You are on page 1of 19

Ecological Engineering 24 (2005) 201–219

Determination of COD, BOD, and suspended solids loads during


combined sewer overflow (CSO) events in some combined
catchments in Spain
J. Suárez ∗ , J. Puertas
Civil Engineering School, Universidade Da Coruña, Campus de Elviña, 15192 A Coruña, Spain

Received 25 March 2004; received in revised form 15 September 2004; accepted 16 November 2004

Abstract

The main topic of the research project presented here in part is the determination of pollutant loads during combined sewer
overflow events in several experimental catchments in Spain that were determined to obtain a pollution pattern which may be
applied to other catchments in Spain or an order of magnitude of pollution loads in Spanish catchments. The existence or non-
existence of a first-flush is also examined. Five catchments were monitored in five Spanish cities, and both hydraulic parameters
and pollutant amounts were measured in dry conditions and during rainfall events. From the data obtained, some parameters
were extracted for each catchment and an order of magnitude of the pollution amounts was proposed for Spanish catchments,
in terms of event mean concentration (EMC) and maximum concentrations, since the moderate differences between catchments
permit an estimation of a mean value.
© 2004 Elsevier B.V. All rights reserved.

Keywords: Combined sewer overflow; Water pollution; Urban drainage

1. Introduction impact on rivers and lakes caused by CSO events is ac-


cepted as an important source of pollution. This prob-
Urban drainage projects are mainly concerned with lem was first pointed out in the 1970s and has been
the protection of the quality habitat of receiving wa- studied over the course of the last few decades (Butler
ters. A major reason for the long-term persistence of and Davies, 2000).
poor quality waters is the continued existence of uncon- European regulations concerning water quality stan-
trolled or poorly controlled discharges from combined dards in rivers, which classifies the rivers according
sewer overflows (CSOs) and surface water runoff. The with the standards that different kind of fish need to
live in, are mainly broken because of CSO discharges.
∗ Corresponding author. Fax: +34 981167170.
Some countries have made an effort to determine pol-
E-mail addresses: suarez@iccp.udc.es (J. Suárez), lution levels in the sewers as well as receiving waters,
jpuertas@udc.es (J. Puertas). and structures and best management practices (BMPs)

0925-8574/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.ecoleng.2004.11.005
202 J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219

have been designed to control this pollution source (i.e. one country to another, and even from one catchment
WRC, 1987). to another within one country. No general trends have
Generally two types of the wet weather pollutant been defined. Some of the results on mean values are
sources exist: summarized below:
• Discharges form combined sewer systems (CSOs), • Ellis (1989): SS: 425 mg/L; BOD: 90 mg/L; COD:
by which a mixture of industrial wastewater, urban 380 mg/L.
surface runoff, domestic wastewater and sewer de- • Lager et al. (1977): SS: 370 mg/L; BOD: 115 mg/L;
posits are discharged into receiving waters. COD: 367 mg/L.
• Discharges form separate systems, by which mainly • Metcalf and Eddy (1991): SS: 270–550 mg/L; BOD:
the runoff from urban surfaces is discharged. 60–220 mg/L; COD: 260–480 mg/L.
• NWRW (1991): SS: 105–320 mg/L; BOD:
These types of discharges are different with respect
40–124 mg/L; COD: 148–389 mg/L.
to the volume discharged, the concentrations of pollu-
tants in the water phase and the periodicity or discharge. The ATV-128 German standard (1992) made con-
It is very important to know the maximum concentra- siderable progress in the design of structures to reduce
tion of every pollutant, which permits evaluation of CSO effects. This standard uses COD as a pollution
the acute impact (immediate effects) over the receiving parameter, and considers CSO pollution to be calcu-
waters and the habitats involved, and the event mean lated as the dilution of wastewater, with a mean COD
concentration, which permits knowing the total mass value of 700 mg/L, in rainwater, with a COD value
discharged (immediate effects and long-term effects). of 117 mg/L. The German standard, however, is based
These contaminants will affect stream organisms and upon the hypothesis that separate catchments produce
alter the character of the ecosystem. reasonable levels of pollution, which is not true in
Table 1 shows the impacts of some key contam- many cases. This topic has been discussed or is un-
inants on receiving waters (modified from Lijklema der revision in many countries. Some field work has
et al., 1993) by discharges of CSO, sewage treatment been done on separate catchments, indicating that the
plants, and surface water runoff. pollution of rainwater – when it reaches the receiving
Pollution from rainwater discharges to receiving wa- media – is not negligible (Bertrand-Krajewski et al.,
ters was first identified in early 1970s and this subject 1998). Even in Spain, some field work is being done
has been studied in several research projects, such as the by our group to measure pollution loads in separate
Nationwide Urban Runoff Program, US-EPA (1983). systems.
From these projects, an order of magnitude was ob- As a preliminary approximation, if we consider a
tained to evaluate pollution from rainwater discharges. dilution overflow criteria of about 6:1, as was pro-
However, the main result is the evidence of the com- posed in the early British approaches to the problem
plexity of the problem, with many degrees of freedom, (Butler and Davies, 2000), according to the ATV-128
and pollution values that show enormous variations German standard, the concentration of the CSO should
from one catchment to another. This is also true if we be (6 × 117 + 1 × 700)/7 = 200, in terms of COD. These
compare CSO pollution data. Soil use, population den- pollution values have been considered too low for some
sity, traffic intensity, etc. are parameters whose vari- real catchments in Spain, whose measured pollution
ations produce huge differences in pollution amounts values seem to be quite a bit higher. As these mea-
(Butler and Davies, 2000). surements are isolated tests done on a small number
As geographic and demographic aspects appear of CSO events, more accurate information about these
to be important, there may be noticeable differences phenomena would be necessary in order to ascertain
among countries which have different city planning the real COD concentrations of CSO events in Span-
and sanitary policies. In Europe, some field work has ish catchments. The possible reasons for the increase
been done. The projects by Ellis (1991) and others pre- in COD values in CSO events may be attributed to the
sented in Valiron and Tabuchi (1992) are worthy of removal of sediments from the sewer network. These
mention. Both separate and combined sewer networks sediments cannot be removed by wastewater, whose
have been studied. The results differ substantially from hydraulic energy is moderate, but can be removed by
J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219 203

