.
ter. Discharge events are short, but release large volumes and consider- A full conversion from combined to separated system of an existing
able pollutant loads. This means not only that carbon and nutrient sewer is often unfeasible for cost-benefit reasons. Therefore, the manag-
compounds enter surface waters without any treatment, but also ing of CSOs has been seen as a way to address their environmental im-
micropollutants, heavy metals and pathogens (Tondera et al., 2013; pact. Although solutions such as (1) continuous primary treatment (e.g.
A. Rizzo et al. / Science of the Total Environment 727 (2020) 138618 3

grit removal, sedimentation, enhanced flocculation); (2) increasing the 2. Literature analysis and treatment schemes
use of available storage capacity through integrated sewer manage-
ment; (3) storage methods such as first flush tanks; and (4) the combi- The peer reviewed literature was screened using SCOPUS and
nation of these with disinfection or chemical treatment (EPA, 1995; searching the three keywords which were used to target CW treatment,
Lainé et al., 1998; De Cock et al., 1999; Zukovs and Marsalek, 2004; “constructed wetland”, “treatment wetland”, and “retention soil filter”
Barco et al., 2008; Gasperi et al., 2012b; Fuchs et al., 2014; Tondera (see Section 2.1), and the keyword “combined sewer overflow”. Forty-
et al., 2016) are often implemented, Nature-Based Solutions (NBS) for two articles were found; each article was labelled according to its key
continuous CSO treatment are rapidly gaining momentum and form topic. The details of the literature analysis are reported in Supplemen-
the topic of our review. Constructed wetlands (CWs, referred to also tary online material.
as treatment wetlands - Kadlec and Wallace, 2009) are widely known In order to identify suitable designs, grey literature was also consid-
NBS, and have various designs to treat different sorts of polluted waters. ered. This included the conference proceedings of two major CW con-
According to the path of flow, CWs can be distinguished in surface or ferences: the “IWA International Conference on Wetland Systems for
subsurface-flow CWs, both a state-of-the-art CSO treatment. In case of Water Pollution Control” and the “International Symposium on Wetland
sub-surface flow CWs, the wastewater flows through a porous media, Pollutant Dynamics and Control” (WETPOL). Three studies were identi-
which is composed of sand or gravel with sufficient and stable hydraulic fied accounting for: (i) an aerated wetland of Hawes et al. (2016), (ii)
conductivity. Macrophytes create a root zone of enhanced interaction the US case study of Harbor Brook presented by Monge et al. (2017),
between the roots, the biofilm and the water which flows through the and (iii) the CW treating CSO upstream the centralized WWTP of
system. The media has the physical filtering role that enhances solid re- Carimate reported by Masi et al. (2019).
tention, and provides surface for adsorption and a stable environment The results of the literature analysis are summarised in Fig. 1, while
for biofilm development. In the case of surface flow wetlands (usually the full results are in the Supplementary online materials (Table S1).
referred as free water surface – FWS - systems), water is normally shal- Eight papers (19% of all) regard pilot-scale systems, which are usually
low and the CW is planted with either emergent or free-floating macro- tested to prepare new CW schemes to go full-scale. This recites the
phytes. FWS systems imitate the natural treatment capacity of surface early 2000s. Sixteen papers (35%) regard full-scale systems. This con-
waters such as wetlands, lakes, and ponds. Constructed wetlands for firms the technological readiness of CSO-CWs to reduce CSO impacts
combined sewer overflow treatment (CSO-CWs) were initially on inland waters and the shoreline. Eleven papers (28%) address the
engineered to reduce total suspended solids (TSS) and chemical oxygen conditions faced by CSO-CWs, such as the stochastic variability of influ-
demand (COD) while providing additional retention volume for CSOs. ent loads via modelling studies .
Since then, they have been subsequently optimized to remove a large We categorised eight CW treatment schemes in our review (Fig. 2).
number of other pollutants. Such differences in CSO-CW approaches can be attributed to a general
The first CSO-CWs were built in Germany in the late 1980s, using a fragmentation in how legislation deals with CSO pollution control.
vertical subsurface flow system. Pilot sites in the middle of the 1990s Since a uniform legislation for CSO is generally lacking, both in EU
were then subject to research in the German states of Baden- (European Commission, 2019) and in US (Levy et al., 2014; Tao et al.,
Wuerttemberg and North Rhine-Westphalia (NRW) to evaluate CSO- 2014), varying CSO-CW designs have been proposed in different coun-
CW performance (Grotehusmann et al., 1997; LfU, 1998; Liebeskind, tries, in order to meet local needs in water pollution control and/or ef-
2001). Today, there are 139 CSO-CWsin NRW (www.elwasweb.nrw. fluent water quality targets.
de; accessed on 21st October 2019). Other states and countries have CSOs convey large amount of solid particles, including sands and
been building CSO-CWs as well, from small-scale pilot studies in sediments. Therefore, CSO-CWs usually have preliminary treatments
Portugal (Amaral et al., 2013; Pisoeiro et al., 2016), to a quickly growing (e.g. sand and grease trap). Since sand and other kind of coarse ma-
number of full-scale systems in France (Pálfy et al., 2017a, 2017b, 2017c), terials can lead to clogging, we included a preliminary treatment
Italy (Masi et al., 2017), and the United States (Tao et al., 2014). Full-scale stage in all proposed treatment schemes. A concise description for
CSO-CW technology is established, showing how it is well suited to con- each CSO-CW scheme is provided in the following subchapters.
trol water pollution from CSOs in line with the sewer system. Meyer et al. Technical details of the German, French and Italian approaches are
(2013) revised European schemes and Tao et al. (2014) reported the US well-described in Meyer et al. (2013), the American ones in Tao
experience in review papers, without reporting detailed treatment per- et al. (2014).
formance. Since then, performance studies have been published,
reporting on conventional and emerging pollutants as well as reduction 2.1. German approach
of indicator microorganisms (Christoffels et al., 2014; Pálfy et al., 2017a;
Tondera, 2019; Tondera et al., 2019). CSO-CWs were also targeted by National guidelines such as those of the German Water Association
modelling, both for better understanding the internal removal processes (DWA) stipulate a unique nomenclature. The term corresponding to
(e.g. Pálfy et al., 2016) and to develop state-of-the-art design tools CSO-CWs translates as retention soil filter (RSF). The name stems from
(Meyer and Dittmer, 2015; Pálfy et al., 2017c; Pálfy et al., 2018). Other the late 1980s, when the first systems were built with soil as filter
aspects highlighted by research were the role of CSO-CWs at watershed
scale (e.g. Schreiber et al., 2016; Tondera et al., 2017; Bachmann-
Machnik et al., 2018), and the ecosystem services of multi-purpose
CSO-CWs (Liquete et al., 2016; Masi et al., 2017; Rizzo et al., 2018).
This manuscript gives a comprehensive review of CSO-CW state-of-
the-art, based on recent evidence published by the scientific commu-
nity. Doing so, our review (1) identifies the proposed treatment
schemes through a literature analysis in Chapter 2; (2) discusses the
treatment performance of standard pollutants, micropollutants, and mi-
crobial contamination in Chapter 3; (3) provides an overview of model-
ling studies in Chapter 4; (4) highlights additional ecosystem services
that can be assured by CSO-CWs in Chapter 5; (5) summarizes key de-
sign parameters and potential issues learnt from long-term operation
and literature-based evidence in Chapter 6; and (6) discusses the cur-
rent knowledge gaps and future research needs in Chapter 7. Fig. 1. Pie chart on the results from the literature analysis on CSO-CWs.
4 A. Rizzo et al. / Science of the Total Environment 727 (2020) 138618

Fig. 2. Schemes of CSO-CW treatment on the basis of literature analysis. AEW: aerated wetland. FWS: free water surface flow. FTW: floating treatment wetland. HF: horizontal subsurface
flow. VF: vertical subsurface flow.

material (Brunner, 1995). Due to problems with clogging, the filter ma- RSF and CSO tanks as one treatment unit, allowing the substitution of
terial is nowadays usually sand (0/2 mm), with a steep sieve curve. Co- missing CSO tank volume with the retention volume of the RSF (DWA,
hesive soil is no longer allowed for new systems. The German approach 2019).
remains distinguished as RSF (“Retention Soil Filter”), although it would In terms of technology, a RSF is a vertical-subsurface flow (VF) wet-
be more logical to refer to it as “Retention Sand Filter”, as done in Meyer land with sand filter material and a detention volume on top of the filter
et al. (2013). level. The filter is planted with reed to prevent clogging. The filtrate is
In Germany, most WWTPs in combine sewer systems are designed collected by perforated drain pipes. The filtration rate is limited by a
to treat 2–6 L·s−1·per hectare of connected impermeable area (Aimp). throttle orifice at the outlet structure to qout b 0.05 L·s−1·m−2 (DWA,
Any runoff exceeding this value is stored in CSO tanks, which are not 2019), and many systems operate with lower values. The limitation en-
necessarily situated at the locations of WWTPs, but are mostly distrib- sures a quasi-uniform percolation rate in the entire active pore system.
uted along the drainage system and connected to different WWTP To achieve optimal aeration, the filter is completely drained after each
sub-catchments. Storage volumes are 10–40 m3·ha−1 Aimp, to catch loading event. The material usually used for the filter layer is quartz
1–4 mm of rainfall. In many catchments, rainfall events exceeding sand (granulation 0.063–2.0 mm) containing up to 20% of limestone,
these values cause fifteen to forty overflows annually. Over the last and the filter bed has a layer of broken limestone on top. The limestone
20 years, several hundred RSFs have been built at these discharge points provides calcium buffering of pH to prevent acidification that could be
to protect the receiving waters. The current national guideline considers caused by biochemical processes such as nitrification.
A. Rizzo et al. / Science of the Total Environment 727 (2020) 138618 5

