Content uploaded by Jeroen Langeveld
Author content
All content in this area was uploaded by Jeroen Langeveld on Jan 16, 2014
Content may be subject to copyright.
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
Langeveld et al. 1
Storm water sewers: pollution levels and removal rates of three
full scale storm water treatment facilities in Arnhem
J.G. Langeveld1,3, H.J. Liefting1 and H. Velthorst2
1 Royal Haskoning, Barbarossastraat 35, P.O. Box 151, 6500 AD, Nijmegen, the Netherlands
2 Department water, municipality of Arnhem, P.O. Box 9200, NL-6800 HA, Arnhem, the Netherlands
3. Delft university of Technology. Department of Sanitary Engineering, Faculty of Civil
Engineering and Geosciences, Delft University of Technology, P.O. Box 5048, NL-2600 GA,
Delft, the Netherlands
*Corresponding author, e-mail j.langeveld@royalhaskoning.com
ABSTRACT
The urban water systems of the municipality of Arnhem (Netherlands) are facing water
quality problems that have to be addressed properly due to WFD. Therefore, the Municipality
of Arnhem has set up an ambitious three year research project aiming to gain knowledge on
the ecological status of receiving urban waters and the cost effectiveness of rehabilitation
options of these receiving waters, such as development of natural banks or end-of-pipe
treatment facilities for treating urban runoff discharged by storm sewers.
This article describes the results of the monitoring campaign focussing on the emission from
storm sewers and the removal efficiency of three full scale storm water treatment facilities,
comprising a sand filter, a lamella separator and an infiltration (soil) filter. The results of
monitoring over 150 storm events in 2006 and 2007 show that the emission from storm
sewers is indeed significant with respect to nutrients and heavy metals and, in addition, storm
water is unsafe from a microbiological point of view. The treatment efficiencies of all
treatment facilities are in the range to be expected given the dominant treatment processes,
with the sand filter and soil filter performing at a higher removal rate than the lamella settler.
KEYWORDS
monitoring, stormwater treatment, stormwater characteristics, lamella settler, sand filter, soil
filter, EMC
INTRODUCTION
Separated sewer systems are widely applied in Europe as well as in the United States. Storm
sewers are known to contribute significantly to the annual pollution loads into the receiving
waters. Measuring programs show that this annual pollution load per ha can easily exceed the
load per ha discharged by combined sewer overflows. The pollution load discharged by storm
sewers origins form the pollution associated with the rainfall itself, the pollutants taken with
the flow during the rainfall runoff process and from illicit (or wrong) connections (Salvia-
Castellvi et al., 2005).
The available measurement data on the emission from separated sewer systems, however, is
relatively sparse. Moreover, available data is often published without a well description of
meta data, like measurement setup, sewer system and catchment characteristics. Even
essential details on the monitoring itself, such as whether the data is derived from grab
samples or flow proportional samples is often lacking (Boogaard and Lemmen, 2007).
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
2Storm water sewers: pollution levels and removal rates of three full scale storm water
treatment facilities in Arnhem
The European Water Framework Directive (WFD) demands an enhanced protection of the
aquatic environment. As a consequence, the WFD requires municipalities to address the
emission from wastewater systems properly and to take action when these emissions whether
they origin from storm water systems, combined sewer overflows or wastewater treatment
plants, affect the water quality of receiving waters.
The urban water systems of the municipality of Arnhem (Netherlands) are faced with a
number of problems: soil and water quality do not always comply with standards, banks are
very unnatural, receiving waters are fed with polluted storm water and upward nutrient rich
seepage from canals and rivers. In addition, knowledge on the ecology of urban water systems
and the annual pollutant loads discharged by the storm water outfalls of storm sewers, is
limited.
Therefore, the Municipality of Arnhem has set up an ambitious three year research project
within the EU Interreg IIIB project URBAN WATER aiming to gain knowledge on
(Vermonden and Velthorst, 2008):
• ecological status of receiving urban waters in Arnhem
• cost effectiveness of rehabilitation options of these receiving waters:
o improvements of receiving water by dredging, development of natural banks
o end-of – pipe treatment facilities for treating urban runoff discharged by storm
sewers
This article describes the results of the monitoring campaign focussing on the emission from
storm sewers and the removal efficiency of three full scale storm water treatment facilities.
The full scale installations comprise a sand filter, a lamella separator and an infiltration (soil)
filter, with hydraulic capacities ranging from 50 to 250 m3/h.