Table 1
COD: chemical oxygen demand, BOD: biochemical oxigen demand, NH4 + : ammonium ion, Ntot: total nitrogen, Ptot: total phosphorus, CSO:
combined sewer overflow, STP: sewage treatment plant (Lijklema et al., 1993)
Contaminant (state variable) Environmental effects Ecological impact
1. Oxygen demanding Dissolved oxygen Food web
From CSOs, sewage
reduction
treatment plants
COD, BOD
(STP), and surface
water runoff
Biodiversity
Energy dynamics
Biomass
Food web
accumulation
Ecosystem development
Dissolved oxygen Food web
From CSOs, sewage
reduction
treatment plants
NH4 +
(STP), and surface
water runoff
Biodiversity
Energy dynamics
Biomass
Food Web
accumulation
Ecosystem development

2. Nutrients From CSOs, STP and Energy dynamics


Ntot Enrichment
surface water runoff
Food web
From CSOs, STP and Critical species
Ptot Enrichment
surface water runoff
Ecosystem development

3. Toxicants Food web


NH4 + Toxicity
Biodiversity
Critical species
Acute Toxicity Food web
Metals Biodiversity
Cumulative Toxicity
Critical species
Ecosystem development
Food Web
Organic
Cumulative Toxicity
micropollutants
Biodiversity
Critical species

Public health (primary Energy dynamics


4. Hygiene Fecal bacteria effect), biomass
(secondary effect)
Food web
Ecosystem development
204 J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219

Table 1 (Continued )
Contaminant (state variable) Environmental effects Ecological impact
5. Physical Energy dynamics
Temperature
Temperature rise + long term
change
Food web
Genetic diversity
Dispersal and migration
Blanketing + harm to Critical species
Suspended solids
fish
Dispersal and migration
Food web
Washout; morphology
Flow
changes
Critical species
Ecosystem development
Food web
Excess dissolved
Chloride
solids
Genetic diversity
Ecosystem development

combined waters, which have more discharge, velocity, cleaning policy, slopes of the catchment and the sewer
and shear stress. system, all influence the temporal variability in the con-
If this removal is produced during the first stages centration and the load of the pollutants (Saget et al.,
of the storm, first-flush phenomena have to be con- 1996; Bertrand-Krajewski et al., 1998). These authors
sidered not only as a wash off of the catchment but defined the first-flush as being when at least 80% of
also as a result of the cleaning of the sewer network. the pollution load is transferred in the first 30% of
If sediment levels are substantial and if any colloidal the volume. From an analysis of 197 rainfall events
or cohesive structures have been formed, the removal in 12 separate and combined sewer systems, the au-
processes could be very slow and difficult to analyze, thors showed that in separate sewer systems, 80% of
possibly generating rather large amounts of pollution the total pollutant mass is transported in the first 74% of
during the full process without any relevant first-flush the total volume for 50% of the rainfall events. In com-
effects. bined sewer systems, 80% of the total pollutant mass
The first-flush of pollutants has been identified in a is transported in the first 79% of the total volume for
relatively high proportion of the total storm pollution 50% of the rainfall events. Whether or not a first-flush
load that occurs in the initial part of the combined sewer does exist is very important when attempting to design
runoff. Thornton and Saul (1986) and Pearson et al. CSO control structures.
(1986), mentioned by Gupta and Saul (1996), defined The variation of the pollutant concentration during
the first-flush as the initial period of storm flow during an event is described by the pollutograph. For a com-
which the concentration of pollutants was significantly parison of different events, a non-dimensional curve is
higher than those observed during the latter stages of more appropriate (Saget et al., 1996). For this reason
the storm event. Previous studies have highlighted the it is necessary to study the pollution load distribution.
fact that the previous dry weather conditions, the length For each event and each pollutant, we used the cumu-
of the previous dry weather period, the magnitude and lative load divided by the total pollution load versus
pollutant characteristics of the dry weather and storm the cumulative volume divided by the total volume of
flows, together with the characteristics of the sewer the event curves (curves L(V)). Each curve can be fit-
system, including the area of the catchment, soil use, ted by a function of the Y = Xa type, where “Y” is the
J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219 205