2.2. French approach separation unit, the system implements a floating treatment wetland
(FTW) as first stage, which maintains a vegetated cover while accu-
The French approach treats CSO in a VF wetland without CSO tank or mulating the volume of a CSO to be treated with a variable water
sedimentation basin. Similarly to French domestic wastewater treat- depth. The accumulated CSO is then treated in a second VF stage
ment wetlands (Molle et al., 2005), the objective is to minimize sludge (filled with sand) and a final polishing FWS wetland. The scheme is
management. French regulation assigns new CSO-CWs as treatment based on the full-scale pilot plant of Harbor Brook, built in Syracuse
plants that need to ensure the same outlet quality as the WWTPs. To (New York). The different stages can work both in series and in paral-
maximize the performance, several design specificities were included lel; to the best of our knowledge, the monitoring campaign is still on-
in the first guideline (ANR, 2013). The bottom layer with the drainage going and clear indications on the best configuration have yet to be
pipes is kept saturated limiting reed stress during dry periods and also identified (Monge et al., 2017).
hydraulic short-circuiting at the beginning of events, as well as improv-
ing TSS removal. An aeration pipe is implemented above the saturation 2.5. Combination of CSO tanks and conventional CW solutions
layer to favor oxygen transfer during resting periods. As the German
systems, the French systems also have a detention volume above the fil- We have identified a number of treatment schemes that are addi-
ter surface and the outflow rate is limited by a throttle orifice in the out- tional to those reviewed by Meyer et al. (2013) and Tao et al. (2014).
let structure with a targeted percolation rate of 0.01–0.05 L·s−1·m−2. One is based on the Portuguese trials on pilot-scale horizontal subsur-
Clogging is delayed by two specificities: by the use of coarse sand face flow (HF) systems, which show the capability of HF wetlands to
(d10 N 0.4 mm) and by building two filter cells to allow alternating feed- treat CSOs (Amaral et al., 2013; Pisoeiro et al., 2016). However, a proper
ing and resting periods. Small but frequent events use only one cell, treatment chain must be established, one that allows to counteract CSO
while the other cell rests. In case of events with larger inflow, both variability and let the wetlands work in conditions that are comparable
cells are flooded in order to make use of the full detention volume. to those experienced by CWs at municipal wastewater systems. There-
The resting periods promote good sludge mineralisation. The cells in op- fore, a CSO tank is mandatory. The tank coincides with the first flush
eration are switched on a monthly frequency to uniformly distribute the tank usually proposed to reduce the impact of CSO pollutant loads on
sediments on both. Interestingly, a similar scheme is also proposed for water bodies (e.g. Barco et al., 2008). First flush tanks are used to cap-
German RSFs by the recent German guidelines (DWA, 2019). When am- ture the most polluted fraction of the CSO, which is pumped back
monium reduction is required to be high, zeolite replaces a 0.06 to again in the sewer when the CSO event ends. Different from the conven-
0.18 m thick layer of sand with similar particle size distribution. This is tional use of first flush tank, the schemes here proposed conjugates con-
estimated to assure an outlet concentration of NH+ 4 -N below 5 mg/L. ventional first flush tanks with CWs, treating the captured CSO volume
in situ. This permits to reduce the pollutant loads conveyed to central-
2.3. Italian approach ized WWTPs, which does not have to treat the diluted CSO conveyed
by a conventional use of first flush tanks. For instance, a modelling
The Italian approach developed from the peculiarities of the Gorla study of Masseroni et al. (2018) investigates and proposes a combina-
Maggiore CSO-CW (Meyer et al., 2013; Masi et al., 2017). Local regula- tion of first flush tank and CW similar to those discussed here. We report
tion (regional law of Lombardy on CSO, R.R. 3/2006), at the year of de- on three further CW schemes on the basis of the most relevant subsur-
sign, required the treatment of the first flush volume – defined as 50 face flow systems (Dotro et al., 2017):
m3·ha−1Aimp – and the limitation of flow rates into the water bodies for
flood risk mitigation. - preliminary treatment, CSO tank, primary treatment (e.g., septic
Consequently, designers proposed the following treatment scheme. tanks) and HF wetland;
Firstly, a VF wetland as first stage treats first flush, with analogies to - preliminary treatment, CSO tank, primary treatment and VF wetland
the German and French approaches. Similar to the French systems, the filled with sand;
preliminary treatment is simple (grit, sand, and oil removal), the bed - preliminary treatment, CSO tank, VF wetland with gravel (in accor-
of the actual filter unit is filled with a coarse media (gravel 2–6 mm) dance to the French Systems for wastewater treatment (Molle
and the outflow rate is limited by a throttle valve. The value of the out- et al., 2005))
flow rate was chosen equal to 0.004–0.020 L·s−1·m−2. Secondly, a free
water surface (FWS) wetland polishes the effluent as second stage. The 2.6. Aerated systems
FWS wetland receives the effluent from the VF first stage and the CSO
surpassing the first flush. It is designed with high free banks to also Hawes et al. (2016) report a successful application of intensified
work as an extended retention basin. CWs for CSO treatment. Therefore, the possibility to use aerated systems
cannot be excluded from our treatment schemes. Aerated wetlands are
2.4. American approaches expected to increase performance, robustness, and resilience of the
CSO-CW, in comparison to a conventional CW solution, requiring a
Due to the typically higher available space to implement NBS, as slightly more complex implementation and input of energy for the aer-
compared to Europe, the US approaches were mainly based on exten- ation system. A FWS wetland polishing stage can be included.
sive FWS systems with larger footprints. According to Tao et al.
(2014), a few CSO-CWs have already been implemented in the US. 3. Treatment performance
Those plants are difficult to compare in terms of design characteristics;
indeed, they satisfy and function under different site specific conditions, Treatment performance depends on wetland design and operation.
highlighting the lack of homogeneous guidelines as pointed out by Levy But it is also impacted by the stochastic nature of sewer overflows, in
et al. (2014). However, the experiences of Tao et al. (2014) allow the term of intensity, length, frequency and pollutant levels. Variable inlet
definition of two different American treatment schemes: conditions affect removal and system behaviour. Consequently, the di-
lution level of inlet wastewater, the load arriving on the wetland during
• American approach I: the scheme is highly naturalistic, consisting of a an event or the period between two overflow events are important.
simple preliminary treatment and a subsequent treatment by a FWS The majority of data comes from German RSFs. The State of NRW
wetland. An example is the CSO-CW of Arch/Bar, situated in Elkhart provides a database, and reports 139 sites in operation on combined
(Indiana). sewer systems and 48 on separate sewer systems (www.elwasweb.
• American approach II: after a preliminary treatment with a swirl nrw.de; accessed on 21st October 2019). Removal processes and
6 A. Rizzo et al. / Science of the Total Environment 727 (2020) 138618

efficiencies have been extensively analysed in various research projects, COD is eliminated by 42–47% only. Numerous case studies have permit-
including small-scale lab experiments, filter columns, pilot-scale and in- ted statistical analysis on data from several systems. Results from
tensive monitoring of full-scale plants (Dittmer, 2006; Frechen et al., Tondera, 2019 are summarised in Table 1, regarding 23 RSF systems.
2006; Henrichs et al., 2007; Woźniak et al., 2007; Woźniak, 2008; Further confirmation of CSO-CW capability to control conventional
Waldhoff, 2008; Henrichs et al., 2009; Dittmer and Schmitt, 2011; pollutants comes from the experience with the French approach.
Meyer, 2011; Tondera et al., 2013, 2019; Christoffels et al., 2014; Fournel et al. (2013) and Fournel (2012) report results from nine differ-
Scheurer et al., 2015; Schreiber et al., 2016; Tondera, 2019; Ruppelt ent pilots, each of them with a surface of 20 m2. Pálfy et al. (2017a,
et al., 2018; Ruppelt et al., 2019; Ruppelt et al., 2020). These studies 2017b) discuss results from the first full-scale system at Marcy-
allow a consistent description of key processes in RSFs. CSO-CW tech- L'Étoile as the successor of the Fournel et al. (2013) and Fournel
nology has spread into other countries too, where treatment schemes (2012) pilots, with a construction dating to 2011. COD and TSS removal
have evolved further. We report the key aspects and removal efficien- in the French system is described to be in correlation with the inflow
cies of RSF systems according to their importance, in conjunction with concentration. For COD, a background constant (below 23 mgCOD·L−1)
the removal efficiencies of the full-scale and pilot plants of other CSO- can describe outlet concentration at diluted inflows (below 70
CW schemes. It must be stressed that a direct comparison in terms of mgCOD·L−1). Removal rates are then different for COD between 70 and
treatment performance among different CSO-CWs is out of the scope 250 mgCOD·L−1 and above, with much higher removal in the upper
of this chapter. Indeed, different local regulations and targeted pollut- band (Pálfy et al., 2018). This is because pollution above 250 mgCOD·L−1
ants led to different design approaches, making it difficult to compare mostly consists by particulates that are filtered very efficiently, but re-
them. moval depends on the biological activity too. Namely, dry period length
and evapotranspiration impact subsequent removal, since they can im-
3.1. Conventional pollutants pact the biomass by the depletion of the substrates or drying out (Pálfy
et al., 2017a).
CSO treatment targets (1) solids and pollutants associated to particle At the full-scale site of Marcy L'Étoile, removal performance was es-
transport, (2) the reduction of organic loads, and (3) the reduction of timated in terms of event mean concentration (EMC) and was in agree-
peak concentrations of ammonium nitrogen. Suspended solids (SS) ment to that of RSFs. TSS removal was very high regardless of the
are retained on the filter surface, forming a sludge layer on top and to coarser media (median: 97%). The removal of total COD was at 80%,
a lesser extent within the filter media. Concentrations of SS and partic- with a lower performance for the dissolved fraction (37%) and the par-
ulate COD (CODX) are reduced to a background level. Mineralisation ticulate fraction comparable to that of TSS (93%). NH+ 4 -N removal was
happens between events (inter-event period) when aerobic conditions 85% for events that were equal or smaller than the design load (arriving
favor heterotrophic degradation. The sludge layer is nonetheless perma- volume with a one-year return period). The amendment of a 0.12 m ze-
nent, improves adsorption capacity and hosts the highest biological olite layer enhanced NH+ 4 -N retention compared to purely sand-based
activity. filters. The filter retained around 45% of total phosphorus, but this was
Concentration of active biomass is expected low in RSF, comparing probably due to the filtration of particulate phosphorus, since the
with wetlands used to treat sewage. Due to this, intra-event degrada- PO3−
4 -P retention was about 18% on average with a decreasing trend
tion of solute COD compounds (CODS) is also expected to be limited, over time to almost zero.
and, to some extent, CODS is mostly retained by adsorption. The elimi- The Italian approach was described based on monitoring of the full-
nation rate can be described as constant. In case of extremely low inflow scale plant of Gorla Maggiore (Masi et al., 2017). Influent EMCs were in
concentrations a minimum background concentration remains in the line with literature data and equal to 382 ± 217 and 7.1 ± 1.3 mg/L for
effluent. As for particulate compounds, degradation mainly takes place COD and NH+ 4 -N, respectively. Due to lack of effluent flow meters, re-
inter-event. moval efficiencies were estimated with two different approaches, both
Ammonium elimination has two steps: during the events, ammo- highlighting good removal performance: (1) mean concentration re-
nium is adsorbed almost completely by biofilms within the filter layer. moval of 89.6% (range 74–98%) and 93.5% (range 72–99%) for COD
−1
Effluent concentrations of NH+ 4 -N are reduced to b0.2 mg·L . The and NH+ 4 -N, respectively; (2) annual pollutant load reduction of 68.0%
retained ammonium is nitrified and the sorption capacity regenerates for COD and 94.8% for NH+ 4 -N (based on a theoretical approach using
as nitrate is washed out at the beginning of the next loading event. measured flow rates from sixty-nine CSO events). The estimation of dif-
The activity of nitrifying bacteria is highest immediately after the ferent CSO pollutant loads was assigned to (i) washout from the sewer
draining and re-aeration of the filter layer. Sorption capacity is re- and the drained watershed, (ii) sewage to centralized WWTP, and (iii)
established almost completely within a few days. In cases of prolonged second flushes. Particularly, an estimation of the load in second flushes
loads, the input of ammonium can exceed the adsorption capacity, was carried out using a conservative approach, showing such load to be
which leads to a breakthrough of the infiltrating concentration only due to diluted blackwater. Masi et al. (2017) report that washout
(Dittmer, 2006; Pálfy et al., 2017b). effect is prevalent, but treatment of second flushes is not negligible,
Direct plant uptake has been observed to play a minor role in re-
moval processes. On the other hand, plants are fundamental since
Table 1
they play an active role in supplying of oxygen and root exudates, help-
Treatment performance of RSF from Tondera, 2019.
ing to maintain a healthy microbial life within the wetland even during
drying periods (Brix, 1997). Moreover, stems movement due to wind COD TSS TP
and roots within the media work against clogging, helping to maintain In Out In Out In Out
a proper drainage capacity of the wetland.
Number of samples 226 284 274 284 335 338
RSFs have been extensively studied in terms of treatment perfor- Sampled RSFsa 12 14 13 13 15 16
mance for conventional pollutants, showing stable and high removal ef- Concentrations [mg·L−1] Minimum 15 15 1 1 0.1 0.1
ficiencies. For instance, Uhl and Dittmer, 2005 report removal higher Median 44 16 25 3 0.6 0.2
Mean 69 18 39 6 0.9 0.3
than 84% and 96% for COD and NH+ 4 -N, respectively. Other research pro-
Maximum 918 89 1123 214 16.1 2.7
jects confirm reduction rates of 67–93% for COD based on data covering Removal Median 61% 87% 70%
over two decades and several systems (Liebeskind, 2001; Tondera et al., Mean ± SD 49% ± 44% 38% ± 251% 34% ± 112%
2013). The results reflect those obtained with column tests (Woźniak, a
Values based on samples taken on 23 RSFs designed according to the same standard;
2008). Dittmer (2006) distinguishes between dissolved and particulate values provided by operator treated as if taken on one site; not all samples were corre-
COD: while particulate COD is almost completely retained, dissolved sponding ones, therefore, the number of outflow samples N number of inflow samples.
A. Rizzo et al. / Science of the Total Environment 727 (2020) 138618 7