METHODS
The southern part of the city of Arnhem, developed between 1960 and 2000, comprises 300
ha of separated sewer systems, with over 300 storm water outfalls. At three of these outfalls, a
full scale storm water treatment facility has been located, fully equipped with monitoring
devices in order to be able to determine:
• EMC’s of major pollutants in storm water
• removal efficiencies under various loading conditions of the treatment facilities
• impact of operation and maintenance on performance of treatment facilities
The effort necessary to guarantee the quality of the measurement data obtained is described in
detail by (Liefting and Langeveld, 2008).
Lamella settler.
The lamella settler is located at the Dordrechtweg. The contributing area comprises 3.8 ha,
developed between 1970 and 1980. The majority of the houses are owned by a housing
corporation, with mainly terraced housing. Figure 1 gives a schematic of the lamella settler.
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
Langeveld et al. 3
3. Automatic sampling of influent and effluent
2. Lamella filter in manhole
1. Storm sewer
with flow sensor
4. Filtered storm
water is
discharged to
the canal.
Q
Figure 1. Schematic of lamella filter
Upstream of the lamella settler an overflow weir is installed in order to maximise the flow at
250 m3/h to the lamella settler. This maximises the hydraulic surface loading at 1 m/h. The
lamella settler has a nominal design capacity of 50 m3/h, at which the hydraulic loading is
equivalent to a surface loading of 0.2 m/h. The flow to the lamella settler is measured at a 1
minute interval by an ultrasonic flow sensor (Endress+Hauser PROMAG 50 W), using a
Doppler shift to determine the velocity in the fully submerged sewer Ø 250 mm. The
automatic sampling system of the influent and effluent of the lamella settler is switched on
and off when the flow is above the threshold of 30 m3/h. The samples taken are stored at 4º C
in the Efconomy sampler system. As soon as a sample is taken, a SMS signal is send to the
operator, who takes care of collecting the samples for subsequent analysis in the laboratory of
waterboard Rivierenland in Tiel. This procedure also applies to the soil filter and sand filter.
Table 1 gives an overview of the parameters analysed in the laboratory.
Table 1. Water quality parameters analysed
Water quality problem related
parameter
Oxygen depletion SS, BOD, COD
Eutrophication NH3-NH4+, TKN, Total-P, Otho-P
Toxicity Pb. Zn, Cu, PAH (16 compounds)
Hygienic quality E.Coli, Tot. Coli, Feac. Strept., Therm. Coli
Soil filter
The soil filter is located at the Burg. Matsersingel. The contributing area comprises 5.6 ha,
developed between 1980 and 2000. The majority of the houses are privately owned semi-
detached dwellings. Figure 2 gives a schematic of the soil filter. The soil filter is operated by a
pump with a capacity of 25 m3/h. The pump starts operating as soon as the flow in the storm
sewer exceeds 50 m3/h. The flow is also measured at a 1 minute interval by an ultrasonic flow
sensor (Endress+Hauser PROMAG 50 W) in a 400 mm contraction in the 800 mm sewer. The
soil filter, with a surface of 300 m2, is located at the banks of the receiving water. The filter
layer consists of sand, with a design hydraulic conductivity of 1.5 m/d. At a depth of 60 cm,
well above the groundwater table, drains collect the treated stormwater and discharge via a
sampling manhole to the receiving water. The hydraulic performance of the soil filter is
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
4Storm water sewers: pollution levels and removal rates of three full scale storm water
treatment facilities in Arnhem
monitored by two water level sensors at a 1 minute interval, located in the soil filter. The
influent of the soil filter is sampled at an interval of 5 minutes as long as the influent pump is
running, using the same type of sampler as applied at the lamella filter. The effluent of the soil
filter is manually taken at the sampling manhole, as automatic sampling from the drains is too
difficult due to the long and varying retention time of the storm water in the soil filter. The
sampling manhole provides a completely mixed sample per storm event.
5. A drain collects effluent and discharges to the canal.
1. Storm sewer with flow sensor
3. Automatic sampling of influent
6. Manual sampling of effluent via a manhole
4. Soil filter
2. Storm water
is pumped into
soil filter
Q
Figure 2. Schematic of soil filter
Sand filter
The sand filter is located at the Brabantweg. The contributing area comprises 4 ha, developed
around 1980. 40% of the mainly terraced houses ore owned by a housing corporation, 60 % is
privately owned. Figure 3 gives a schematic of the sand filter. The sand filter has a design
capacity of 25 m3/h. The sand filter is designed at a surface loading of 10 m/h. the sand filter
consists of two parallel pressure tanks. A soon as the hydraulic resistance of the sand bed
exceeds a threshold, the filter is backwashed automatically. The back wash, containing the
retained pollutants, is discharged to a nearby sanitary sewer.