fraction of the discharged pollution load and “X” is the of pollution loads in Spanish catchments. The existence
corresponding fraction of volume. The value of param- or non-existence of a first-flush is also examined.
eter “a” characterizes the deviation of the curve from The catchments were chosen to cover different cli-
the diagonal. These curves permit an analysis of the matic regions and areas all over Spain. All of them
existence of first-flush. have combined sewer systems and the soil is mainly
There has been some effort – although rather sparse used for residential purposes, although there are some
– over the course of the last 10 years in Spain, differences in population density.
directed at preventing CSO discharges, mostly in One of the topics to be taken into account is the con-
the northern catchments, carried out by the Confed- trol of inputs and outputs in the catchments. Prelimi-
eración Hidrográfica del Norte, which has constructed nary studies were conducted to ensure that the catch-
more than 100 first-flush tanks, and in Barcelona, ments observed had a single lowest point in their sewer
by the Clavegueram de Barcelona (CLABSA), which systems, and also that no parasite waters (inflows) were
has built huge tanks in the city with a double entering the system. In many cases, these preliminary
aim: flood prevention and pollution control. Sev- studies were done on the basis of the sparse or obsolete
eral academic groups have also been working in this data available on the sewer system, which have since
field—in Barcelona and its surroundings (Universitat been up-dated and completed during this project.
Politècnica de Catalunya), in Santander and all around As will be pointed out in Section 4, one of the catch-
the Cantabrian coast (Universidad de Cantabria), and ments (Valencia) was discovered to have parasite inputs
in Santiago de Compostela and Asturias (Universidade that were not previously detected, for which reason its
Da Coruña). results were clearly different from the others. Sewer
Some municipalities are now interested in this prob- systems in Spain started to be constructed many cen-
lem, and this interest has been taken up by the Spanish turies ago (dating back as far as the Roman Empire or
Ministry of Environmental Protection (Ministerio de to the Islamic Kingdoms, in some cases), and the im-
Medio Ambiente), which has invested its own funds in provements and additions which they have undergone
the field work that is partially presented in this paper. have seldom been documented. Only today are all the
This work was sponsored by the Spanish government, sewer systems being put into GIS systems. A GIS sys-
and most of large municipalities in Spain, grouped into tem requires a topographic study that will allow for the
the GADU (Group on Advanced Wastewater Manage- location of each pipe and every singular device in the
ment) and the AEAS (Spanish Association of Drinking system. It was during one of these topographic jobs
Water and Sewerage Management). that the municipality of Valencia noticed the presence
of inflows coming from an irrigation system whose pol-
lution levels were obviously much lower than those of
2. Definition of the research project an urban sewer system, as can be observed in the re-
sults.
The main topic of the research project which is par- The cities and catchments studied are presented in
tially presented in this paper is the determination of Table 2, which includes some data on their size, and in
pollutant charges during CSO events in several experi- Fig. 1, which shows the location of the cities.
mental catchments in Spain, for the purpose of obtain- Barcelona and Valencia are two of the biggest cities
ing a pollution pattern which may be applied to other on the Mediterranean coast, and their climate may
catchments in Spain or, at least, an order of magnitude be considered completely Mediterranean, with mod-

Table 2
Experimental catchments (ha: hectares; inh./ha: inhabitants per hectare)
City Barcelona Madrid Sevilla Vitoria Valencia
Catchment name Bac de Roda Arroyo del Fresno Los Remedios Almendra Malvarrosa
Area (ha) 170 3800 135 132 89
Population density (inh./ha) 235 60 380 140 200
Climate Mediterranean Continental Semi-arid Oceanic Mediterranean
206 J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219

Fig. 1. Map of Spain. Location of the experimental catchments.

erate annual rainfall heights, but with a very great risk Madrid is in the very center of Spain, and its climate
of high intensity storm events, with normal values of is continental, with a low annual rainfall height, high
100–200 mm/h an hour, for return periods of no more summer temperatures and cold winters. The receiving
than 5 years. river is the Manzanares, whose source can be located
In addition, both cities are quite flat, with sewer at one of the WWTPs of the city. So, quality is the
slopes of about 0.1–0.2%, so their sewer systems con- main issue for the municipality of Madrid, as the qual-
sist of enormous sections which are designed as flood ity of the Manzanares River is quite poor during CSO
evacuation structures, with CSO pollution being a sec- events.
ondary problem in most cases. During the last 5–10 Vitoria, in the central, northern part of the country,
years some effort has been made in both cities – mostly has an oceanic climate with rainy winters of moderate
in Barcelona – to limit CSO events, by constructing intensity. Vitoria is considered to have one of the high-
huge reservoirs with a double aim: limiting the flood est standards of living of all the cities in Spain, and
peaks and driving most of water to the WWTP. its municipality is extremely concerned with environ-
Sevilla is located in the south of Spain and its climate mental conservation. The Zadorra River, tributary of
is semi-arid, with high temperatures, and a moderate the Ebro River, is the receiving river, and the improve-
rainfall regime, and no pronounced heavy rainfall in- ment of its quality is a main issue for the municipality
tensities. As floods are not a major problem in the city, of Vitoria.
future investments are to be earmarked for the reduc- The Atlantic area has not been covered in this study,
tion of CSO pollution. as its climate is substantially different from the general

Fig. 2. Example of properly measured event. The marked area corresponds to the total sampling time. Event Sevilla 3.
J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219 207

Spanish climate. Further research needs to be done in • Water quality parameters, which may be grouped by
this area. families.
◦ COD, BOD5 , filtered TOC.
◦ Turbidity, total solids (TS) suspended solids (SS),
3. Development of the experimental work dissolved solids (DS) and volatile fractions (VSS,
VDS).
3.1. Definition of the parameters ◦ Kjeldahl nitrogen (TKN), ammonia (NH4 + ), total
phosphorus (P).
As CSO events include both hydrologic and qualita- ◦ Metals (Cu, Pb, Zn, Cd).
tive variables to be measured, the experimental work is ◦ Total hydrocarbons (HC).
based upon the measurement of numerous parameters ◦ Conductivity, pH, temperature.
of both types:
• Hydrologic parameters, including the preceding dry These parameters were measured both in dry
weather period (days), rainfall measured at 5 min weather conditions and during rainfall events. Con-
intervals, and level and discharge at the lowest point ductivity, pH and temperature, as well as hydro-
of the main pipe in each catchment. logic parameters were measured on a continuous basis

Fig. 3. Examples of hydrographs and pollutographs. (a) Event Sevilla 4 (discharge, BOD, COD); (b) Event Sevilla 5 (discharge (Q), suspended
solids (SS), suspended volatile solids (SSV), dissolved solids (SD), dissolved volatile solids (SDV)).
208 J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219