since second flushes intercepted an additional 12.1 tCOD·year−1 and concentration of diclofenac in the outflow was 65% lower than that in
0.28 tN-NH4·year−1 (14.5% and 43.9% in terms of annual loads) as com- the inflow.
pared to annual pollutant load estimated from CSOs. Another group of micropollutants is reported from the full-scale
Performance information regarding conventional pollutants is also plant of Marcy-L'Etoile (French approach), including seven Polycyclic
available from the other CSO-CW approaches. Tao et al. (2014) report Aromatic Hydrocarbons (PAHs) that are listed as priority substances
that performance data on CSO-CWs in the United States are limited, by the EU Water Framework Directive (2000/60/EC). Pálfy et al.
since plants are often authorized without mandatory monitoring of ef- (2017a) report high removal of three PAHs that had been present in
fluent water quality. However, monitored U.S. systems show good per- concentrations sufficient enough to quantify reduction of benzo(a)
formance in terms of conventional pollutants. The Arch/Bar and the pyrene (80%), phenanthrene (80%), and naphtalene (92%). Effluent con-
Washington CSO-CWs have a FWS wetland as main treatment stage centration of these and other PAHs are summarised in Table 3.
and represent American approach I. The Arch/Bar is reported to have Pálfy et al. (2017a) confirm good removal for the majority of the
mean removal efficiencies of 91% for COD, 92% for TSS, 98% for NH+ 4 -N, monitored metals, too, as shown in Table 4. The authors report on
and 96% for TP. The Washington CSO-CW is known to release twenty-five elements in dissolved and particulate phases. Most metals
TSS b 15 mg·L−1, and BOD5 b 7 mg·L−1. The American approach II were bound to particles in CSO, which is in accordance with Dembélé
(FTW + VF wetland or HF wetland + FWS wetland) from the full- et al. (2009) reporting these to make up 55–85% of the pollution in a
scale pilot plant of Harbor Brook removes BOD5 66%, TSS 56%, NH+ 4 -N comparable catchment. It was concluded that Al, Ti, Cr and Ba are re-
42%, and TP 47% (Monge et al., 2017). moved well during all phases of CSO-CW operation. Mn, Fe, Cu, Zn and
The Portuguese experiences on pilot-scale experiments confirm that Sr are released at the early stages of CSO loads but improve importantly
HF -CW can properly treat conventional pollutants in CSOs. Amaral et al. after long-term filter operation.
(2013) and Pisoeiro et al. (2016) highlight that one day of hydraulic re-
tention time (HRT) is already sufficient to get high removal on COD and 3.3. Microbiological parameters
TSS, as summarised in Table 2. Hawes et al. (2016) also report satisfying
removal in the first aerated systems, with averages of 88% for BOD and Information on pathogen removal performance is widely available
94% for NH+ 4 -N. for RSFs. This includes studies that investigated the reduction of bacteria
and bacteriophages in full-scale systems.
3.2. Micropollutants and heavy metals Waldhoff (2008) is the first study with a comprehensive investiga-
tion of pilot- and full-scale CSO-CWs in terms of reduction rates of
A number of studies investigate the removal of micropollutants in Escherichia coli (E. coli; median reduction of 1.2 log10) and intestinal En-
RSFs, and heavy metal reduction was extensively studied in the full- terococci (I.E., median reduction of 0.9 log10) reduction rates. Ruppelt
scale plant of Marcy-L'Etoile. The results help identifying prevailing re- et al. (2018) report similar reductions on pilot-scale systems; reduction
moval pathways of emerging pollutants. of faecal indicator bacteria ranges between 1 and 2 log10, with filtration,
Micropollutants removal depends on both the chemical characteris- straining processes and adsorption serving as the main removal mech-
tics and operational parameters of the system, and the varying concen- anisms. Somatic coliphages decrease 1.0 to 1.2 log10. Reductions could
trations of CSOs are a common challenge for the evaluation and be further improved by using a series connection of two RSFs with a fil-
comparison for this field of research (Tondera et al., 2013). Aerobic deg- tration layer of 0.75 m (E. coli 2.3 log10, intestinal Enterococci 2.2 log10,
radation of adsorbed micropollutants is the main purification process. somatic coliphages 2.1 log10; median reduction of all events sampled).
Therefore, sufficient oxygen supply in between two events is crucial Other results from full-scale systems confirm these values. Merkel
(Matamoros and Bayona, 2007). Precipitation, sorption, and plant- and Schaule (2010) quantify the performance of four established but
uptake are expected to be main heavy metal removal processes in sparsely-loaded RSFs in NRW after 3.5 to 4 years of operation. In these
CSO-CWs. However, the changes in redox conditions driven by the RSFs, the reduction of E. coli varies between a median of 1.4 to almost
CSO variability could promote remobilization of precipitated metals, 3 log10. Tondera et al. (2019) conclude on long-term performance.
such as iron (Kadlec and Wallace, 2009; Tondera et al., 2018), and it They observe slightly decreasing reduction for indicator bacteria, and
must be properly addressed during the design of CSO-CWs if the quantify removal to be 1.1 log10 for E. coli and 1.3 log10 for I.E. after
heavy metal removal is a key target. Additionally, filtration can be a seven years, compared to 0.8 log10 for E. coli and 0.8 log10 for I.E. after
key process for removal of both metals and organic pollutants, since ten years. In contrast, somatic coliphages reduction evolves in the oppo-
they are often linked to solid particles. Filtration and sorption processes site direction, with 0.6 log10 and 1.0 log10 reduction after seven and ten
can promote important degradation pathways for organic pollutants, years, respectively.
due to very long residence times. Table 3 shows removal efficiencies Schreiber et al. (2016) describe the reduction performance of a RSF
for micropollutants in the French and German systems. Table 4 presents located in the river Swist catchment area (289 km2) in the federal
the heavy metal removal efficiencies. state of NRW of 2 to 4 log10 depending on the microbiological parame-
For RSFs, Scheurer et al. (2015) report a reduction for diclofenac of ter. This performance is in the same range of the removal of microbio-
81% on average, based on sampling inflow and outflow during five logical parameters in conventional WWTPs. Christoffels et al. (2014)
CSO events. Christoffels et al. (2014) used a different evaluation proce- rely on event composites CSO samples and show that this RSF had a
dure, leading to the limit of quantification (LOQ) of diclofenac in only higher reduction capacity than other investigated systems. Indeed, the
68% of water samples (n = 343, 33 events over 14 months), and only median reduction capacity of the RSF studied by Christoffels et al.
9% of the outflow samples (LOQ: 100 ng·L−1). Hence, the maximum (2014) were at least 3 log10 for faecal indicator bacteria, faecal patho-
genic bacteria and somatic coliphages. This median is calculated from
Table 2 all single reduction capacities of every mean concentration, out of
Performance of Portuguese HF pilot in function of different hydraulic retention time 250 L sample volume each, from all events sampled. It is argued that
(HRT), from Pisoeiro et al. (2016).
the better results observed by Christoffels et al. (2014) are due to the
Substance Experiments HRT 1 day HRT N 1 daya lower regulated flow rate (0.015 L·s−1 m−2) in comparison to the
COD Amaral et al. (2013) 82–88% 93–97% higher value requested by national guidelines (0.03 L·s−1 m−2).
Pisoeiro et al. (2016) 63–85% 77–90% Pathogen removal performance is indicated also for CSO-CWs
TSS Amaral et al. (2013) 86–95% 94–99% different to RSFs, confirming the CSO-CW potentiality of bacteria re-
Pisoeiro et al. (2016) 85–96% 88–99% moval in general. For American approach I, the Arch/Bar CSO-CW
a
Duration not further specified in the references. show a reduction of 5 to 6 log10 in terms of E. coli (Tao et al., 2014).
8 A. Rizzo et al. / Science of the Total Environment 727 (2020) 138618

Table 3
Removal efficiencies for micropollutants of German and French approaches reported in literature.

Micropollutant Removal Effluent conc.[μg/L] CSO-CW scheme References

Acenaphtene NA 0.011 French Pálfy et al., 2017a


Benzo(a) anthracene NA 0.028 French Pálfy et al., 2017a
Benzo(a) pyrene 80% 0.002 French Pálfy et al., 2017a
Benzo(b) fluoranthene NA 0.006 French Pálfy et al., 2017a
Benzo(g,h,i)perylene NA 0.006 French Pálfy et al., 2017a
Bisphenol A 64 ± 35% 0.232a RSF Tondera et al., 2019
53 ± 162% 0.071a RSF after 7 years
33 ± 104% 0.077 ± 0.066 RSF 0.75 m sand layer Ruppelt et al., 2020
28 ± 89% 0.103 ± 0.093 RSF 0.5 m sand layer

65 ± 28% 0.065 ± 0.062 Two RSF 0.75 m each in series connecetion


Benzotriazole 66 ± 20% RSF Scheurer et al., 2015
a
26 ± 42% 1.019 RSF after 7 years Tondera et al., 2019
−39 ± 117% 3.580a RSF after 10 years
29 ± 35% 1.203 ± 0.773 RSF 0.75 m sand layer Ruppelt et al., 2020
28 ± 24% 1.354 ± 0.830 RSF 0.5 m sand layer
26 ± 56% 1.028 ± 0.566 Two RSF 0.75 m each in series connection
Carbamazepine 21 ± 49% 0.206a RSF after 7 years Tondera et al., 2019
−4 ± 84% 0.211a RSF after 10 years
−41 ± 74% 0.138 ± 0.043 RSF 0.75 m sand layer Ruppelt et al., 2020
−32 ± 64% 0.132 ± 0.041 RSF 0.5 m sand layer
−69 ± 145 0.151 ± 0.047 Two RSF 0.75 m each in series connection
Chrysene NA 0.010 French Pálfy et al., 2017a
Diatrizoic acid −4 ± 20% RSF Tondera et al., 2013
Diclofenac 67 ± 23% 0.227a RSF after 7 years Tondera et al., 2019
34 ± 47% 0.418a RSF after 10 years
65% RSF Mertens et al., 2012
29 ± 23% 0.302 ± 0.172 RSF 0.75 m sand layer Ruppelt et al., 2020
19 ± 18% 0.339 ± 0.179 RSF 0.5 m sand layer
61 ± 20% 0.159 ± 0.095 Two RSF 0.75 m each in series connection
EDTA 12 ± 51% 1.479a RSF after 7 years Tondera et al., 2019
52 ± 58% 7.538a RSF after 10 years
Fluoranthene NA 0.013 French Pálfy et al., 2017a
Fluorene NA 0.011 French Pálfy et al., 2017a
Ibuprofen 55% RSF Mertens et al., 2012
Indeno (1,2,3-cd)pyrene 0.006 French Pálfy et al., 2017a
Metoprolole 58 ± 23% 0.227a RSF after 7 years Tondera et al., 2019
36 ± 46% 0.247a RSF after 10 years
72 ± 24% 0.079 ± 0.041 RSF 0.75 m sand layer Ruppelt et al., 2020
69 ± 25% 0.090 ± 0.044 RSF 0.5 m sand layer
88 ± 9% 0.038 ± 0.024 Two RSF 0.75 m each in series connecetion
Naphthalene 92% 0.034 French Pálfy et al., 2017a
Phenanthrene 80% 0.010 French Pálfy et al., 2017a
Pyrene NA 0.013 French Pálfy et al., 2017a
Sulfamethoxazole 4 ± 101% 0.117a RSF after 7 years Tondera et al., 2019
−544 ± 928% 0.239a RSF after 10 years
−21 ± 66% 0.366 ± 0.693 RSF 0.75 m sand layer Ruppelt et al., 2020
−5 ± 47% 0.375 ± 0.786 RSF 0.5 m sand layer
−98 ± 162% 0.313 ± 0.436 Two RSF 0.75 m each in series connecetion
TCPP 34 ± 54% 1.511a RSF after 7 years Tondera et al., 2019
−14 ± 109% 0.216a RSF after 10 yearsb

NA: concentrations too close to the quantification limit to calculate removal performance (Pálfy et al., 2017a).
a
Mean value, 5–12 samples.
b
Inflow concentrations after ten years of operation much lower than after seven years (Tondera et al., 2019).

The American approach II shows a mean removal of approx. 0.6 log10 for
Table 4 faecal coliform (Monge et al., 2017). Amaral et al. (2013) and Pisoeiro
Removal efficiencies for heavy metals of French approach reported in literature. et al. (2016) investigate the dependence of Enterococci removal to HRT
Source: Pálfy et al. (2017a). in Portuguese HF wetland pilot plants, showing 1.6 to 2.7 log10 removals
Metals Removal efficiencies [%] with HRT of one day and 4.6 to 5.7 log10 removal with HRT greater than
one day.
Al 39–99%
Ti 46–91%
Cr 76–84% 4. Modelling tools
Mn Up to 36%, but with releasing events
Fe Up to 86%, but with releasing events
Different models have been developed to simulate CW performance
Ni 96–97%
Cu Up to 96%, but with releasing events and to fit different purposes (Meyer et al., 2015). Generally, models can
Zn Up to 79%, but with releasing events be listed into two groups: complex models, used to better understand
Sr Up to 61%, but with releasing events internal functioning (process-based), and generally simpler models,
Ba 61–100% aimed to support wetland design. CSO-CW complex modelling has pro-
Pb 56–65%
vided good results, but process-based models remain too complex and
A. Rizzo et al. / Science of the Total Environment 727 (2020) 138618 9