This pump starts operating as soon as the flow in the storm sewer exceeds 50 m3/h. The flow
in the 700 mm storm sewer is measured by an ultrasonic flow sensor (OCM PRO), with
ultrasonic velocity measurement. The measuring range of the sensor is adjusted to 0-200 m3/h
in order to guarantee stable operation of the pump. The influent and effluent of the sand filter
is sampled automatically at a 5 minute interval.
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
Langeveld et al. 5
3. Two sand filter installations in a cellar.
Automatic sampling of influent and effluent.
2. Storm
water is
pumped into
sand filter.
1. Storm sewer with flow sensor
power
supply
4. Filtered
water
returns to
storm
sewer.
5. Storm sewer discharges to canal.
Q
Figure 3. Schematic of sand filter. The sand filter is placed subsurface in the banks of the
receiving water
RESULTS AND DISCUSSION
The three full scale storm water treatment facilities have been in operation from august 2006
till December 2007. In this period, over 60 storm events have been monitored at the lamella
filter and soil filter, and only approximately 30 at the sand filter due to operational problems
with the flow sensor. For each storm event, EMC’s have been determined by analysing a
completely mixed sample of all samples taken during the events. In addition, for a number of
events and parameters, the variation in concentration levels of suspended solids during the
event has been analysed.
Storm water characteristics: EMC’s
Table 2 shows the storm water characteristics measured in Arnhem. For all parameters the
mean of the EMC’s is significantly higher than the median of the EMC’s. This is due to the
large variation in concentrations, with maximum EMC’s that can be extremely high. The
same phenomenon was observed in a comparable research project in Luxembourg. (Salvia-
Castellvi, 2005). In the Netherlands, STOWA has launched a project aiming at archiving all
monitoring data on urban runoff quality. This resulted in a comprehensive database (Boogaard
and Lemmen, 2007). Comparing the Arnhem data with the Dutch national database learns that
the storm water in Arnhem is comparable with the average storm water quality in literature.
This means that the loading of the treatment facilities in Arnhem was not exceptional, thus
making the results more widely applicable. The microbiological parameters show that, albeit
urban runoff is often considered ‘clean’, stormwater exceeds by far the standards of 200
E.Coli/100 ml for swimming water. The PAH concentrations were in most storms below
detection limits and therefore not listed in table 2.
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
6Storm water sewers: pollution levels and removal rates of three full scale storm water
treatment facilities in Arnhem
Table 2. EMC’s in storm sewers in Arnhem compared with median concentrations of all
samples in the Dutch STOWA national database (Boogaard and Lemmen, 2007) (n = number
of events in Arnhem)
Dutch STOWA
database
Lamella settler Soil filter Sand filter
Median n mean median n mean median n mean median
SS 20 338* 46* 14* 50 19 13 35 45 20
BOD 4.0 20 4.6 2.7 17 5.3 3.2 10 10.4 4.3
COD 32 56 29 22 48 25 22 32 38 24
NH3-NH4 - 27 0.5 0.3 26 0.62 0.35 21 0.3 0.2
TKN 1.7 58 2.9 1.2 50 1.5 1.3 28 1.9 1.1
TP 0.26 55 0.24 0.14 48 0.26 0.21 25 0.25 0.14
Pb 12 63 19 7 47 15 9 29 39 13
Zn 95 63 128 80 47 88 68 29 90 60
Cu 10 63 22 12 47 21 19 29 24 15
E. Coli 1.E+4 15 3.E+03 5.E+02 13 3.E+04 3.E+04 11 5.E+03 5.E+03
Tot. Coli - 16 6.E+07 4.E+04 14 4.E+07 1.E+05 11 5.E+06 5.E+04
Strep.faecalis - 11 7.E+03 6.E+02 9 2.E+04 2.E+04 9 1.E+04 4.E+03
Therm. Coli - 15 4.E+03 8.E+02 13 2.E+07 4.E+04 11 3.E+04 3.E+03
* The SS concentration of the lamella settler is the mean and median of all samples, rather
than an EMC
Variability in storm water concentrations
The variation of EMC’s between storm events is significant, as shown in the previous
paragraph. Figure 4 and 5 show the variation in the concentration of suspended solids during
two storm events. The storm event of 7 December 2006 consisted of two sub-events, the first
with a total precipitation height of 15 mm in 6 hours and the second with 5 mm in 30 minutes,
see figure 4. During the first sub-event the flow is moderate at 50 m3/h and influent
concentrations range between 20 and 40 mg/l. During the second sub-event the flow peaks at
200 m3/h accompanied by suspended solids concentrations in the effluent peaking above 200
mg/l. The difference between influent and effluent concentrations is significantly higher for
elevated suspended solids concentrations. This phenomenon is observed for many storm
events.