(every 5 min), while the other parameters were mea- search. The authors will provide this information to any
sured in the laboratory from the samples obtained by group upon request.
using a sampler (24 bottles).
Dry weather conditions were characterized by the
data from the continuous parameters as well as from 3.2. Event definition and parameters involved
specific sampling processes developed both in the win-
ter and summer months. During rainfall events, the wa- Once an event has been considered valid, that is to
ter level is used as a flag, so if a target level is reached, say, the hydrograph has been properly covered by the
the sampler starts working. Sampling time increments sampling process, an analysis of the data has to be done.
were studied in more frequent intervals during the first This analysis consists of both graphic information (i.e.
stages, when the hydrograph is expected to peak, and at Figs. 2, 3a and b), and parameter generation.
longer intervals during the latter, to get an idea of the The graphic information consists of a series of hy-
recession curve shape. Obviously, as hyetographs do drographs and associated pollutographs, where we can
not have a standard shape, some of the sampling series see if there is any decay in the pollutant concentration.
do not cover the full process accurately, so only a cer- In the graphs included as examples, a clear decay was
tain number of events have been considered in the data observed, but this is not always the case. Packs of these
process. When the sampling process was considered to graphs are available for each event and pollutant. In this
be incorrect, the samples were discarded. paper, only data regarding COD, BOD, and SS will be
In many cases, the minimum water level used to trig- discussed. Fig. 3 is given as an example.
ger the sampler was too high and portions of the initial Apart from the graphic results, which give us a no-
overflows were not sampled. If the water level used to tion of the evolution of pollution during the event, gen-
trigger the sampler is set too low, daily dry weather eral quantitative parameters have been extracted for
flow variations may trigger the sampler and, if some each event, including:
hours have passed after the occurrence of a rainfall, the
sampler bottles, which are already full, cannot capture • Maximum concentration (C. max) (mg/L) of each
the real event. It is not easy to capture an event properly pollutant.
from the very beginning. • Event mean concentration (EMC) (mg/L) of each
An EXCEL-format form was filled in for each sam- pollutant during the full event.
pling pack, including general, hydrologic, and pollu- • Total mass per surface unit (g/m2 ) of each pollutant
tion parameters. This information, which includes both during the full event.
numerical and graphic data, is available for further re- • Mass maximum flux (kg/s) of each pollutant.

Fig. 4. Example of L(V) curves. The “a” exponent value is 0.886 for COD and 0.853 for BOD, for this singular event (Vitoria 4).
J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219 209

The kind of information that can be extracted from by singular values, as is the case of maximum values.
these parameters is considerably different. Thus, max- Given that this is not a concentration value, it is possi-
imum concentration and maximum flux tell us about ble to have an idea of the pollutant mass which has been
pollution peaks. They may show a broad variation from removed. If there is some source of clean water in the
one catchment to another, and they have some depen- catchment, such as infiltration water through phreatic
dence on hydraulic values such as the slope of catch- levels, concentration values decrease, but total mass per
ments and sewers. So it is not possible to compare data surface unit values do not change. This can be observed
if the catchment characteristics are not similar. when analyzing the catchment of Valencia, whose con-
Also, the maximum flux depends on the magnitude centration values are clearly below average, but whose
of the catchment (area), since the full mass per sec- total mass values are above average. As values are ob-
ond is measured in the control section. So this value is tained per surface unit, this value can be extrapolated
very difficult to compare or extrapolate even in similar from one catchment to another.
places if catchment sizes are not the same. A reference value in the literature is usually the
The total mass per surface unit provides a global event mean concentration (EMC) which gives a mean
value for the whole event, which is not conditioned weighted value for the concentration during the event.

Table 3
Relevant parameters obtained for Madrid—Arroyo del Fresno catchment (11 events)
Event 2 3 4 5 6 7 8 9 10 11 12
Qmax (m3 /s) 3.967 3.156 3.342 3.440 3.472 3.344 3.140 3.252 2.820 3.322 3.438
Qmean (m3 /s) 3.505 2.750 2.837 2.481 3.178 2.278 2.452 3.051 2.263 3.031 3.174
Volume (m3 )
Total volume 12084 17939 18283 16519 22384 16175 8749 22036 8122 22325 22445
Runoff volume 6384 6141 5235 10285 7200 6815 5719 13171 5242 9005 11420
Runoff volume/total volume 0.528 0.342 0.286 0.623 0.322 0.421 0.654 0.598 0.645 0.403 0.509
Maximum concentration (mg/L)
COD 1782 1624 1346 1782 1560 2384 2089 250 230 990 752
BOD5 993 670 750 1060 758 1324 1625 186 178 545 625
SS 1465 1002 1117 1120 1400 2773 2000 980 186 604 434
EMC (mg/L)
COD 770 712 694 602 584 1208 1472 135 154 641 506
BOD5 409 372 412 266 313 673 961 93 107 300 322
SS 824 469 442 550 780 1086 1378 152 116 425 347
Surface loads (g/m2 )
COD 0.178 0.213 0.176 0.242 0.190 0.372 0.295 0.045 0.026 0.259 0.213
BOD5 0.071 0.101 0.112 0.104 0.087 0.188 0.195 0.035 0.018 0.113 0.139
SS 0.238 0.164 0.132 0.231 0.368 0.410 0.301 0.069 0.022 0.189 0.159
Mass discharges (kg/s)
COD 4.253 4.045 2.913 5.861 3.522 7.107 5.871 0.709 0.572 2.705 2.220
BOD5 1.779 1.730 1.825 3.482 1.529 3.882 4.817 0.535 0.446 1.522 1.622
SS 3.231 2.748 2.499 3.517 4.304 8.837 6.087 3.114 0.485 1.702 1.154
Runoff mass/total mass
COD 0.729 0.633 0.527 0.923 0.552 0.723 0.871 0.574 0.803 0.688 0.712
BOD5 0.547 0.578 0.567 0.897 0.475 0.654 0.882 0.654 0.800 0.642 0.729
SS 0.908 0.742 0.620 0.966 0.801 0.886 0.950 0.788 0.865 0.755 0.775
Shape parameter “a”
COD 0.576 0.731 0.702 0.695 0.637 0.620 0.736 0.802 0.670 1.011 0.934
BOD5 0.577 0.775 0.695 0.738 0.689 0.650 0.674 0.768 0.609 0.904 0.921
SS 0.804 0.656 0.578 0.550 0.701 0.540 0.807 0.418 0.692 0.978 0.993
210 J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219