too time-consuming to be included in the design phases of CSO-CWs. and (ii) the smallest filter area to satisfy effluent flow and concentration
Simulation tools for CSO-CWs have been developed therefore to be thresholds.
fast, easy-to-use, and, in the end, to target practitioners and decision TSS and COD removal are based on empirical kinetic equations. For
makers. The following sections discuss the modelling tools and the re- TSS, these return a nearly constant background value. For COD, Orage
sults of CSO-CW modelling. works with a background value too, but only at low input concentra-
tions. Above a concentration limit, the COD correlation changes to first
order, and again, at even higher concentrations, a higher COD removal
4.1. Process-based model experiences
rate is used as the extension of this three-stage linear approach. More-
over, the tool adjusts COD kinetic parameters based on the inter-event
The HYDRUS modelling toolkit with the CW2D biokinetic model (©
duration, the season and the climate region. Indeed, dry and hot days
PC Progress s.r.o.) is communicated to be the first process-based model
cause COD removal to drop in CSO-CWs, that is assumed to be caused
that has been used for CSO-CW modelling. Its capability to simulate CWs
by bacterial starvation and, likely in extreme cases, die-off due to
with stochastic CSO loads is investigated by several authors with various
draught as well (Pálfy et al., 2016; Dittmer, 2006).
levels of success. Henrichs et al. (2007, 2009,) model ammonium suc-
Removal of NH+ 4 -N is modelled as a two-step process. These steps
cessfully, and also COD removal for single loads by setting adsorption
are adsorption and subsequent nitrification, although the model does
for slowly biodegradable COD. Fournel et al. (2013) study the methods
not account for nitrogen as product. For calculating adsorption, Orage
of fitting hydraulic conditions. Meyer (2011) fits hydraulic conditions,
has a database of two-stage linear adsorption isotherms that approxi-
but also effluent concentrations for single events. The author adjusts
mate Freundlich isotherms of different filter materials. Equilibrium is
manually initial bacterium concentrations to this aim, in order to have
calculated for the whole mass of the process layer. During hydraulic
a vertical simulated bacteria profile that is similar to measured DNA/
short-circuiting, the water-contacted mass is estimated and only the ad-
RNA/ATP concentrations in the experimental columns that accompa-
sorption capacities of this mass is considered. Nitrification, that is lim-
nied the experiment. Additionally, Meyer (2011) implements a particu-
ited in the tool to the regeneration of adsorption capacities, starts as
late deposition and transport sub-model too, and highlights a gap in the
soon as the process layer begins to empty. The rate is the function of
knowledge about the bacteria dynamics in inter-event (dry) periods.
the mean seasonal temperature in a given climate region and the
Pálfy et al. (2016) experiment further to overcome limits of previous
mass in the solid phase. Calibration matches field measurements at
studies. To this aim, Pálfy et al. (2016) report new biokinetic parameters
full-scale, involving consecutive events with closed adsorption and ni-
like (i) hydrolysis rate; (ii) growth rate of bacteria; (iii) COD fractions
trate flush cycles (Pálfy et al., 2017b).
generated in biomass lysis, (iv) yield coefficients and the non-
As a result, Orage is a relatively simple model that allows simulating
biokinetic settings like COD adsorption. With the new settings, a
one year functioning of a CSO-CW (only French and German ap-
quasi-stable biomass is reached by simulating 160 days of inoculation.
proaches) in short simulation time (about 5 min). The graphical inter-
The previous assumptions permit Pálfy et al. (2016) to successfully sim-
face is user-friendly and it requires a maximum of eight input
ulate a series of multiple loads into laboratory CSO columns, meaning
parameters relatively easy to be obtained: (i) selection from climate re-
good fit to measured COD and NH+ 4 -N in the effluent.
gions by name; (ii) entering available land; (iii to v) entering threshold
concentrations on three different pollutants; (vi) selecting if the sewer
4.2. Models for CSO-CW design support is combined or separated (Orage is also suitable to simulate CW perfor-
mance for stormwater treatment); (vii) the maximum allowed outflow
The first design-support modelling tool is the RSF model family rate; and (viii) desired volumetric storage, based on return periods.
(RSF_Disc and RSF_SIM), described by Dittmer (2006) and Meyer and These features allow the use of Orage in real design phases, such as
Dittmer, 2015. RSF_Sim aims to support the RSF design (German ap- the study on CSO pollutant mitigation in Rhône-Alpes (unpublished
proach). It is still used only for scientific purpose, but it could be imple- data).
mented into sewer system models to estimate potential CSO pollutant
removal from RSF positioned overall the sewer catchment. 5. Ecosystem services additional to water quality improvement
Results from France and Germany were merged into a design-
support modelling tool called Orage, which is based on RSF_SIM and Usually, CSO-CWs are designed with the single purpose to be techni-
permits the design optimization of both the German and the French cal treatment systems. In these cases, despite their “natural” appear-
CSO-CW systems. The tool is based on results from column, pilot, and ance, they are not specifically designed to provide other benefits.
full-scale CSO CWs as well as simulation studies (Dittmer et al., 2005; Taking biodiversity as an example, the choice of plants (monocultures
Dittmer, 2006; Frechen et al., 2006; Heinrichs et al. 2007, 2009; of Phragmites australis) is often motivated only by technical concerns
Woźniak et al., 2007; Waldhoff, 2008; Woźniak, 2008; Dittmer and (e.g. limiting the risk of clogging) and practical reasons (suppression
Schmitt, 2011; Fournel, 2012; Felmeden, 2013; Fournel et al., 2013; of other planted species in past trials).
Pálfy et al., 2016; Pálfy et al., 2017a). Orage optimizes dimensions and However, CSO-CWs are part of the larger family of NBS, which are, as
materials of CSO CWs site-specifically, based on inflow CSO time series referred by the WWAP (2018), “inspired and supported by nature and
and a low number of input parameters. Optimization is based on a use, or mimic natural processes to contribute to the improved manage-
core model (Pálfy et al., 2018) that is handled by an optimization algo- ment of water” and “usually offer multiple water-related benefits and
rithm (iterative shell). Simulations are based on inflow data and desired often help address water quantity, quality and risks simultaneously”.
wetland parameters (Pálfy et al., 2017c). The core model simulates CSO- Therefore, CSO-CWs should be designed, wherever possible, to provide
CW hydraulics and removal performance in terms of TSS, COD, and multiple benefits.
NH+ 4 -N. Despite a simplified process description, Orage remains a dy- The most representative CSO-CW in terms of multipurpose design is
namic and deterministic model. Indeed, flow and transport processes an Italian case study, the Water Park of Gorla Maggiore (Liquete et al.,
are simplified and represented as three continuously stirred tank reac- 2016; Masi et al., 2017). The CSO-CW design prioritized effective treat-
tors by filter cell, plus a common one representing their common stor- ment, changing an abandoned poplar plantation in proximity of the
age basin (seven tanks altogether). The simulation of outflow rates is Olona River into a water park accessible for recreation. Moreover, the
based on a constant, which represents the deterministic effect of the FWS wetland was designed to also be a detention basin for flood mitiga-
outflow orifice in the real system hydraulic. In order to allow easy han- tion and a site of high biodiversity. To assess ecosystem services, this
dling, Orage has a design-optimization algorithm that follows pre- water park is compared with grey infrastructure (i.e. a first flush tank
defined logic to look for a filter with (i) the simplest possible material plus occasionally a dry detention basin) through a multi-criteria
10 A. Rizzo et al. / Science of the Total Environment 727 (2020) 138618

analysis (MCA) by Liquete et al. (2016) and is hydraulically modelled by infrastructure (Liquete et al., 2016). Similar ecological benefits are ex-
Rizzo et al. (2018). pected also from other CSO-CW approaches including a FWS stage, i.e.
The modelling study of Rizzo et al. (2018) investigates the flood mit- for the American approaches, Tao et al. (2014) report a high ecological
igation effect of the park. It models the hydraulic functioning of both the value of the CSO-CW installed in Arch/Bar, where 30 different species
VF and FWS wetlands. The model simulate different CSO events: of native aquatic and transitional vegetation were put in place.
(i) from single to up to five consecutive CSO peaks (i.e. different increas- The Water Park of Gorla Maggiore has also become a recreational
ing and decreasing phases within the same CSO event); (ii) the design park (Fig. 3) with restored riparian tree vegetation, green open space,
CSO hydrograph with return time of ten years; and (iii) a complete walking and cycling paths and supporting services (e.g. picnic table, toi-
time series that covers seventy CSO peaks and corresponding dry pe- lets, bar). The park is maintained by a voluntary association (http://
riods between March and August 2014. Dynamic behaviour and overall www.calimali.org/). The MCA of Liquete et al. (2016) estimates the
attenuation effect are comprehensively analysed. The simulations show sub-criterion “Improve people recreation and health” based on visitor
peak flow reductions from 52.7% to 95% and a retention volume of about numbers, uses and frequency of visits from a successful mail survey.
8800 m3. Interestingly, also the VF first stage contributes well by itself to The results confirm the validity of the multi-purpose approach of the
flood reduction, even if the available storage volume on top of the VF CSO-CW, since the score for the recreation sub-criterion results were
bed is less in comparison to French and German CSO-CW systems. The 85% for the green infrastructure and 40% for grey infrastructure. Indeed,
mean of CSO flow peaks can be reduced by the VF stage as much as by the grey infrastructure is marked less attractive due to the lack of biodi-
95.4%, increasing the hydrograph duration by a factor of twenty-one, versity and related educational facilities.
and storing 92% of the volume. The stored volume is important (12%)
also for the CSO event with a return time of ten years. The VF stage at- 6. Design and operational experiences
tenuation capacity gets less relevant only in the case of consecutive
CSO peaks, where CSO-CW overall attenuation mainly relies on the This chapter presents the design indications and operational experi-
FWS second stage. As main evidence of the simulations, the Water ences from the two most detailed sources of practical information avail-
Park significantly contributes to restore the hydrological response of able up to now: (i) the experience gained by the long-term German use
the catchment with Gorla Maggiore town from a post-development of RSFs, and correlated guidelines; (ii) the detailed monitoring of the
(high peak and short duration) back to a pre-development status (low French System at the Marcy L'Étoile full-scale plant, including the infor-
peak and high duration), in accordance with urban hydrology concepts mation gained from the modelling studies used to define French guide-
(e.g. Fletcher et al., 2013). Flood mitigation benefits can be also consid- lines. Many of the reported German and French experiences should be
ered, even if not targeted to, for single VF stages of German and French considered valuable for CSO-CWs in general and, therefore, also useful
approaches, in which accumulation depth above the surface can be up for the design of other CSO-CW schemes proposed in this review. The
to 2 m (Meyer et al., 2013; DWA, 2019). The capacity to attenuate CSO chapter ends with design recommendation for the different schemes,
peak flow was also highlighted for extensive FWS systems proposed also based on a deep understanding of German and French CSO-CW
in the US (Tao et al., 2014). schemes.
In terms of ecological benefits, the FWS second stage was designed
to enhance biodiversity. Its bottom topography is diversified to favor 6.1. German experience
different autochthone macrophytes, both emergent (Typha angustifolia,
Lythrum salicaria, Mentha aquatic, Iris pseudacorus, Lysimachia vulgaris) Today, RSFs are standardly applied in Germany. Over two hundred
and floating (Nymphaea alba, Nuphar lutea, Ranunculus aquatilis, full-scale systems have been implemented during the last twenty-five
Hydrocharis morsus-ranae L., Ceratophyllum demersum). The MCA of years, and a lot of experience has been gained in design and operation.
Liquete et al. (2016) quantifies the biodiversity contribution with the Sand is nowadays used as a reliable filter media instead of the cohesive
sub-criterion “wildlife support”, which compares the CSO-CW Italian soil materials in the 1990s, which are prone to clogging. The accumula-
scheme used in the Water Park of Gorla (VF + FWS), defined as a tion of sediments and dead biomass from the Phragmites coverage cre-
green infrastructure, to a first flush tank plus a dry detention basin ates a dynamic layer on the filter surface. These secondary layers
with grass bottom, defined as grey infrastructure. The presence of contain mostly organic material and contribute to the overall removal
open water is advantageous in terms of biodiversity, leading to a score by adding to the sorption capacity. This is a process of maturation rather
for the sub-criterion “wildlife support” of about 85% for the green infra- than ageing of the filter. An aerobic environment within the bed is nec-
structure, while the same criterion is equal to 40% for the grey essary for a successful long-term functioning. Since filter feeding is

Fig. 3. Pictures from the Water Park of Gorla Maggiore (Italy). Courtesy: Giacomo Galimberti.
A. Rizzo et al. / Science of the Total Environment 727 (2020) 138618 11