Figure 4. Variation of suspended solids concentration during storm event of 7 December
2006 in influent and effluent of lamella settler, Arnhem.
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
Langeveld et al. 7
Figure 5 illustrates the improved functioning of the lamella settler at higher concentrations of
suspended solids for the storm of 11 January 2007. An explanation for this behaviour may be
the sudden release of in-sewer sediments during peak storms. This material, which has
previously settled in the sewer, is much more settleable than the suspended solids transported
during moderate storm events, see also figure 7.
The variability of pollutant concentrations during storm evens in storm sewers has not
received much attention in literature. As a consequence, in many monitoring projects only a
few grab samples are taken to research the quality of the storm water. This may easily lead to
erroneous results, as the concentration may vary a factor 40 within 30 minutes, as illustrated
in figure 5.
Figure 5. Variation of suspended solids concentration during storm event on 11 January 2007
in influent and effluent of lamella settler, Arnhem. The peak concentration of suspended
solids was as high as 2000 mg/l. The storm water sampled during this event was reported to
be ‘black’.
Removal rates
Figure 6 gives an overview of the removal rates of the most relevant parameters. The
calculated removal rates are based on the total mass removed. The removal efficiency of
lamella settlers is significantly lower than the removal efficiency of the soil filter and sand
filter. This clearly represents the difference in dominant removal processes. The removal of
pollutants in the lamella settler is based on settling, whereas the sand and soil filters rely on
filtration and possibly adsorption of pollutants.
Soil filter and sand filter. The removal rates of the soil filter and sand filter are in accordance
with rates reported in literature, such as the BMP reports of the Daywater project (Rombout et
al. 2007). E.g. for suspended solids, a removal efficiency between 70-90% is to be expected
for a soil filter, with 68% in Arnhem and between 80-90% for a sand filter, with 84% in
Arnhem. For heavy metals, the removal efficiency of a soil filter ranges between 80 and 90%
and of a sand filter between 50 and 80%.(Rombout et al. 2007). In Arnhem, these removal
rates vary between 65 and 84% for the soil filter and between 32 and 82% for the sand filter,
thus also confirming literature.
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
8Storm water sewers: pollution levels and removal rates of three full scale storm water
treatment facilities in Arnhem
Lamella settlers. With respect to lamella settlers, Daligaut et al. (1999) report a removal
efficiency of 54% in a project in Brunoy and 30% in Vignuex. The removal efficiency of the
lamella settler in Arnhem, based the volume-averaged removal rate of 60 storm events, is
37% for suspended solids. In addition, the removal rates for heavy metals are also in the same
range. E.g. for lead the rates are 44% in Brunoy and 28% in Vignuex, while the removal rate
in Arnhem is 39%.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
SS COD NKj Ptot Cu Pb Zn
Removal efficiency
lamella settler
sand filter
soil filter
Figure 6. Removal rates of lamella settler, sand filter and soil filter in Arnhem
Settling velocities
Within the framework of a BSc project, van Geelen en Kloppenburg (2007) performed
additional analyses on the settling velocities of pollutants in the influent of the lamella settler.
The settling velocities have been determined using a column of 1500 mm height and 42 mm
diameter. The column was filled with a fully mixed homogeneous sample, without further
sample preparation. Sampling of water took place at desired moments via the lowest sampling
point, 50 mm from the bottom of the column. The suspended solids concentration was
determined using a laboratory turbidity meter HACH 2100N and subsequent lab analysis.
For 13 storm events, the settling velocities have been determined for 2 grab samples, resulting
in 26 settling curves as shown in figure 5. The results show a significant variation between
events. For the majority of events, the percentage of particles with lower velocities than 0.2
m/h, the nominal design capacity of the lamella settlers, is at least 65%. Consequently, the
removal rate of the lamella settler in Arnhem would be expected to be maximal 35%, which is
in accordance with the observed removal efficiency of 37%. Only in the storm event of 11
January 2007, shown in figure 5, the percentage of particles with lower velocities than 0.2
m/h is approximately 50%, the value found in a comparable study in Le Marais, Paris
(Gromaire-Mertz, 1999).