It is a value that can be extrapolated regardless of the the value of “a”, the more severe the first-flush
size of the catchments, as it is a concentration value and becomes.
it does not depend on the magnitude of the catchment.
Apart from these basic parameters, a first-flush pa-
rameter must be calculated. As explained earlier, L(V) 4. Results
curves (Saget et al., 1996) allow a quantitative evalua-
tion of the existence of the first-flush by means of the 4.1. Data obtained
so-called “a” exponent, a shape parameter which fits
the data by using a Y = Xa expression. As an example, As the main result of this paper, the full set of sig-
Fig. 4 shows one set of curves for COD and BOD be- nificant data is presented. The whole series of source
longing to the Vitoria 4 event. A minor first-flush effect data from which these parameters have been obtained
can be observed, as both curves are above the diagonal are available to the scientific community for further
line. research.
Values for the shape parameter “a” of less than A set of parameters is presented for every catch-
1 denote the existence of a first-flush. The lower ment (Tables 3–7), including only valid events. Some

Table 4
Relevant parameters obtained for Sevilla—Los Remedios catchment (10 events)
Event 1 2 3 4 5 6 7 8 9 10
Qmax (m3 /s) 0.265 0.946 0.496 0.901 0.754 0.277 0.215 0.919 0.394 0.838
Qmean (m3 /s) 0.197 0.629 0.279 0.675 0.620 0.253 0.176 0.611 0.295 0.521
Volume (m3 )
Total volume 1434 4483 1750 4119 4188 344 1249 4225 2010 3718
Runoff volume 609 3589 1074 3492 3547 179 636 3406 1270 3029
Runoff volume/total volume 0.425 0.801 0.614 0.848 0.847 0.520 0.509 0.806 0.632 0.815
Maximum concentration (mg/L)
COD 3260 2158 1472 1840 2080 1320 506 1041 1805 1683
BOD5 1150 758 593 1043 530 705 410 777 780 845
SS 2939 1233 1335 1075 1433 991 921 978 2923 3394
EMC (mg/L)
COD 1873 887 977 689 781 1133 277 417 722 582
BOD5 842 396 402 202 274 643 177 247 328 377
SS 1249 678 851 535 845 909 346 407 757 755
Surface loads (g/m2 )
COD 1.623 2.562 1.131 1.833 2.224 0.234 0.166 1.058 0.745 1.392
BOD5 0.666 1.067 0.458 0.472 0.728 0.139 0.118 0.620 0.282 0.907
SS 1.082 2.077 1.050 1.511 2.556 0.213 0.240 1.182 0.906 2.011
Mass discharges (kg/s)
COD 0.762 0.820 0.266 0.715 1.005 0.281 0.075 0.375 0.317 0.398
BOD5 0.249 0.362 0.108 0.288 0.280 0.172 0.065 0.207 0.095 0.231
SS 0.733 0.645 0.493 0.760 1.066 0.226 0.150 0.417 0.486 0.609
Runoff mass/total mass
COD 0.816 0.870 0.893 0.872 0.918 0.809 0.645 0.811 0.694 0.869
BOD5 0.745 0.811 0.880 0.765 0.855 0.851 0.722 0.801 0.579 0.874
SS 0.816 0.922 0.952 0.925 0.975 0.921 0.751 0.927 0.804 0.967
Shape parameter “a”
COD 0.809 0.855 0.867 0.690 0.669 1.009 0.690 0.683 0.834 0.647
BOD5 0.855 0.841 0.769 0.567 0.688 1.003 0.722 0.650 0.821 0.730
SS 0.834 0.888 0.762 0.617 0.724 0.919 0.610 0.632 0.820 0.627
J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219 211

Table 5
Relevant parameters obtained for Vitoria—Almendra catchment (eight events)
Event 2 4 5 6 8 9 10-a 10-b
Qmax (m3 /s) 0.694 1.368 0.742 0.877 2.307 1.553 1.101 2.053
Qmean (m3 /s) 0.653 0.712 0.316 0.708 1.078 0.571 0.463 0.914
Volume (m3 )
Total volume 2339 2623 1313 4309 6010 4207 1790 1359
Runoff volume 1709 2047 737 3196 4961 3487 1171 1113
Runoff volume/total volume 0.731 0.780 0.561 0.742 0.825 0.829 0.654 0.819
Maximum concentration (mg/L)
COD 942 673 2626 600 1006 1232 2376 1711
BOD5 235 377 1028 273 489 525 740 503
SS 1094 1192 1776 380 1245 847 820 372
EMC (mg/L)
COD 311 431 1293 291 252 622 1129 1038
BOD5 136 203 617 146 128 294 587 480
SS 200 346 1239 182 240 390 599 174
Surface loads (g/m2 )
COD 0.387 0.758 1.043 0.478 0.957 1.797 1.236 0.957
BOD5 0.135 0.344 0.452 0.125 0.480 0.812 0.576 0.427
SS 0.247 0.652 1.077 0.389 0.955 1.209 0.712 0.152
Mass discharges (kg/s)
COD 0.592 0.883 0.877 0.224 0.868 1.806 1.246 1.824
BOD5 0.124 0.348 0.371 0.074 0.604 0.755 0.651 1.024
SS 0.625 0.723 0.883 0.185 0.947 1.242 0.796 0.393
Runoff mass/total mass
COD 0.701 0.886 0.811 0.504 0.833 0.907 0.808 0.895
BOD5 0.559 0.853 0.735 0.263 0.825 0.866 0.723 0.864
SS 0.699 0.950 0.873 0.654 0.873 0.973 0.877 0.849
Shape parameter “a”
COD 0.787 0.665 0.838 0.750 0.478 0.476 0.728 0.756
BOD5 0.783 0.650 0.958 0.761 0.408 0.572 0.843 1.000
SS 0.320 0.429 0.931 0.648 0.373 0.432 0.674 0.589