linked to mostly short wet weather events, sufficient oxygen is provided during the research period reported by Pálfy et al. (2017b) was 85% or
both under submerged and dry conditions. more, even during intensive storms.
RSFs have returning aerobic conditions and this promotes nitrifica- For a regular CSO event, flows allow rapid filter saturation, usually
tion, but inhibits denitrification and thus total nitrogen elimination. within less than half an hour after its start. Hydraulic short-circuits
Therefore, the need of a permanently submerged zone within the filter should remain insignificant in terms of total operation time; instead,
has been investigated in various stormwater management studies. the filter saturates progressively from the bottom and allows a ponding
However, filters have various shortcomings when it comes to denitrifi- depth of up to 2.3 m height aboveground. The wastewater passes
cation, not just the lack of a permanently saturated bottom. For exam- through the filter within typically around 5 to 8 h which favors pollutant
ple, Pálfy et al. (2017b) investigate filters similar to RSFs with degradation.
saturated bottom and find no indication of the process. We can conclude For small events or low inlet flow, the primary filter cell in a CSO-CW
that denitrification needed addressing the following, non-favorable fea- does not saturate and all wastewater infiltrates close to the CSO inlet
tures of CSO-CWs: (1) usually, there is no permanently saturated bot- structure. Under these conditions, water passes rapidly through the fil-
tom in the filters; (2) organic carbon is required to drive ter and builds a hydraulic short-circuit that causes low removal perfor-
denitrification, but its concentration is always low in the water of the mance. To limit short-circuiting, the depth of the saturated layer at the
saturated bottom (if there is one); (3) organic carbon-containing bottom is a design parameter of importance, because stored water di-
water arrives with each new CSO load, but contact times with nitrate lutes the percolating water upon drainage. Pálfy et al. (2017b) show
are too short and nitrate is flushed out whether there is a permanently that a larger filter surface is becoming active as the filter is getting closer
saturated zone or not; and (4) the stochasticity of CSOs arriving into the to saturation; above 0.45 m of saturation from the drainage pipes, the
filter mean anoxic conditions are unstable. short-circuiting flows have little impact on the effluent quality.
The German experience shows that denitrification in RSFs is possible In terms of treatment capability, both dissolved COD and NH+ 4 -N re-
with a special filter setup. Saturating the bottom was put under trial at moval performance decrease once the volume that the filter supposed
several full-scale plants, mainly to store water for the reed (Dittmer to be able to treat (design volume) passes the outlet. Extreme events
et al., 2016). In the long run, this led to a degradation of the filter mate- might lead to an important pollutant loss, but still with decreased flux.
rial and odour development, which could then be reduced installing Because inflow, concentration and return period impact performance,
aeration pipes directly above the saturation zone (as in the French de- target conditions need to be defined in the design phase to properly de-
sign). Other potential negative impacts were the iron precipitates in sign the CSO-CW size.
the filter outlet structure, the degradation of limestone (if present), The design-support model Orage (Pálfy et al., 2018, 2017c), de-
and the corrosion due to hydrogen sulphide generation. scribed in chapter 4, provides special design values and indications for
An additional design parameter is the infiltration speed through the the French approach and for CSO-CWs in general. Pálfy et al. (2017c)
filter bed. Different studies report better bacteria removal when limiting show that treatment of any CSO might be satisfied by different designs.
the effluent rate to 0.015 L·s−1 m−2 (e.g. Merkel and Schaule, 2010). They show also why both under- and over-scaled filters are problem-
Hence, the State guideline of NRW for planning, construction and oper- atic. Under-scaled filters: (i) lack treatment potential for the adsorption
ation of RSFs (MKULNV, 2015) recommends an effluent rate limited to capacities of NH+ 4 -N; (ii) can be easily saturated by larger events; and
0.01 L·s−1 m−2 if the removal of pathogens is targeted; (iii) their annual hydraulic load might be higher than considered opti-
0.02 L·s−1 m−2 for NH+ 4 -N and 0.03 L·s
−1
m−2 for TSS, COD, BOD, mal by the present practice, probably leading to clogging issues (espe-
heavy metals and phosphorus. cially if fed with raw wastewater). Large filters can also be
One reason why public water authorities invest in RSFs is the avail- problematic, because the orifice that limits outflow is proportional to
ability of clear guidelines. According to the first guidelines, a maximum the filter area. This means large filters suffer lasting or even continuous
hydraulic load of 40–60 m3·m-2·year−1 was recommended (Uhl and hydraulic short-circuiting that similarly to an under-scaled filter will
Dittmer, 2005). Based on the evaluation of more than fifty systems de- trigger elevated NH+ 4 -N, but also COD concentrations in the effluent
signed to these standards (Grotehusmann et al., 2017), this design ap- (Pálfy et al., 2017a). Similar issues can be present for all the CSO-CWs
proach was revised from hydraulic to pollutant loading rates, namely using VF wetland plus throttle orifice systems, as confirmed by short-
fine solids. The tolerable annual load of fine solids (TSSfine, defined as circuiting in oversized RSF systems observed in Germany. Generally
TSS b 0.63 μm) is estimated today through long-term sewer simulations speaking, a favorable range of net filter areas of potential solutions can
and should not exceed 7 kg·m−2·year−1 (DWA, 2019). Moreover, new be considered, between over- and under-scaled filters. Simulations con-
design guidelines try to fix the main problem encountered during the firm this range to include filters where the net filter area is significantly
years of operation of RSFs, i.e. oversizing of the surface area, which can smaller than the area needed for a storage basin, if zeolite is mixed to
result in short-circuiting and an under loading in case of small events. the filter media. Simulations in Orage highlight also the importance to
have a good knowledge on sewer response to rain events. For that, it
is recommended to make flow and rain measurements at the overflow
6.2. French experience structure for at least ten significant events. These can be used to simu-
late the sewer with rain data set of at least one year on a 6–15 min
The French approach is represented by pilot-scale research followed time step. Thanks to that, a series of one year (minimum) overflow
by research of one full-scale system with advanced online instrumenta- data set will be used to design the filter. For quality aspects, five over-
tion, the continuous development of the specialized design-support flow events have to be monitored to better analyse variability of over-
modelling tool Orage (Pálfy et al., 2017c; Pálfy et al., 2018) and at flow concentrations on COD, TSS and ammonia (ANR, 2013).
least seven full-scale sites that are currently in various stages of design On the basis of monitoring data from the full-scale system and of
and implementation. The single operating full-scale plant is at Marcy- modelling analysis, the following rules were fixed to ensure the normal
L'Etoile, and was commissioned to the operator after the investigations. functioning of the French system over time, and can be relevant for CSO-
The system at Marcy L'Étoile was intensively monitored (Pálfy et al., CW design in general:
2017a, 2017b), and provided important insights on key parameters for a
proper design of CSO-CWs. The hydraulic buffer capacity is important as - Outlet throttle has to deliver an outflow between 0.01 and
it attenuates well the hydraulic impact on the minor stream receiving 0.05 L·s−1·m−2
the effluents. The outlet orifice can be regulated to ensure a flow be- - Maximum ponding depth at 2.5 m above the filter's surface and
tween 0.008 and 0.016 L·s−1·m−2 in the wetland, and was fixed to be a N 2 m ponding frequency below two per year.
≤0.05 L·s−1·m−2 in the French design guideline. Hydraulic attenuation - Cumulated hydraulic load below 100 m·yr−1.
12 A. Rizzo et al. / Science of the Total Environment 727 (2020) 138618

- Saturated bottom layer of 0.3 m minimum to limit the impact of hy- top. Saturation time was concluded to be dependent on the (1) inflow
draulic preferential flows. rate; (2) the area of the filter; and (3) the orifice limiting the outflow, im-
plying that filling could be speeded up by using two orifices instead of
one. If a top orifice with a proper diameter is established near the vertical
6.3. Design recommendation for the reviewed treatment schemes level of the filter surface, the drainage time should be just slightly de-
layed and filling could happen much quicker even at low inflows. This
The design of treatment stages for event-driven pollutant loads is a theory is to be confirmed via modelling or at full-scale yet.
challenging task. Quality of stormwater runoff varies over a wide Another design parameter requiring future investigation is the max-
range between sites, between events at a given site and within the imum annual hydraulic load. Pálfy et al. (2017b) evidence that the use
course of each event. Variability is even higher in combined sewers of zeolite or natural materials with high cation exchange capacity
due to the varying mixing ratio with wastewater and the remobilization (i.e., high NH+4 -N sorption capacity) permits to build a new filter cate-
of sewer sediments (e.g. Suarez and Puertas, 2005). Any sampling of gory called compact filters. Decreasing the filter area could push the an-
CSO discharge only reflects a small section of a randomly fluctuating sig- nual hydraulic load higher. Modelling simulations of Pálfy et al. (2017c)
nal. Long-term average concentrations and loads are commonly un- show that up to 250 m·yr−1 could be tested instead of presently
known, since CSO structures are usually scattered within the 40–80 m·yr−1, minding premature clogging is the risk to be overcome
catchment and difficult to monitor. Input concentrations of CSO-CWs at high loads. Studying solid retention techniques at the overflow
are subject to high uncertainty and design approaches have to account point, the mineralisation rate of organic deposits on the filters and filter
for this uncertainty. Therefore, robustness against varying flow and pol- alternation strategy are necessary to test such high hydraulic loading
lutant loads is more important than accuracy in estimating effluent con- rate. Compact filters, on the other hand, might need to have a large
centrations. This complexity and site specific needs (e.g. different target free area next to the filter to store the CSO volume to be treated, since
pollutants) lead to complex design procedures. Therefore, the values re- less area for compact filters also means less available volume on top of
ported in Table 2 must be considered as only indicative; indeed, this re- the bed (maximum allowed water storage height equal to 2.5 m). The
view remains a non-critical enumeration and comparison of available volume to be treated could be implemented next to the filter, for in-
schemes and provides no exact design rules. More design details are stance in grassed areas; however, selecting of the right plants (grass,
available in the technical literature (Meyer et al., 2013; Tao et al., reed, sedge, wet gardens) suitable to be flooded is an open field, and it
2014; ANR, 2013; MKULNV, 2015; DWA, 2019), in literature discussing needs to be studied to promote a good integration into the urban
design-support modelling (Pálfy et al., 2018; 2017c) and related environment.
software. As seen in chapter 6, there is still uncertainty on the usefulness of
saturated bottom layer in VF filters for CSO treatment. The French expe-
7. Discussion and conclusions: gap of knowledge and future research rience confirms the positive effects of a saturated bottom layer in terms
needs of water reservoir for plants during dry periods and of avoiding hydrau-
lic shortcuts for small CSO events. On the other hand, German experi-
This work has identified eight CSO-CW configurations to effectively ence doubts its effectiveness in terms of denitrification. If the
treat CSOs. For the German approach, a high number of full-scale appli- denitrification is a design target for CSO-CWs, an alternative option is
cations have been built and data from them are available. The number is the use of a natural second stage, which could be able to provide the
less for the other approaches, but the systems are recent and were mon- needed carbon source from the natural environment itself (e.g. decayed
itored with the newest technologies in certain cases (e.g. Marcy plant biomass (Hang et al., 2016) or root exudates (Zhai et al., 2013)).
L'Étoile). Treatment performance is promising for all treatment For instance, Tanner and Kadlec (2013) suggest that FWS wetlands
schemes; a critical comparison of the configurations is neither possible can successfully denitrify agricultural runoff under time-variable influ-
nor necessary at this stage. Firstly, different CSO-CW approaches have ent water flow patterns, which have comparable temporal flow patterns
risen from different targets. Both European (European Commission, to CSO-CW releases.
2019) and US (Levy et al., 2014; Tao et al., 2014) cases have shown a Regarding system optimization, the greatest challenge remains the
fragmented permitting and law framework regarding CSO. Therefore, collection of data for accurate simulation, such as using the Orage
single states have decided to address CSO issue on their own, trying to model. Collecting long-term CSO flow and concentration series of
fit with general water quality objectives; this led to different objectives. good quality would be crucial for design-support modelling and CSO-
For instance, the goal for using RSFs in Germany is to reduce surface CWs in general. For example, data gaps will be interpreted by the tool
water pollution in order to reach a good quality of water bodies as re- Orage as regeneration periods that restore NH+ 4 -N adsorption capacity.
quired by the European WFD, thereby not fixing strict effluent water The same gaps, where too long, will degenerate COD removal perfor-
quality standards and aiming mostly to remove TSS, COD and ammonia, mance as compared to reality. Using other than a full inflow
whereas TN removal was not specifically targeted and the systems have pollutograph leads, therefore, to incomplete optimization of filter area
not been optimized for denitrification. The French approach differs sig- or even material. Sub-optimal wetland design might provide adequate
nificantly, since the effluent water quality standards for CSO-CWs are treatment, but there are cost implications (land area, construction)
those of centralized WWTPs (COD 125 mg/L, BOD b25 mg/L, TSS that can be huge for one system and could add up due to the number
b35 mg/L). Comparing CSO-CWs designed to meet different water qual- of overflow points. Concerning the reliability of design-support models
ity targets can result in misleading conclusions. Secondly, CSO-CW re- themselves, apart from data, Orage is in use for design-support. But as
search needs more results in order to further optimize design schemes other models, it should be involved in further research to gain feedback.
with a lower number of installations and to properly compare the differ- Intensive monitoring of a model-designed full-scale system could pro-
ent CSO-CW configurations. Even if monitoring of CSOs remains a chal- vide feedback for how to build and calibrate the model themselves so
lenge, as pointed out by Tao et al. (2014), we hope that future studies that it provides the best for practitioners.
are going to estimate the treatment performance with EMC and mass The in-situ management of pollutant loads can be considered as the
balance approaches rather than simply based on concentrations. most cost-effective solution for CSO pollution control, if compared to the
Intensive monitoring of full-scale CSO-CWs is an option to investigate development of a fully separated sewer starting from an existing com-
potential optimization needs. For example, Pálfy et al. (2017a, 2017b) re- bined one. A relevant example is given by the study for solving the
port results from online monitoring, and relies on scientific as well as CSO pollution issue of the Yzeron catchment, a west peri-urban area of
data science methods to obtain them. Findings show how hydraulic the Greater Lyon Metropolitan Area. A feasibility study was developed
short-circuiting diminishes as the filter pores saturate from bottom to after the successful installation of the CSO-CW at Marcy-L'Étoile (Pálfy
A. Rizzo et al. / Science of the Total Environment 727 (2020) 138618 13