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
Langeveld et al. 9
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
100.0
0.01 0.1 1 10
Settling velocity (m/h)
% of particles (in mass) with lower velocities
21-11-2006
21-11-2006
23-11-2006
23-11-2006
4/5-12-2006
4/5-12-2006
7-12-2006
7-12-2006
8-12-2006
8-12-2006
16-12-2006
16-12-2006
30/31-12-2006
30/31-12-2006
1-1-2007
1-1-2007
4-1-2007
4-1-2007
6-1-2007
6-1-2007
10-1-2007
10-1-2007
11-1-2007
11-1-2007
17/18-1-2007
17/18-1-2007
Figure 7. Settling velocities of suspended solids in the influent of the lamella settler, Arnhem.
The settling velocities are based on grab samples at two stages per storm event.
The settling velocities of the storm water samples in Arnhem indicate that applying the
normal Dutch standard design surface loading of 1 m/h for a lamella settler (Rombout et al.,
2007) will result in unsatisfactory removal efficiencies.
CONCLUSIONS
The municipality of Arnhem has significantly increased the available knowledge on storm
water quality and the potential of treatment techniques by launching the three year
INTERREG III B project. During the monitoring campaign, over 150 storm events at three
locations have been monitored. The results of the monitoring campaign over 2006 and 2007
show that
• the emission from storm sewers is indeed significant with respect to nutrients and
heavy metals and, in addition, storm water is unsafe from a microbiological point of
view;
• the concentrations of pollutants in discharges from storm water sewers vary strongly
between en during storm events. This is especially relevant with respect to the
interpretation of available literature data on storm water emissions, often derived from
grab samples;
• local treatment of storm water is a viable option;
• the achievable removal rate of storm water treatment options strongly depends on the
characteristics of the stormwater itself, e.g. in terms of settleability of specific
pollutants and the ratio between dissolved and suspended material;
• the treatment efficiencies of all treatment facilities are in the range to be expected
given the dominant treatment processes, with the sand filter and soil filter permofing
at a higher removal rate than the lamella settler.
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
10Storm water sewers: pollution levels and removal rates of three full scale storm water
treatment facilities in Arnhem
ACKNOWLEDGEMENTS
The authors would like to gratefully acknowledge the funding by the EU Interreg project IIIB
and the additional funding by STOWA.
REFERENCES
Boogaard, F. en Lemmen, G. (2007). STOWA Regenwaterdatabase. De feiten over de kwaliteit van afstromend
regenwater. STOWA 2007-21 (in dutch: storm water database)
Daligaut, A., Meaudre, D., Arnault, D and Duc, V. (1999). Stormwater and lamella settlers: efficiency and
reality. Wat.Sci.Tech. 39(2)-93-101
Geelen, van, F. and Kloppenburg,W. (2007). Ontwerp lamellenfilter bedrijventerrein “de Overmaat” BSc thesis
Hogeschool Arnhem and Nijmegen (in Dutch: Design of a lamella settler for industrial area De Overmaat)
Gromaire-Mertz, M.C., Garnaud, S., Gonzalez, A. and Chebbo, G. (1999). Characterisation of urban runoff
pollution in Paris. Wat.Sci.Tech. 39(2)-1-8
Liefting, H.J. and Langeveld, J.G. (2008). Sewer monitoring projects: data collection, data handling and data
quality in Arnhem. Submitted to 11th Int. Conf. on Urban Drainage
Rombout, J., Boogaard, F., Kluck, J. en Wentink, R. (2007). Zuiverende voorzieningen regenwater. STOWA
2007-20 (In dutch: storm water treatment facilities)
Salvia-Castellvi, M., Iffly, J.F., Guignard, C. and Pfister, L.(2005). Characterisation of urban stormwater runoff
from separated sewer overflows in Luxembourg. 10th Int. Conf. on Urban Drainage, Copenhagen/Denmark,
21-26 August 2005
Vermonden, K and Velthorst, H.(2008). Monitoring ecology in urban water systems in Arnhem and Nijmegen,
the Netherlands: monitoring helps saving money! Urban Water Final Conference, Dortmund, Germany, 9th-
10th April 2008