hydraulics data, such as mean and maximum dis- 4.2. Data analysis
charges, and total volumes involved in each event are
included. Runoff volumes have been calculated as the The data obtained may be used on a comparative
total volume minus the base wastewater flow. The per- basis, either by simply comparing the mean values of
cent of the pollutant mass which is associated with each catchment, or by using probability functions that
runoff processes has also been calculated by subtract- can fit the whole series of data. This technique seems
ing the mass corresponding to dry weather conditions to be more reliable, but we have also done a simple,
in the same catchment and in the same climatic con- preliminary analysis based upon mean values, which is
ditions at the same hour and day of the week (i.e. presented in Table 8. BOD values are not available for
dry weather conditions for a Friday in summer at the Barcelona catchment.
13:00). The main points to be highlighted after an analysis
No BOD values were measured in the Barcelona of Table 8 are as follows:
catchment, so the data presented for this catchment only • Even though the Madrid catchment has the largest
include COD and SS. surface area, higher discharges and volumes are
212 J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219

Table 6 which, in both cases, is extremely flat with huge sec-


Relevant parameters obtained for Valencia—Malvarrosa catchment tions designed for the evacuation of floods. When
(four events)
low discharges circulate along such large channels
Event 1 2 3 4 (rectangular boxes over 6 m wide), the water veloc-
Qmax (m3 /s) 1.351 0.871 7.901
Qmean (m3 /s) 1.346 0.557 4.542
ity is quite slow and sedimentation may occur. High
discharges promote the removal of these sediments,
Volume (m3 )
and high surface load values are recorded. A separa-
Total volume 2428 4151 30773
Runoff volume 2204 3207 29884 tive sewer system would probably avoid most of the
Runoff volume/total volume 0.908 0.773 0.971 pollution in these catchments, as the optimum sizes
Maximum concentration (mg/l)
for a wastewater sewer system and a rainwater sewer
COD 356 427 967 525 system have too many differences to work properly
DOB5 260 273 505 325 as a combined system. The maximum flux, which
SS 148 117 1167 812 is more difficult to analyze, also shows the highest
EMC (mg/L) values for Barcelona’s catchment.
COD 287 422 171 • Most of the mass measured during a rainfall event
BOD5 194 199 105 is associated with runoff. This pollution may come
SS 96 480 112
from the catchment or the sewer system, which
Surface loads (g/m2 ) would seem to be the main origin of the pollution
COD 0.709 1.570 5.480 measured in Barcelona. Catchments with high rain-
BOD5 0.473 0.648 3.341
SS 0.222 1.894 3.521
fall intensities seem to have higher runoff percents.
The values from Valencia are not reliable because of
Mass discharges (kg/s)
the irrigation inflows.
COD 0.448 0.788 1.266
BOD5 0.327 0.399 0.855 • A slight first-flush was noticed in every catch-
SS 0.181 0.966 0.750 ment, but in quantitative terms, a device which only
Runoff mass/total mass
recorded the first part of the hydrograph would not
COD 0.917 0.807 0.939 show a good enough efficiency. The mean value
BOD5 0.905 0.705 0.929 (0.76) means that the pollutant mass which is re-
SS 0.856 0.855 0.921 moved if the first 30% of the runoff is stored (which
Shape parameter “a” is quite a large amount of water) is less than 50% of
COD 1.048 0.555 0.989 the total mass. These data are in keeping with the re-
BOD5 1.094 0.528 0.920 search developed by CLABSA in Barcelona, whose
SS 1.340 0.513 0.834
tanks are big enough to promote the decantation of
solids, and also with the current trends in the north
observed in Barcelona and Valencia, due to their of Spain, where tank volumes are over 40 m3 /ha.
Mediterranean climate, with high rainfall intensities.
Moreover, the Valencia and Barcelona catchments A preliminary statistical analysis of the data has
are totally urban, while the Madrid catchment has a been done, including single plots for every catchment
rural area. Values for %runoff are clearly lower in and fully grouped data. The distribution of pollutant
Madrid. amounts, in terms of the parameters considered (C.
• The maximum concentration and EMC values for max, EMC, Unit mass, Flux. Max) have been plotted on
COD, BOD, and SS are very similar in Madrid, probability scales, corresponding to Lognormal distri-
Sevilla, Vitoria, and Barcelona (excluding BOD). bution. Some other distributions, such as Weibull, Nor-
Clearly lower values have been measured in Valen- mal, Logistic, and Exponential have also been tested.
cia, and this is due to the existence of a considerable Lognormal distribution provided the best approach
amount of parasite inflows from an irrigation system. with our data, in terms of the Anderson–Darling statis-
• The highest surface load values were found in tic, for both C. max and EMC values. As pointed out
Barcelona and Valencia. The most likely reason is by Butler and Davies (2000), most authors used this
the removal of pollutants from the sewer system, probability distribution.
J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219 213

