Stofkoncentrationer i regnbetingede udledninger fra fællessystemer

4. Concentration levels reported in international investigations

4.1 Dutch study in Loenen, etc., 1986
4.2 German study in Munich-Harlaching, etc., 1977
4.3 Collation study from United Kingdom, 1985
4.4 Canadian study in Ontario, 1980ties
4.5 French measurement programmes
4.6 Norwegian study in Oslo, 1974
4.7 Measurement programs in the Unites States

The studies presented in this paragraph are both studies from individual sites and comparative studies. They present large variations with regard to experimental procedures, to what extent findings have been related to catchment characteristics and the level of detail reported.

In chapter 5.2 the statistical distribution of the data is analyzed whereas chapter 5 relates the SMC-values to catchment characteristics.

4.1 Dutch study in Loenen, etc., 1986

The study commenced in 1981 with the aim of looking into storm water discharges from sewer systems (Onderdelinden and Timmer, 1986). The fact that this is a Dutch study makes it interesting, as the average sewer system in the Netherlands is flat and often with some infiltration water due to the high groundwater levels. This results in relatively high dry weather flows (DWF). The small hydraulic gradients make the systems prone to depositing of sediments. Relatively large volumes of fine sediments in the systems can potentially be flushed out during rain events, but this is somewhat counterbalanced by the low mean slope of the sewer systems which does not entail high bed shear levels.

Six drainage areas with typical Dutch sewer systems were monitored. The four combined sewer systems are located at Loenen, Oosterhout, Bodegraven, and Kerkrade. Table 4.1 shows the main characteristics of these sites. The gradients in the systems vary between 1:500 and 1:1000 (1-2 0/00) and all chosen systems have relatively large diameter sewers and only one CSO structure connected.

Some general conclusions were reached. The pollution emissions depend on rainfall intensity and overflow discharges, i.e. extreme loads of pollutants occur when overflow discharges are large. Contrary to expectations there were no correlation with the average dry weather period (ADWP). One explanation is that the sewer system is not flushed clean each time an overflow occurs.

Another subject treated within the project were the correlation between different measured pollutant parameters. Table 4.2 shows computed correlation between parameters.

As seen in table 4.2 there are rather high correlations between the shown parameters. This means that the concept of using one or a few parameters such as COD as overall pollutant indicators is quite valid in an on-line situation. The level of the different pollutant parameters needs, however, to be measured before the correlations can be used for predicting e.g. BOD from measurements of COD. Results obtained from the four combined systems are seen in table 4.3.

4.2 German study in Munich-Harlaching, etc., 1977

This study has been presented by Geiger (1986b) and besides presenting results from the measuring site of Munich-Harlaching reference is also made to several other studies. The field data was collected during the period 1977 to 1981. Munich-Harlaching is a rather big catchment with a drainage area of 540 hectares. This should result in less pronounced first flush due to the long concentration time, i.e. that all the sediments and water does not arrive at the CSO at the same time. The catchment characteristics are seen in table 4.4. Note the difference between the average surface inclination (1.7%) and the average sewer slope (0.5%). Such differences underline the importance of checking information carefully when correlating runoff concentrations with catchment characteristics.

Some main conclusions were reached based on the study (Geiger, 1986b). For the catchment of Munich-Harlaching first flush seems to depend on the level of the DWF whereas correlation with ADWP was not found. The basis for the conclusion with respect to the DWF is calculated contributions of DWF to the storm water run-off. In this case the DWF contributes with 23% to the run-off volume, 32% to TSS, 31% to BOD₅, 44% to COD, 53% to N_Kj, and 43% to P_tot.

Some flush effects were seen for the events especially for TSS and to some degree for COD whereas other pollutant parameters showed dilution effects. The flush effects were most apparent for night events where the DWF is small.

I data.n connection with the investigation presented by Geiger (1986b) a literature search resulted in mean concentrations (not EMC's) for combined sewers, see table 4.5. Unfortunately, attaining catchment areas, etc. were not reported, neither the original source of the

The first conclusion to be drawn on table 4.5 is that the mean values exhibit large variations. Some of the discrepancies can of course arise from differences in measuring procedures and methods and calculation methods, cf. chapter 3, but a large part of the variation must be explained by differing rainfall and catchment characteristics at each site.