et al., 2017a) to test different options (Canto, 2018). The price compar- combined use, treating both CSO and WWTP effluent in one RSF. The
ison shows that separating sewage from stormwater for 14 km would study was conducted at a RSF pilot plant under real conditions and
cost 41 million Euro, while separating sewage from stormwater for using artificial CSO. They could show that when applying a dry period
only 5 km and implementing CSO-CWs on the catchment would cost of 18 h (minimum) between CSO and WWTP feeding, the dual use of
23 million Euro. The Greater Lyon example is in line with experiences RSF was most effective.
in the United States reviewed by Tao et al. (2015), which evidences con- Another attractive added value is the use of CSO-CWs as polishing
struction cost savings if using CSO-CWs instead of separating sewers step for micropollutants and emerging contaminants (Brunsch et al.,
and enlarging the WWTP (Akron: $0.7 million for CSO-CW, $5.1 million 2018; Brunsch et al., 2020). Indeed, trapping persistent organic pollut-
for grey option) or physical/chemical treatment (Washington: CSO-CW ants in an ecosystem characterised by a high biodiversity can offer sev-
$40 million, $26.4 million for grey solution). Considering the manage- eral pathways of degradation, with the long retention time and
ment of CSO pollutant loads at the discharging point, a cost comparison adsorbed organic molecules in a complex organic matrix, and therefore
between grey infrastructure (e.g. first flush tank) and CSO-CW was de- CWs could represent one of the most suitable alternatives for dealing
veloped by Liquete et al. (2016), who report a negligible difference in with the currently quite uncontrolled release of these substances into
terms of both construction and maintenance costs. Liquete et al. the environment. Although the effectiveness of CSO-CWs in treatment
(2016) also highlight that the added value of CSO-CWs in comparison of emerging pollutants has already been reported, as reviewed in this
to grey infrastructure lies in the additional ecosystem services provided work, this is surely a hot topic to be further investigated in the future.
by the wetland system, in particular biodiversity support and social ben- Finally, a greater inclusion of CSO-CWs into urban and peri-urban
efits. Further evidences are needed to confirm this statement also for spaces are expected and desired. NBS for urban environment are gaining
other CSO-CW schemes. Particularly, the advantages of using CSO- momentum, especially after the release of the UN WATER Report
CWs optimized only for CSO treatment (i.e. with less ecosystem ser- (WWAP, 2018). Increasing attention is given to NBS for sustainable
vices) in comparison to conventional first flush tank needs to be inves- stormwater management, not only by engineers, but particularly by ar-
tigated, considering also the advantage of CSO-CWs to avoid CSO chitects and landscapers. NBS have the capability to be multipurpose
volume to be treated by WWTP at the end of rainfall events, as it is and easily implementable into urban green spaces and green-blue infra-
the case for first flush storage tanks (Barco et al., 2008). Tondera et al. structure. Following these principles, urban design concepts like Sus-
(2017) simulated 21 scenarios for reduction of bacteria from CSOs in a tainable Drainage Systems (SuDS – Woods-Ballard et al., 2015), Water
catchment in Germany. Although the study shows that combinations Sensitive Urban Design (WSUD – Wong and Brown, 2009), Low Impact
of different treatment techniques are necessary to reduce a maximum Development (LID – Dietz, 2007) or Sponge Cities (Li et al., 2017) are
amount of bacteria, the most cost-effective scenarios always include gaining increasing attention. All these concepts, however, are only fo-
CSO-CWs. cused on stormwater generating urban runoff, while similar negative
Authors expect a great diffusion of CSO-CWs also outside of Germany. impacts are generated also by stormwater mixed with wastewater con-
One key element for a vast diffusion is the presence of successful imple- veyed by combined sewers and discharged as CSOs. Similar to urban
mentation of full-scale CSO-CWs, which could work as reference for stormwater runoff, not only the water quality of receiving water bodies
other CSO-CW systems. Italy, and in particular the Lombardia Region, is is threatened by CSOs; flood reduction is also required (Rizzo et al.,
an example. The success of the Gorla Maggiore Water Park (see chapter 2018); an urban green space with a stream receiving a CSO can compro-
5), in which the Italian approach was tested (Masi et al., 2017), pushed mise the use by the citizens of the green space itself (Liquete et al.,
the Lombardy Region to invest in a number of similar installations to fur- 2016). Therefore, CSO- CWs should be included into the portfolio of
ther investigate the use of NBS for CSOs. Four further full-scale CSO-CWs NBS for sustainable stormwater management in future SuDS, LID, and
have been realized. Two are very significant in terms of sizes, since they WSUD manuals and guidelines. Currently, there are limited examples
treat the CSOs upstream two centralized WWTPs (Carimate, 70,000 PE, for the implementation of CSO treatment facilities in these concepts,
total CSO-CW area 13,000 m2, presented in Masi et al. (2019); Merone, which also lies in a limited collaboration between the expert groups of
120,000 PE, total CSO-CW area 4800 m2, in which the main treatment constructed wetlands and urban drainage. Another reason might be mi-
stage is an aerated wetland). All the systems are designed to be multipur- crobial and pollutant contamination that is higher compared to separate
pose, having a free access for visitors and being designed with an open- stormwater outlets, pointing towards public health and possible liability
water stage for biodiversity purposes. The exploitation of multiple eco- concerns in municipal administration. Nonetheless, our reviews shows
system services could be, therefore, another important driver to push CWs are ready to be used for CSO pollution control at full-scale as they
the application of CSO-CWs; this highlight the importance to consider provide solutions for the ever-increasing pressure on receiving urban
the implementation of more extensive FWS systems (e.g. Italian or waters. Research with societal discourse on the use of CSO-CWs in
American approaches), when space is available, increasing ecosystem urban spaces is necessary to foster the diffusion of this technology in
services, especially biodiversity support. Lombardia Region also expects urban spaces and parks.
to increase the use of CSO-CWs with a new regulation (no 6/2019),
which clarifies the amount of CSO to be treated on-site (up to Declaration of competing interest
750–1000 L⋅d-1 per PE) and that NBS can be used for CSOs. Lombardia Re-
gion is also releasing a new guideline to support the design of both tech- The authors declare that they have no known competing financial
nological (mainly primary treatments) and NBS. The experience of the interests or personal relationships that could have appeared to influ-
Lombardia Region was here reported as an example of policies aimed ence the work reported in this paper.
to diffuse CSO-CWs, which encounter also the agreement of literature.
For instance, the need of guidelines was advocated as an important Appendix A. Supplementary data
way to promote the diffusion of CSO-CWs systems by Levy et al. (2014).
Although the main locations of CSO-CWs are the dispersed CSO Supplementary data to this article can be found online at https://doi.
discharging points within the sewer catchment, they can also be used org/10.1016/j.scitotenv.2020.138618.
to treat the CSO upstream big centralized WWTPs. In this case, the at-
tractiveness of CSO-CWs could theoretically be increased by using
References
them in dry periods to polish effluent from centralized WWTPs
(Griffin, 2004; Brunsch et al., 2020). However, the implication on treat-
Al Aukidy, M., Verlicchi, P., 2017. Contributions of combined sewer overflows and treated
ment performance and long-term effects for the CW itself needs to be effluents to the bacterial load released into a coastal area. Sci. Total Environ. 607,
part of future research. Brunsch et al. (2020) use the concept of the 483–496.
14 A. Rizzo et al. / Science of the Total Environment 727 (2020) 138618