Table 7
Relevant parameters obtained for Barcelona—Bac de Roda catchment (13 events)
Event 1 2 3 4 5 6 7 8 9 10 11 12 13
Qmax (m3 /s) 1.864 14.850 2.762 7.495 3.277 22.200 4.439 7.293 6.898 16.321 18.498 4.757 8.331
Qmean (m3 /s) 1.676 12.394 1.317 4.890 2.229 12.476 3.442 6.522 5.580 9.475 10.555 4.321 6.026
Volume (m3 )
Total volume 8876 90251 8448 28345 18425 43082 22706 28397 34477 110062 78365 14902 54025
Runoff volume 8120 89251 7835 27463 17350 42620 21824 27784 33587 108491 77046 14406 52706
Runoff volume/total volume 0.915 0.989 0.927 0.969 0.942 0.989 0.961 0.978 0.974 0.986 0.983 0.967 0.976
C. max (mg/L)
COD 711 2333 1960 1075 1300 1471 542 394 1202 1484 616 670 1996
SS 210 1178 2257 1100 1710 980 768 1233 2893 2300 1485 827 1702
EMC (mg/L)
COD 296 689 1393 606 330 190 434 139 444 214 128 446 615
SS 61 331 1074 514 483 703 397 349 883 1379 486 361 518
Surface loads (g/m2 )
COD 1.438 36.461 6.836 9.984 3.423 4.738 5.674 2.241 8.889 13.645 5.735 3.843 19.371
SS 0.266 17.477 5.295 8.500 5.160 17.782 5.236 5.792 17.839 89.137 22.293 3.129 16.370
Mass discharge (kg/s)
COD 1.185 19.906 3.920 4.486 2.926 5.602 2.373 2.105 6.975 12.329 2.420 3.154 16.595
SS 0.391 16.150 5.412 3.299 3.537 14.013 3.393 6.674 16.698 34.801 12.301 3.655 14.162
Runoff mass/total mass
COD 0.931 0.996 0.988 0.988 0.958 0.987 0.979 0.963 0.986 0.984 0.969 0.982 0.991
SS 0.834 0.996 0.992 0.993 0.986 0.998 0.988 0.993 0.997 0.999 0.996 0.989 0.994
Shape parameter “a”
COD 1.405 0.519 0.938 1.183 0.519 0.564 1.326 0.573 0.519 0.390 0.607 1.251 0.566
SS 0.680 0.584 0.727 2.010 0.584 0.878 0.997 0.469 0.584 0.879 0.613 0.908 0.485

The Anderson–Darling statistic is a measurement Fig. 5 represents the data and their adjustment lines
of how far the plot points fall from the fitted line in a by using the Lognormal distribution. The more grouped
probability plot. A smaller A–D statistic indicates that together the lines appear, the more homogeneous the
the distribution fits the data better (MINITAB, 2000). catchments are considered to be. The accuracy of each
For total mass per surface unit, better fits have been individual fit may be read by its A–D statistic value.
obtained using the Normal distribution, but there was Fig. 6 shows EMC for COD. The results are quite
not a big difference, so the same distribution was cho- closely grouped, considering that the x-range axis is
sen for all the parameters involved to make it easier to much narrower than the one in Fig. 5. Data from Va-
understand the set of graphs. The fitted lines have been lencia show the lowest values, as mentioned earlier.
included in every single case and the 95% confidence Fig. 7 shows the total mass per surface unit for COD.
intervals have also been used in the fully grouped data As these are total values, Valencia data are not lower
plots. The complete series of statistical plots are avail- than average. For the analysis of these data, morpho-
able. A set of single plots is included as an example. logic data have to be taken into account. The Madrid
The maximum concentrations for SS are shown in catchment, which is much larger than the others, is not
Fig. 5. As can be seen, all the catchments present a fully urbanized, and has a lower population density.
similar trend except Valencia (V, in the legend), whose Therefore, its pollution tax per surface unit was ex-
values are below average. This is a constant for every pected to be the lowest, and was observed accordingly.
pollutant, as observed in Table 8, owing to the existence No explanation has been found for the high values in
of parasite irrigation waters. Barcelona if we consider only its catchment, whose
214 J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219

Fig. 5. Maximum concentration values for suspended solids. Lognormal fit. Notation: Vi—Vitoria, M—Madrid, Se—Sevilla, V—Valencia,
B—Barcelona. Location and scale are the Lognormal distribution parameters. AD: Anderson–Darling statistic.

Fig. 6. EMC distribution for COD.


J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219 215

Fig. 7. Total mass per surface unit (COD). Lognormal fit.

characteristics are more or less the same as those of geneous cluster of points. Therefore, when analysing
Sevilla, Valencia, and Vitoria. The only point to be Fig. 7, the points are clearly separated from each other.
mentioned is the existence of public works in the catch- So, it would seem that a general curve might make some
ments, which may lead to larger amounts of suspended sense (practically speaking, at least) if the EMC and C.
inorganic solids and some metals, whose values are max values are considered, as this general curve would
very high. The influence of these works on COD values be not useful for the total mass or for maximum flux.
is not very clear. So, it may be considered that this pol- EMC and C. max curves are shown in Figs. 8 and 9
lution has been removed from the sewer system, which for the full set of catchments except Valencia, whose
has been accumulating sedimentation during dry peri- values are considered to be affected by parasite inflows.
ods. BOD values are not shown as there are no values for
It makes no sense to show a grouped plot for the the Barcelona catchment.
total mass per surface unit, as the morphologic char- The main conclusions to be drawn from the grouped
acteristics of the catchment have a considerable influ- probability functions are that they offer the most useful
ence. Maximum flux depends on the size, so it cannot percentiles for EMC and C. max. Table 9 summarizes
be extrapolated either. these percentiles.
In spite of the fact that EMC values and C. max val- As clearly seen here, very high values were obtained
ues are also related to catchment characteristics, these not only for the maximum concentration, which may be
are the values that are usually used as a basis for com- considered an extreme value, but also for EMC, which
parison. represents the mean value of the pollution during the
When analysing Figs. 5 and 6 (C. max and EMC), full event.
the points corresponding to different catchments are COD amounts of roughly 500 mg/l during the mean
not clearly separated, but they are relatively intermixed, event mean that Spanish catchments cannot be de-
which means that if only one fit were done for the full signed with standards such as the ATV-128, whose
set of data, no grouping into five packs of data would pollution CSO values are around 200 mg/l, in terms of
be observed, but rather there would be a single homo- COD. The capacity of the structures designed to avoid
216 J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219

Fig. 8. EMC values for COD and SS for Madrid, Barcelona, Sevilla and Vitoria. Log, normal fit, including confidence intervals (95%).
J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219 217

CSO events must be able to ensure that very few events during a rainfall event. In any case, further research is
take place or there must be an improvement in the clean- necessary from a practical point of view, in order to
ing policy of the sewer network and the streets in order establish a standard which would be considered useful
to reduce the amount of pollution that can be removed for Spanish catchments.