As can be noticed in table 4.5 the values from the Munich-Harlaching catchment are not included. The investigations in Munich-Harlahing consists of a very high number of samples taken which have been used for calculation of mean concentrations and station mean concentrations. Table 4.6 shows some of the obtained results.

When combined run-off concentrations and load figures from Münich-Harlaching was plotted it demonstrated that the values were lognormal distributed. This distribution has frequently been shown to be applicable (VanBuren et al, 1997), though it did not fit distributions of dissolved solids, chloride, sulphates, and COD.

4.3 Collation study from United Kingdom, 1985

This reference (Ellis, 1986) does not present one specific investigation but presents pollution aspects concerning urban runoff and EMC-values condensed from mainly UK sources.

It is evident that treatment of the dry weather flow (DWF) from combined sewers is not sufficient for a sustainable receiving water quality. It is estimated that 35% of the annual pollutant load in the United Kingdom originates from CSO’s, that operates only 2-3% of the time. Comparing annual water volumes the dry weather runoff approximately equals the stormwater runoff. A summary of encountered values of mean pollutant concentrations is reproduced in table 4.7.

Table 4.7   Look here!

As seen (table 4.7) the values are given in ranges indicating the variation of EMC's. It is interesting in this case to see the attempt to discriminate between different pollution sources. It can be seen that the pollution load from CSO is significantly higher than from the storm sewer outlets.

4.4 Canadian study in Ontario, 1980ties

An interesting comparison is made between raw sewage, treated sewage, surface runoff, and combined sewer overflow. This comparison can be seen in Table 4.8. Data originates from Ontario, Canada and are rendered by Waller and Hart (1986).

Mean concentrations of TSS from combined sewer overflows are rather high (190 mg/l). This is explained by the fact that combined sewer systems often serve older urban areas with higher population density. Separate and combined sewer systems are compared (Waller and Hart, 1986) and according to this comparison the combined system discharges about twice the loads of the separate system, also taking into account the discharges from the WWTP. This seems to be valid for TSS, P_tot, N_tot, and chlorides. However, such conclusions depend on e.g. the storage volume and the interceptor capacity of the catchment and such information was not given.

4.5 French measurement programmes

In this reference (Hémain, 1986) a national French study containing four sites is presented. The French results originate from four catchments (Maurepas, Les Ulis, Aix Zup, and Aix Nord) with relatively long and consistent measurements, (Hémain, 1986). A summary of the results is shown in Table 4.9. Data can also be seen in (Deutsch and Hémain, 1984).

4.5.2 Study in "Le Marais", 1998

The Le Marais catchment is a monitored catchment, which has been thoroughly investigated for a number of years (Gromaire-Mertz et al, 1998a,b). As for other studies it was revealed that the mean concentrations are not the same at different levels in the catchment. The catchment characteristics are shown in Table 4.10 and Table 4.11 shows results for TSS, COD, BOD, and VSS. The EMC-values for roof, yard, and street are calculated as means of EMC for a number of events and the ranges are the result of different roof, yard and street types. Since there was only one outlet from the studied catchment no range is given here.

The different surfaces (roofs, yards, etc.) do not contribute to the total load with the same percentage from event to event, though pollution originating from resuspension of sewer sediments is rather constant 45 - 66%. Waste water contributes with 6 – 37%, roofs 3 – 23%, yards 3 – 10% and streets 10 –17%. The findings equal those of other French studies according to (Gromaire-Mertz et al, 1998a).

4.6 Norwegian study in Oslo, 1974

This study was initiated in 1974 and is comprised of seven field sites, three combined systems and four separate sewer systems (Lindholm and Balmér, 1978). The objective of this investigation was to estimate annual levels of pollution discharges and in what way this was influenced by

• the sewage system,
• different degrees of urbanization,
• antecedent dry weather period,
• time from the start of the rain event, and
• runoff intensity.

The study did, however, not show clear correlations, but it did show that combined sewers for most pollutants contributes to the annual load with larger annual loads than separate sewers. Catchment characteristics can be seen in (Lindholm and Balmér, 1978). Table 4.12 renders the results for combined sewers.