Amaral, R., Ferreira, F., Galvao, A., Matos, J.S., 2013. Constructed wetlands for combined Languedoc PhD thesis. http://www.biu-montpellier.fr/florabium/jsp/nnt.jsp?nnt=
sewer overflow treatment in a Mediterranean country, Portugal. Water Sci. Technol. 2012MON20111;, Accessed date: 14 August 2019 (in French, English abstract).
67 (12), 2739–2745. Fournel, J., Forquet, N., Molle, P., Grasmick, A., 2013. Modelling constructed wetlands with
ANR, 2013. Système extensifs pour la Gestion et le Traitement des Eaux Urbaines de variably saturated vertical subsurface-flow for urban stormwater treatment. Ecol.
temps de Pluie. Extensive Systems for the Management and Treatment of Urban Eng. 55, 1–8.
Stormwaters. Guideline. https://epnac.inrae.fr/wp-content/uploads/2012/08/ Frechen, F.B., Schier, W., Felmeden, J., 2006. The plant-covered retention soil filter (RSF):
ANRsegteup-GUIDE-Vfin-novembre2013-corrige.pdf, Accessed date: 10 March 2020 the mechanical and biological combined sewer overflow (CSO) treatment. Plant. Eng.
(in French). Life Sci. 6 (1), 74–79.
Bachmann-Machnik, A., Meyer, D., Waldhoff, A., Fuchs, S., Dittmer, U., 2018. Integrating Fuchs, S., Mayer, I., Haller, B., Roth, H., 2014. Lamella settlers for storm water treatment –
retention soil filters into urban hydrologic models–relevant processes and important performance and design recommendations. Water Sci. Technol. 69 (2), 278–285.
parameters. J. Hydrol. 559, 442–453. Gasperi, J., Zgheib, S., Cladière, M., Rocher, V., Moilleron, R., Chebbo, G., 2012a. Priority pol-
Barco, J., Papiri, S., Stenstrom, M.K., 2008. First flush in a combined sewer system. lutants in urban stormwater: part 2 – case of combined sewers. Water Res. 46,
Chemosphere 71 (5), 827–833. 6693–6703.
Brix, H., 1997. Do Macrophytes play a role in constructed treatment wetlands? Water Sci- Gasperi, J., Laborie, B., Rocher, V., 2012b. Treatment of combined sewer overflows by
ence Technology 35 (5), 11–17. ballasted flocculation: removal study of a large broad spectrum of pollutants. Chem.
Brunner, P.G., 1995. Regenwasserbehandlung in Bodenfilterbecken. Treatment of rainwa- Eng. J. 211–212, 293–301. https://doi.org/10.1016/j.cej.2012.09.025.
ter in soil filter basins. Wasserwirtschaft 85 (3), 134–138 (in German). Griffin, P., 2004. Ten years experience of treating all flows from combined sewerage sys-
Brunsch, A.F., ter Laak, T.L., Christoffels, E., Rijnaarts, H.H.M., Langenhoff, A.A.M., 2018. Re- tems using package plant and constructed wetland combinations. Water Sci. Technol.
tention soil filter as post-treatment step to remove micropollutants from sewage 48 (11−12), 93–99.
treatment plant effluent. Sci. Total Environ. 637, 1098–1107. https://doi.org/ Grotehusmann, D., Kerpen, T., Ruß, H.-J., 1997. Mischwasserbehandlung in
10.1016/j.scitotenv.2018.05.063. Bodenfilteranlagen. Treatment of combined sewage in soil filter plants.
Brunsch, A.F., Zubieta Florez, P., Langenhoff, A.A.M., ter Laak, T.L., Rijnaarts, H.H.M., 2020. Umweltqualitäten und Wirtschaften? Was wurde erreicht? Wo geht es hin?, 38th
Retention soil filters for the treatment of sewage treatment plant effluent and com- Essener Tagung für Wasser- und Abfallwirtschaft, Aachen, Germany. 158.
bined sewer overflow. Sci. Total Environ. 699, 134426. https://doi.org/10.1016/j. Gewässerschutz, Wasser, Abwasser, pp. 10/1–10/14 (in German).
scitotenv.2019.134426. Grotehusmann, D., Lambert, B., Fuchs, S., Uhl, M., Leutnant, D., 2017.
Canto, A., 2018. Les filtres plantés de roseaux pour le traitement des eaux pluviales se Erhebungsuntersuchung zur Optimierung der Retentionsbodenfilter in NRW. In:
développent. Reed beds for the treatment of stormwater runoff are developing. Ministry of Environment, Nature Conservation, Agriculture and Consumer
Environnement & Technique 380 (in French). Protection of the German Federal State of North Rhine-Westphalia (Ed.), Investiga-
Chocat, B., Cathelain, M., Mares, A., Mounchel, J.M., 1994. La pollution due aux rejets tion to Optimize Retention Soil Filter in NRW Final Report. Düsseldorf. https://
urbains de temps de pluie: impacts sur les milieux récepteurs. Pollution Caused by www.lanuv.nrw.de/fileadmin/tx_mmkresearchprojects/Abschlussbericht_RBF_NRW.
Urban Stormwater: Impacts on Receiving Waters. vols. 1-2. La Houille Blanche, pdf, Accessed date: 6 August 2019 (in German).
pp. 97–105 (in French). Hang, Q., Wang, H., Chu, Z., Ye, B., Li, C., Hou, Z., 2016. Application of plant carbon source
Christoffels, E., Mertens, F.M., Kistemann, T., Schreiber, C., 2014. Retention of pharmaceu- for denitrification by constructed wetland and bioreactor: review of recent develop-
tical residues and microorganisms at the Altendorf retention soil filter. Water Sci. ment. Environ. Sci. Pollut. Res. 23 (9), 8260–8274.
Technol. 70 (9), 1503–1509. Hawes, P., Wallace, S., Cooper, D., 2016. Combined sewer overflow (CSO) treatment by
De Cock, W., Blom, P., Vaes, G., Berlamont, J., 1999. The feasibility of flocculation in a stor- aerated wetland. 15th IWA International Conference on Wetland Systems for Water
age sedimentation basin. Wat. Sci. Tech. 39 (2), 75–83. Pollution Control, ICWS 2016. 4–9 September (Gdansk, Poland).
Dembélé, A., Becouze, C., Bertrand-Krajewski, J.L., Cren-Olivé, C., Barillon, B., Coquery, M., Henrichs, M., Langergraber, G., Uhl, M., 2007. Modelling of organic matter degradation in
2009. Quantification des pollutants prioritaires dans les rejets urbains de temps de constructed wetlands for treatment of combined sewer overflow. Sci. Total Environ.
pluie. Quantification of priority pollutants in urban stormwater 104 (4), 60–76. 380 (1), 196–209.
https://doi.org/10.1051/tsm/200904060 TSM. (in French, English abstract). Henrichs, M., Welker, A., Uhl, M., 2009. Modelling of biofilters for ammonium reduction in
Dietz, M.E., 2007. Low impact development practices: a review of current research and combined sewer overflow. Water Sci. Technol. 60 (3), 825–831.
recommendations for future directions. Water Air Soil Pollut. 186 (1–4), 351–363. Kadlec, R.H., Wallace, S.D., 2009. Treatment Wetlands. 2nd ed. CRC Press, Boca Raton, FL,
Dittmer, U., 2006. Prozesse des Rückhaltes und Umsatzes von Kohlenstoff- und USA.
Stickstoffverbindungen in Retentionsbodenfiltern zur Mischwasserbehandlung. Re- Lainé, S., Poujol, T., Dufay, S., Baron, J., Robert, P., 1998. Treatment of stormwater to bath-
tention and transformation processes of carbon and nitrogen compounds in retention ing water quality by dissolved air flotation, filtration and ultraviolet disinfection. Wat.
soil filters for CSO treatment. PhD thesis. Schriftenreihe des Fachgebiets Sci. Tech. 38 (10), 99–105.
Siedlungswasserwirtschaft der Technischen Universität Kaiserslautern Volume 23. Launay, M.A., Dittmer, U., Steinmetz, H., 2015. Organic micropollutants discharged by com-
Kaiserslautern Technical University. http://nbn-resolving.de/urn/resolver.pl?urn: bined sewer overflows – characterisation of pollutant sources and stormwater-related
nbn:de:hbz:386-kluedo-20509, Accessed date: 4 August 2016 (in German, English processes. Water Res. 104, 82–92. https://doi.org/10.1016/j.watres.2016.07.068.
abstract). Levy, Z.F., Smardon, R.C., Bays, J.S., Meyer, D., 2014. A point source of a different color:
Dittmer, U., Schmitt, T.G., 2011. Purification processes in biofilter systems for CSO treat- identifying a gap in United States regulatory policy for “green” CSO treatment using
ment. Proceedings of the 12th International Conference on Urban Drainage, Porto constructed wetlands. Sustainability 6 (5), 2392–2412.
Alegre, Brazil https://www.researchgate.net/profile/Ulrich_Dittmer/publication/ LfU, 1998. Bodenfilter zur Regenwasserbehandlung im Misch- und Trennsystem. Soil fil-
268002715_Purification_Processes_in_Biofilter_Systems_for_CSO_Treatment/links/ ters for stormwater treatment in combined and separate sewer system. Handbuch
54c7669f0cf22d626a365adf/Purification-Processes-in-Biofilter-Systems-for-CSO- Wasser 4 Band 10. Landesanstalt für Umweltschutz Baden-Württemberg, Karlsruhe
Treatment.pdf, Accessed date: 14 August 2019. (95 p. + annexes (in German).
Dittmer, U., Meyer, D., Langergraber, G., 2005. Simulation of a subsurface vertical flow Li, H., Ding, L., Ren, M., Li, C., Wang, H., 2017. Sponge city construction in China: a survey
constructed wetland for CSO treatment. Water Sci. Technol. 51 (9), 225–232. of the challenges and opportunities. Water 9 (9), 594.
Dittmer, U., Meyer, D., Tondera, K., Lambert, B., Fuchs, S., 2016. Treatment of CSO in reten- Liebeskind, M., 2001. Untersuchung der Leistungsfähigkeit von Bodenfiltern in NRW. In-
tion soil filters - lessons learned from 25 years of research and practice. Novatech. 9th vestigation on Efficiency of Retention Soil Filter in NRW. Final Report. Forschungsinstitut
International Conference: Planning and Technologies for Sustainable Urban Water für Umwelttechnik und Umweltanalytik, Wasserlaboratorien Roetgen GmbH (un-
Management. Lyon (28/06-01/07/2016). published (in German).
Dotro, G., Langergraber, G., Molle, P., Nivala, J., Puigagut, J., Stein, O., Von Sperling, M., Liquete, C., Udias, A., Conte, G., Grizzetti, B., Masi, F., 2016. Integrated valuation of a
2017. Treatment Wetlands. IWA Publishing. nature-based solution for water pollution control. Highlighting hidden benefits. Eco-
DWA, 2019. Retentionsbodenfilteranlagen. Retention soil filter sites. DWA-A 178, German system Services 22, 392–401.
Water Association. ISBN: 978-3-88721-826-3. Marsalek, J., 1998. Challenges in urban drainage. In: Environment NAS (Ed.),
EPA, 1995. Combined Sewer Overflows - Guidance for Long-term Control Plan. US Envi- Hydroinformatic Tools for Planning, Design, Operation and Rehabilitation of Sewer
ronmental Protection Agency, Office of Water. Systems. Kluwer Academic Publishers, London, pp. 1–23.
European Commission, 2019. Evaluation of the Urban Waste Water Treatment Directive. Masi, F., Bresciani, R., Rizzo, A., Conte, G., 2017. Constructed wetlands for combined sewer
SWD, Brussels. overflow treatment: ecosystem services at Gorla Maggiore, Italy. Ecol. Eng. 98,
European Union, 2006. DIRECTIVE 2006/7/EC OF THE EUROPEAN PARLIAMENT AND OF 427–438.
THE COUNCIL of 15 February 2006 Concerning the Management of Bathing Water Masi, F., Bresciani, R., Martinuzzi, N., Bernasconi, M., Rizzo, A., 2019. Treatment of com-
Quality and Repealing Directive 76/160/EEC. bined sewer overflow upstream centralized treatment plants with nature-based solu-
Felmeden, J., 2013. Phosphorrückhalt in der Mischwasserbehandlung durch tions: the constructed wetland system of Carimate WWTP, Italy. 8th International
Retentionsbodenfilter-Anlagen. Retention of the Phosphorous in Combined Sewer WETPOL Symposium (Aarhus, Denmark, June 17–21).
Overflows Using Retention Soil Filters. Department of Sanitary and Environmental Masseroni, D., Ercolani, G., Chiaradia, E.A., Maglionico, M., Toscano, A., Gandolfi, C.,
Engineering, University of Kassel PhD thesis. publication No. 33. http://www. Bischetti, G.B., 2018. Exploring the performances of a new integrated approach of
unikassel.de/upress/online/frei/978-3-89958-610-7.volltext.frei.pdf, Accessed date: grey, green and blue infrastructures for combined sewer overflows remediation in
4 August 2016 (in German). high-density urban areas. Journal of Agricultural Engineering 49 (4), 233–241.
Fletcher, T.D., Andrieu, H., Hamel, P., 2013. Understanding, management and modelling of Matamoros, V., Bayona, J.M., 2007. Behavior of emerging pollutants in constructed wet-
urban hydrology and its consequences for receiving waters: a state of the art. Adv. lands. In: Barceló, Damià, Petrovic, Mira (Eds.), Emerging Contaminants from Indus-
Water Resour. 51, 261–279. trial and Municipal Waste. Springer, pp. 199–217.
Fournel, J., 2012. Systemes extensifs de gestion et de traitement des eaux urbaines de Merkel, W., Schaule, G., 2010. Bewertung der Leistungsfähigkeit von vier
temps de pluie. Extensive Systems for the Management and Treatment of Urban Run- Retentionsbodenfiltern hinsichtlich der Rückhaltung hygienisch relevanter
offs in Rainy Weather. Universite de Montpellier II, Sciences et Techniques du Mikroorganismen. Evaluating the Retention Capacity of Four Retention Soil Filters
A. Rizzo et al. / Science of the Total Environment 727 (2020) 138618 15