Fig. 9. C. max values for COD and SS for Madrid, Barcelona, Sevilla and Vitoria. Log, normal fit, including confidence intervals (95%).
218 J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219

Table 8
Mean values of the parameters considered
City Madrid Vitoria Sevilla Barcelona Valencia
Summary (mean values)
Qmax (m3 /s) 3.336 1.337 0.600 9.153 3.374
Qmean (m3 /s) 2.818 0.677 0.426 6.223 2.148
Volume (m3 )
Total volume 17005 2994 2752 41566 12450
Runoff volume 7874 2303 2083 40653 11765
Runoff volume/total volume 0.485 0.743 0.682 0.966 0.884
Maximum concentration (mg/L)
COD 1344 1396 1717 1212 569
BOD5 792 521 759 341
SS 1189 966 1722 1434 561
EMC (mg/L)
COD 680 671 834 456 293
BOD5 384 324 389 166
SS 597 421 733 578 229
Surface loads (g/m2 )
COD 0.201 0.952 1.297 9.406 2.586
BOD5 0.106 0.419 0.546 1.488
SS 0.208 0.674 1.283 16.48 1.879
Mass discharges (kg/s)
COD 3.616 1.040 0.501 6.46 0.834
BOD5 2.106 0.494 0.206 0.527
SS 3.425 0.724 0.558 10.35 0.632
Runoff mass/total mass
COD 0.703 0.793 0.820 0.977 0.887
BOD5 0.675 0.711 0.788 0.846
SS 0.823 0.844 0.896 0.981 0.877
Shape parameter “a”
COD 0.738 0.685 0.775 0.797 0.864
BOD5 0.727 0.747 0.765 0.847
SS 0.702 0.550 0.743 0.800 0.896

5. Conclusions

Table 9
The main topic of the research project presented in
Percentiles obtained from the lognormal fit, for COD and SS this paper was the determination of pollutant discharges
Percentile
during CSO events in several experimental catchments
in Spain, with the purpose of obtaining a pollution pat-
50% 90% 95% tern that may be applied to other catchments in Spain.
EMC (mg/l)
The existence or non-existence of a first-flush was also
COD 523 1030 1060 examined.
SS 484 1020 1060 The main conclusions are:
C. max (mg/l)
COD 1199 2600 3300 • The values measured are clearly higher than those
SS 1130 2700 3300 generally reported in the literature (i.e. ATV-128).
J. Suárez, J. Puertas / Ecological Engineering 24 (2005) 201–219 219

A more effective cleaning policy or the construction Bertrand-Krajewski, J.L., Chebbo, G., Saget, A., 1998. Distribution
of separative systems should modify these data, as it of pollutant mass vs. volume in stormwater discharges and the
would seem that the main source of pollution can be first flush phenomenon. Water Res. 32 (8), 2341–2350.
Butler, D., Davies, J.W., 2000. Urban Drainage, E&FN SPON, Lon-
located in the sewer system, at least in flat, Mediter- don, 489 pp.
ranean cities, such as Barcelona and Valencia. Ellis, J.B., 1989. Urban Discharges and Receiving Water Quality
• No severe first-flush has been reported, so first-flush Impacts. Pergamon Press, Oxford, Adv. Water Pollut. Control
prevention structures have a moderate utility, as sub- No. 7.
stantial pollution loads are released. Ellis, J.B., 1991. Measures for control and treatment of urban runoff
quality, Rep. DT3PL/FV/JB, Agence de l’eau Seine-Normandie,
• Some general values of CSO pollution in terms of Paris.
COD, BOD, and SS may be accepted as an order Gupta, K., Saul, A.J., 1996. Suspended solids in combined sewer
of magnitude for Spanish catchments, presented in flows. Water Sci. Tech. 9, 93–99.
Section 4. Lager, J.A., et al., 1977. Urban stormwater management and tech-
nology: Update and users guide. EPA-600/8-77-014, Cincinnati,
OH.
Lijklema, L., Tyson, J.M., Lesouef, A., 1993. Interactions between
Acknowledgements sewers, treatment plants and receiving waters in urban areas:
a summary of the INTERURBA’92 WORKSHOP conclusions.
The project which is partially presented in this pa- Water Sci. Tech. 27 (12), 1–29.
per was sponsored by the Spanish Ministry of Envi- Metcalf, Eddy, 1991. Wastewater Engineering. Treatment, Disposal.
Reuse, third ed. McGraw-Hill International Editions, Civil En-
ronmental Protection (Ministerio de Medio Ambiente), gineering Series, ISBN 0-07-100824-1.
and coordinated by Mr. Angel Cagigas and Mr. José MINITAB, 2000. Statistical Software Release 13. MINITAB Inc.
Ramón Barro, in collaboration with the GADU group NWRW, 1991. Final Report of the 1982-1989 NWRW (National
of AEAS, CLABSA, the municipalities of Madrid, Working Party on Sewerage and Water Quality). Foundation for
Sevilla, Vitoria, and Valencia. The project was devel- Applied Waste Water Research, the Dutch Ministry of Housing,
Physical Planning and environment, 63 pp.
oped by the firm Infraestructura & Ecologı́a, managed Saget, A., Chebbo, G., Bertrand-Krajewski, J.L., 1996. The first flush
by Mr. Luis Ortega, with the collaboration of the Uni- in sewer systems. Water Sci. Tech. 9, 101–108.
versity of A. Coruña. US-EPA, 1983. Results of the Nationwide Urban Runoff Program,
vol. I, Final Report, NTIS PB84-185552, Water Planning Divi-
sion, Washington, DC.
Valiron, F., Tabuchi, J.P., 1992. Maitrise de la pollution urbaine par
References temps de pluie. État de l’art., AGH-TM, Tec-doc, Parı́s, ISBN
2-85206-863-X.
ATV-128, 1992. Standards for the dimensioning and Design of Water Research Centre, 1987. Sewerage detention tanks—a design
Stormwater Structures in Combined Sewers. guide, WRc, Swindon.

You might also like