Table 4.12
Average concentrations in storm runoff, from (Lindholm and Balmér, 1978).

One of the inherent problems of measuring campaigns like this one is whether or not the monitored rain events constitute a suitable representation of the annual precipitation even though the whole year is not monitored. In this case it was noticed that one of the monitored rain events increased the annual loads by a factor of 2-3 as both concentrations and overflow volume were exceptionally large. This was a rain event with a ten-year return period.

4.7 Measurement programs in the Unites States

The United States Nationwide Urban Runoff Program (NURP) was initiated in 1978 by the environmental protection agency. The final report from the investigations was published in 1983 (USEPA, 1983).

The aim of this five year program was to assess the properties of urban runoff pollution to water quality across the United Sates. A range of pollutants were monitored with good consistency at almost all the catchments. The pollutant parameters were: Suspended solids, chemical oxygen demand, biological oxygen demand, total nitrogen, ammonia, total phosphorus, copper, lead, and zinc. The NURP program developed EMC's for these and other pollutants, drawing upon data collected from over 2,300 events at more than 81 sampling stations located in 28 different metropolitan areas. The catchments attaining characteristics were determined by land-use, drainage area, population density, imperviousness, and runoff-coefficient, cf. Table 4.13.

The overall results from the NURP study are shown in Table 4.14 grouped by sampling station; they represent the statistical properties of all the SMC-values found within the NURP program. Between one and 37 EMC values were used to calculate the SMC values for different pollutant parameters. To get an impression of the variation between sampling stations Figure 4.1 shows the 37 SMC-values of COD for residential area as a bar graph. The mean value of the SMC-values shown is 102.4 mg/l. This value is very close to the guideline value of COD stipulated by the U.S. EPA for planning-level calculations (100 mg/l).

Table 4.14   Look here!

Figure 4.1    Look here!
SMC-values for COD from the NURP program (USEPA, 1983).

The data in Table 4.14. is grouped by land-use category (residential, industrial, mixed, commercial and rural). However, the NURP study concluded that the variance of the EMC's when data are grouped by land-use or geographic region is so great that the groups cannot be distinguished statistically. In some cases investigators have attempted to differentiate between land-use categories by applying the NURP data computed by land use, despite the NURP conclusions to the contrary.

It is not clear from the original project report (USEPA, 1983) whether it is concerned with separated or combined sewer systems and contact was thus established with American specialists in the field. Pitt (1999) states that all the catchments included in the NURP database are served by separate systems, however with substantial influence of illicit wastewater connections. Thus data from the NURP database is included in the following to provide a basis for comparison, however sites with less than 50 mg COD/l are excluded from the analysis. American data from combined sewer systems has not been collected and analyzed in a similar systematic way.

Several measurement programs have been conducted in the US since the NURP studies stopped and an effort to compile all data into one common database was recently started by the company Camp Dresser and Mckee (Smullen and Cave, 1998). The database includes data from the following sources:

• The NURP database with more than 2,300 events at more than 81 sampling stations located in 28 different metropolitan areas.

• The USGS storm runoff database which includes data collected through the mid-1980'ies for over 1,100 station-storms at more than 97 urban sites located in 21 metropolitan areas, with only 5 stations common to the NURP data set.
• Data collected by over 30 cities as part of their National Pollution Discharge Elimination System (NPDES) discharge permit applications (816 events in total).
• Data from other major monitoring programs.

Only data from monitoring locations that exhibited little or no base flow and that did not include any stormwater management control practices were allowed in the database. As for the NURP data, this means that it is not clear which data comes from combined sewer systems.

Table 4.15 shows the major findings from comparing the new, pooled database with results from only the NURP data. The differences between the pooled means and those estimated from the NURP data range from a 79% lower estimate for Copper to a 36% higher estimate for BOD. The lower metal concentrations in the newer data could be the result of sampling problems (e.g. poor solids recovery), or they may reflect the use of cleaner technologies that are more prevalent in the 1990'ies for sampling, chain of custody transfers and laboratory methods for metals. These cleaner technologies have evolved over the past decades in response to the criticism of metal determination from previous decades (Smullen and Cave, 1998).

The differences between the very large US data sets illustrate once again that SMC's and EMS's are subject to large variations and that such concentrations should be interpreted with care.