in Terms of Hygienically Relevant Microorganisms Final report. https://www.lanuv. Ruppelt, J.P., Pinnekamp, J., Tondera, K., 2020. Elimination of micropollutants in four test-
nrw.de/fileadmin/tx_mmkresearchprojects/Abschlussbericht_Retentionsbodenfilter_ scale constructed wetlands treating combined sewer overflow: influence of filtration
unterzeichnet.pdf, Accessed date: 3 October 2019 (in German). layer height and feeding regime. Water Res. 169, 1–10. https://doi.org/10.1016/j.
Mertens, F.M., Christoffels, E., Schreiber, C., Kistemann, T., 2012. Rückhalt von Arzneimitteln watres.2019.115214.
und Mikroorganismen am Beispiel des Retentionsbodenfilters Altendorf. Retention of Scheurer, M., Heß, S., Lüddeke, F., Sacher, F., Güde, H., Löffler, H., Gallert, C., 2015. Removal
pharmaceuticals and microorganisms at the example of the retention soil filter Altendorf. of micropollutants, facultative pathogenic and antibiotic resistant bacteria in a full-
59. KA-Korrespondenz Abwasser, Abfall, pp. 1137–1143 (in German). scale retention soil filter receiving combined sewer overflow. Environ Sci: Proc Im-
Meyer, D., 2011. Modellierung und Simulation von Retentionsbodenfiltern zur pacts 17, 186–196. https://doi.org/10.1039/c4em00494a.
weitergehenden Mischwasserbehandlung. Modelling and simulation of constructed Schreiber, C., Rechenburg, A., Koch, C., Christoffels, E., Claßen, T., Willkomm, M., Mertens,
wetlands for combined sewer overflow treatment. PhD dissertation. 31. Kaiserslau- F.M., Brunsch, A.F., Herbst, S., Rind, E., Kistemann, T., 2016. Two decades of system-
tern Technical University. Schriftenreihe des Fachgebiets Siedlungswasserwirtschaft based hygienic–microbiological research in Swist river catchment (Germany). Envi-
der Technischen Universität Kaiserslautern. https://kluedo.ub.unikl.de/files/2843/ ron. Earth Sci. 75 (21). https://doi.org/10.1007/s12665-016-6100-9.
Dissertation_Daniel_Meyer_online.pdf, Accessed date: 4 August 2016 (in German). Suarez, J., Puertas, J., 2005. Determination of COD, BOD, and suspended solids loads during
Meyer, D., Dittmer, U., 2015. RSF_Sim–a simulation tool to support the design of con- combined sewer overflow (CSO) events in some combined catchments in Spain. Ecol.
structed wetlands for combined sewer overflow treatment. Ecol. Eng. 80, 198–204. Eng. 24 (3), 199–217.
Meyer, D., Molle, P., Esser, D., Troesch, S., Masi, F., Dittmer, U., 2013. Constructed wetlands Tanner, C.C., Kadlec, R.H., 2013. Influence of hydrological regime on wetland attenuation
for combined sewer overflow treatment—comparison of German, French and Italian of diffuse agricultural nitrate losses. Ecol. Eng. 56, 79–88.
approaches. Water 5 (1), 1–12. https://doi.org/10.3390/w5010001. Tao, W., Bays, J.S., Meyer, D., Smardon, R.C., Levy, Z.F., 2014. Constructed wetlands for
Meyer, D., Chazarenc, F., Claveau-Mallet, D., Dittmer, U., Forquet, N., Molle, P., Morvannou, treatment of combined sewer overflow in the US: a review of design challenges
A., Pálfy, T., Petitjean, A., Rizzo, A., Campà, R.S., 2015. Modelling constructed wetlands: and application status. Water 6 (11), 3362–3385.
scopes and aims–a comparative review. Ecol. Eng. 80, 205–213. Tondera, K., 2019. Evaluating the performance of constructed wetlands for the treatment
MKULNV, 2015. Retentionsbodenfilter. Handbuch für Planung, Bau und Betrieb. Retention of combined sewer overflows. Ecol. Eng. 137, 53–59. https://doi.org/10.1016/j.
Soil Filter-Planning, Construction, and Operation Manual. Ministry of Environment, ecoleng.2017.10.009.
Agriculture, Conservation and Consumer Protection of the State of North Rhine- Tondera, K., Koenen, S., Pinnekamp, J., 2013. Survey monitoring results on the reduction of
Westphalia https://www.umwelt.nrw.de/fileadmin/redaktion/Broschueren/ micropollutants, bacteria, bacteriophages and TSS in retention soil filters. Water Sci.
retentionbodenfilter_handbuch.pdf, Accessed date: 6 August 2019 (in German). Technol. 68 (5), 1004–1012. https://doi.org/10.2166/wst.2013.340.
Molle, P., Liénard, A., Boutin, C., Merlin, G., Iwema, A., 2005. How to treat raw sewage with Tondera, K., Klaer, K., Koch, C., Ahmed Hamza, I., Pinnekamp, J., 2016. Reducing pathogens
constructed wetlands: an overview of the French systems. Water Sci. Technol. 51 (9), in combined sewer overflows using performic acid. Int. J. Hyg. Environ. Health 219
11–21. (7), 700–708. https://doi.org/10.1016/j.ijheh.2016.04.009.
Monge, Z., Bays, J., Pries, J., Rhoads, T., 2017. Treating combined sewer overflows naturally Tondera, K., Klaer, K., Roder, S., Brueckner, I., Pinnekamp, J., 2017. Improving the microbi-
with engineered wetlands: Harbor brook CSO 018, Onondaga County, New Y. “7th In- ological quality of the Ruhr River near Essen: comparing costs and effects for the re-
ternational WETPOL Symposium” Big Sky, Montana, USA, August 21–25. duction of Escherichia coli and intestinal enterococci. Water Sci. Technol. 75 (11),
Pálfy, T.G., Molle, P., Langergraber, G., Troesch, S., Gourdon, R., Meyer, D., 2016. Simulation 2659–2668. https://doi.org/10.2166/wst.2017.141.
of constructed wetlands treating combined sewer overflow using HYDRUS/CW2D. Tondera, K., Blecken, G.T., Chazarenc, F., Tanner, C.C., 2018. Ecotechnologies for the treat-
Ecol. Eng. 87, 340–347. ment of variable stormwater and wastewater flows. SpringerBriefs in Water Science
Pálfy, T.G., Gerodolle, M., Gourdon, R., Meyer, D., Troesch, S., Molle, P., 2017a. Performance & Technology. Springer, Cham, Switzerland ISBN 978-3319700120.
assessment of a vertical flow constructed wetland treating unsettled combined sewer Tondera, K., Ruppelt, J., Pinnekamp, J., Kistemann, T., Schreiber, C., 2019. Reduction of
overflow. Water Sci. Technol. 75 (11), 2586–2597. micropollutants and bacteria in a constructed wetland for combined sewer overflow
Pálfy, T.G., Gourdon, R., Meyer, D., Troesch, S., Olivier, L., Molle, P., 2017b. Filling hydraulics treatment after 7 and 10 years of operation. Sci. Total Environ. 651, 917–927. https://
and nitrogen dynamics in constructed wetlands treating combined sewer overflows. doi.org/10.1016/j.scitotenv.2018.09.174.
Ecol. Eng. 101, 137–144. Uhl, M., Dittmer, U., 2005. Constructed wetlands for CSO treatment: an overview of prac-
Pálfy, T.G., Gourdon, R., Meyer, D., Troesch, S., Molle, P., 2017c. Model-based optimization tice and research in Germany. Water Sci. Technol. 51 (9), 23–30.
of constructed wetlands treating combined sewer overflow. Ecol. Eng. 101, 261–267. Waldhoff, A., 2008. Sanitation of combined sewer in retention soil filters (RSFs) [in Ger-
https://doi.org/10.1016/j.ecoleng.2017.01.020. man]. Hygienisierung von Mischwasser in Retentionsbodenfiltern (RBF). Department
Pálfy, T.G., Meyer, D., Troesch, S., Gourdon, R., Olivier, L., Molle, P., 2018. A single-output of Sanitary and Environmental Engineering, University of Kassel PhD thesis. publica-
model for the dynamic design of constructed wetlands treating combined sewer tion No. 30. http://www.uni-kassel.de/upress/online/frei/978-3-89958-606-0.
overflow. Environ. Model. Softw. 102, 49–72. volltext.frei.pdf, Accessed date: 14 August 2019 (in German).
Pisoeiro, J., Galvão, A., Ferreira, F., Matos, J., 2016. Potential for CSO treatment with hori- Wong, T.H.F., Brown, R.R., 2009. The water sensitive city: principles for practice. Water Sci.
zontal flow constructed wetlands: influence of hydraulic load, plant presence and Technol. 60 (3), 673–682.
loading frequency. Environ. Sci. Pollut. Res. 23 (20), 20591–20599. Woods-Ballard, B., Wilson, S., Udale-Clarke, H., Illman, S., Scott, T., Ashley, R., Kellagher, R.,
Pistocchi, A., Dorati, C., Grizzetti, B., Udias, A., Vigiak, O., Zanni, M., 2019. Water Quality in 2015. The SuDS Manual, C753. CIRIA, London, UK (ISBN 978-0-86017-760-9).
Europe: Effects of the Urban Wastewater Treatment Directive. A Retrospective and Woźniak, R., 2008. Ermittlung von Belastungsgrenzen an Bodensubstraten zur
Scenario Analysis of Dir. 91/271/EEC, EUR 30003 EN. Publications Office of the weitergehenden Michwasserbehandlung in Retentionsbodenfiltern. Determination
European Union, Luxembourg 978-92-76-11263-1 https://doi.org/10.2760/303163. of load limits on filter substrates of constructed wetlands for combined sewer over-
Rechenburg, A., Kistemann, T., 2009. Sewage effluent as a source of Campylobacter sp. in a flow treatment. PhD thesis at the. Band 24. Department of Architecture/Regional De-
surface water catchment. Int J Environ Heal R 19, 239–249. https://doi.org/10.1080/ velopment and Environmental Planning/Civil Engineering at TU Kaiserslautern,
09603120802460376. Schriftenreihe. https://kluedo.ub.uni-kl.de/files/1948/Diss_Wozniak_SIWAWI.pdf,
Rechenburg, A., Koch, C., Claßen, T., Kistemann, T., 2006. Impact of sewage treatment Accessed date: 4 August 2016 (in German).
plants and combined sewer overflow basins on the microbiological quality of surface Woźniak, R., Dittmer, U., Welker, A., 2007. Interaction of oxygen concentration and reten-
water. Wat Sci Tech 54, 95–99. tion of pollutants in vertical flow constructed wetlands for CSO treatment. Water Sci.
Rizzo, A., Bresciani, R., Masi, F., Boano, F., Revelli, R., Ridolfi, L., 2018. Flood reduction as an Technol. 56 (3), 31–38.
ecosystem service of constructed wetlands for combined sewer overflow. J. Hydrol. WWAP, 2018. The United Nations World Water Development Report 2018: Nature-Based
560, 150–159. Solutions for Water. (United Nations World Water Assessment Programme)/UN-
Ruppelt, J.P., Tondera, K., Schreiber, C., Kistemann, T., Pinnekamp, J., 2018. Reduction of Water. UNESCO, Paris.
bacteria and somatic coliphages in constructed wetlands for the treatment of com- Zhai, X., Piwpuan, N., Arias, C.A., Headley, T., Brix, H., 2013. Can root exudates from emer-
bined sewer overflow (retention soil filters). Int. J. Hyg. Environ. Health 221 (4), gent wetland plants fuel denitrification in subsurface flow constructed wetland sys-
727–733. https://doi.org/10.1016/j.ijheh.2018.04.011. tems? Ecol. Eng. 61, 555–563.
Ruppelt, J.P., Tondera, K., Van der Weken, L., Vorenhout, M., Pinnekamp, J., 2019. Redox Zukovs, G., Marsalek, J., 2004. Planning and design of combined sewer overflow treat-
potential as a method to evaluate the performance of retention soil filters treating ment. Water Qual. Res. J. 39 (4), 439–448.
combined sewer overflows. Sci. Total Environ. 650, 1628–1639. https://doi.org/
10.1016/j.scitotenv.2018.09.043.
Update
Science of the Total Environment
Volume 772, Issue , 10 June 2021, Page

DOI: https://doi.org/10.1016/j.scitotenv.2021.146174
Science of the Total Environment 772 (2021) 146174

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Corrigendum

Corrigendum to “Constructed wetlands for combined sewer


overflow treatment: A state-of-the-art review”
[Sci. Total Environ. 727 (2020) 138618]

A. Rizzo a, K. Tondera b,c,⁎, T.G. Pálfy c,j, U. Dittmer d, D. Meyer e, C. Schreiber f, N. Zacharias f, J.P. Ruppelt g,
D. Esser h, P. Molle c, S. Troesch i, F. Masi a
a
Iridra Srl, Via La Marmora 51, 50121 Florence, Italy
b
IMT Atlantique Bretagne—Pays de Loire, Department of Energy Systems and Environment, 44307 Nantes, France
c
INRAE, REVERSAAL, 69625 Villeurbanne, France
d
Institute for Water, Infrastructure and Resources, Department for Urban Water Management, TU Kaiserslautern, Paul-Ehrlich-Straße 14, 67663 Kaiserslautern, Germany
e
Municipal government City of Mayen, Department 3.1 - City Planning and Construction Supervision, Town Hall Rosengasse. D-56727, Mayen, Germany
f
Institute for Hygiene and Public Health, University Hospital Bonn, GeoHealth Centre, Venusberg-Campus 1, 53127 Bonn, Germany
g
Institute of Environmental Engineering (ISA), RWTH Aachen University, 52056 Aachen, Germany
h
SINT, Société d'Ingénierie Nature & Technique, Chef-Lieu, F-73370 La Chapelle du Mont du Chat, France
i
Eco Bird, 3 route du Dôme, 69630 Chaponost, France
j
University of Sopron, Institute of Geomatics and Civil Engineering, H-9400 Sopron, Hungary

The authors regret that the printed version of the above article review remains a non-critical enumeration and comparison of available
contained a number of errors. The correct and final version follows. schemes and provides no exact design rules.” The authors would like to
The authors would like to apologise for any inconvenience caused. apologise for any inconvenience caused.
In Section 6.3, a table is missing. The sentence “Therefore, the values
reported in Table 2 must be considered as only indicative; indeed, this Declaration of Competing Interest
review remains a non-critical enumeration and comparison of available
schemes and provides no exact design rules” is misleading, it should link The authors declare that they have no known competing financial
to the missing Table 5, which can be found below: “Therefore, the values interests or personal relationships that could have appeared to influ-
reported in Table 5 must be considered as only indicative; indeed, this ence the work reported in this paper.

DOI of original article: https://doi.org/10.1016/j.scitotenv.2020.138618.


⁎ Corresponding author at: IMT Atlantique Bretagne—Pays de Loire, Department of Energy Systems and Environment, 44307 Nantes, France.
E-mail address: katharina.tondera@inrae.fr (K. Tondera).

https://doi.org/10.1016/j.scitotenv.2021.146174
0048-9697/© 2021 Elsevier B.V. All rights reserved.
A. Rizzo, K. Tondera, T.G. Pálfy et al. Science of the Total Environment 772 (2021) 146174

Table 5
Indicative design parameters of CSO-CW treatment schemes and their treatment chain in order.

Treatment scheme Indicative design parameters Consulted references

A) German approach
- sedimentation tank The filtration speed is regulated (e.g. outflow orifice): 0.036–0.180 m·h−1 MKULNV (2015)
- VF wetland (0.01–0.05 L·s−1·m−2 or 10−5 - 5·10−5 m·s−1) DWA (2019)
B) French approach Meyer et al. (2013)
- VF wetland with two filter cells The filtration speed is regulated (e.g. outflow orifice): 0.036–0.180 m·h−1
(0.01–0.05 L·s−1·m−2 or 10−5 - 5·10−5 m·s−1)
C) Italian approach Masi et al. (2016)
- VF wetland The filtration speed (VF) is regulated by (e.g. outflow orifice): 0.036–0.360 m·h−1
- FWS wetland (0.01–0.10 L·s−1·m−2 or 10−5 - 10−4 m·s−1)
The HRT of the FWS is 6–12 h at least
D) American approach I Tao et al. (2014)
- FWS wetland The HRT is 72–120 h at least
E) American approach II Tao et al. (2014)
- FTW wetland The HRT of the FTW is 12–24 h at least Pisoeiro et al. (2016)
- HF or VF wetland The HRT of the HF is 24–48 h at least Dotro et al. (2017)
- FWS wetland The HLR of the VF is based on design values for domestic wastewater and never tops 80–120 L·m−2·d−1
Hydraulic retention time (FWS) is 6–12 h at least
F) Portuguese approach Pisoeiro et al. (2016)
- first flush tank The HRT of the HF is 24–48 h at least Dotro et al. (2017)
- HF or VF wetland with sand The HLR of the VF is based on design values for domestic wastewater and never tops 80–120 L·m−2·d−1
G) Meyer et al. (2013)
- first flush tank The filtration speed is regulated (e.g. outflow orifice): 0.036–0.360 m·h−1
- VF wetland with gravel (0.01–0.10 L·s−1·m−2 or 10−5 - 10−4 m·s−1)
H) Masi et al. (2016) and
- aerated wetland The minimum emptying time of the aerated wetland fed by the CSO is 12–24 h unpublished
- FWS wetland (optional) The HRT (FWS) is 6–12 h at least data on aerated systems

You might also like