COMBINED SEWER OVERFLOW ASSESSMENT IN SLOVAKIA

Daniel Sztruhar, Marek Sokac, Stefan Stanko, Adrian Holiencin, *Andrej Markovic, *Ingrid Labska

Slovak Technical University, Department of Sanitary Engineering, Radlinského 11, 813 68 Bratislava, Slovakia

*Water Research Institute, nabr. a.g. L. Svobodu 5, 812 49 Bratislava, Slovakia

INTRODUCTION

In the majority of towns in Slovakia combined sewer systems prevail. Because waste water treatment facilities are usually designed to handle only dry weather flows but not storm water runoff it is necessary to provide diversion or overflow structures within the sewer systems. Combined sewer overflows (CSO) are designed to reliese some part of storm discharges during heavy rainfalls into the receiving waters or to storm tanks. An inventory and structural evaluation of CSOs in towns with more than 30,000 inhabitants was carried out during a three years joint research (1996-1999) of the Water Research Institute and the Department of Sanitary Engineering of the Slovak Technical University in Bratislava (Markovic, 1999). As part of the research, detailed monitoring of eight storm events in four towns was carried out. Waste water samples were taken and analysed for common constituents and storm flow and pollution transport modelling was carried out by MOUSE (Sokac and Markovic, 2000). Also, recommendations for the use of limiting rainfall for CSO design have been worked out (Rusnak and Stanko, 2000). This paper describes results of the CSO structural evaluation and wet weather monitoring program.

CSO CATEGORISATION

Under the umbrella of state funded water authorities over 700 CSOs are registered. The most common types of CSO structures are the following:

It was the aim of the research project to provide detailed information on the existing CSOs, including description of their structural condition. The results of evaluation of general condition and structural defects of CSOs are summarised below:

• Several CSOs were unreasonably oversized; thus they can not function as overflows. The most common defects of this type were seen at CSOs with one side weir, where structures with over 50 m weir in length have been found. Using two side weirs could have lead to reduction of weir length by 40%.
• On the other hand, several CSOs are overloaded even during dry weather flows and raw sewage is released into the receiving waters. To overcome this defect in many places simple repairs were realised: weir heights have been raised by folders without any calculation.
• Another common reason of frequent overflows is extreme sedimentation in the CSO chamber or just upstream of it. In some places over 50 cm of sediments were measured, most commonly in CSOs with transverse weirs.
• Poor design has been found to be a common reason of wrong hydraulic conditions of many CSOs. One of the most frequently observed defects were upstream pipes laid in steep slopes leading to supercritical flows upstream of the weir and in most cases to overflows of raw sewage. Transverse weirs can be typically grouped into this category of defects.
• A next category of defects was made up by joining several inflow pipes in one CSO chamber. This poor design makes it impossible to calculate and/or measure discharges at the CSO due to turbulent flows at the chamber. Moreover, these chambers have been found covered with thick layers of sediments and/or gravel.
• Sedimentation, except were caused by poor design, was found to be a common defect of mainly large CSOs situated on large sewers. Frequent cleaning of such CSOs is necessary which contributes to higher operational costs.
• The most frequently observed problem of transverse weir CSOs was found blockage of outflow pipes by floating materials (pieces of wood, polystyrene, cans, etc.). There is no universal recipe to overcome this problem; pipe replacement seems to be the only alternative to avoid blockage, increase capacity and decrease excess overflows.
• The last group of problems was flooding of CSOs during high water levels in the receiving waters. To overcome this problem tide gates were installed at some CSOs.

The above findings demonstrate the most frequently observed defects at combined sewer overflow structures in Slovakia. It has to be stated, however, that the structural condition of the majority of overflows were found in satisfactory state and in over 80% of CSOs there were not reported any noticeable problems.

WET WEATHER MONITORING

The objective of the wet weather monitoring program was to collect rainfall, flow and concentration data, which could be used to calculate CSO volumes and pollutant loads as well as for calibration of mathematical models. Three locations were selected for the field campaign, Liptovský Mikuláš, Lučenec and Trnava. The selected sewer systems represent different sizes and geographic conditions (varying from hilly regions to typical low-land regions).

Rainfall was measured by an ISCO tipping-bucket raingauge located in the towns’ centres. Measurements were maid in different seasons of the year (spring, summer, autumn), thus, we recorded rainfalls with different peak intensities and duration. Typical return periods of measured rainfalls were one year or lower.

Flow levels were measured by ISCO ultrasonic flowmeters located at sewer manholes upstream of CSO chambers. Special attention was taken to avoid improper installation of the ultrasonic probes (e.g. in sections with backwater effects, sections with relevant curvature or places with large amount of sediments, were calibration could have lead to incorrect calculated flows). All instruments were equipped with data loggers, which were fully synchronised and set for 1-minute recording intervals.

Wastewater samples were taken manually; automatic sampling was ruled out due to potential vandalism and pertaining difficulties with dual synchronisation to both raingauge and flowmeter. Sampling intervals varied according to the flow rate; at the start of the storm, samples were collected every 1 to 5 minutes to characterise any potential first flush of organic constituents, but as the storm progressed, the sampling interval was increased to 15 minutes. During each time interval only one sample with a volume of 2L was collected. All analyses were carried out at the Water Research Institute according to valid Slovak standards for water quality analyses.

Altogether six rainfall events were monitored during the recent research and two additional events were added to the data from an earlier study carried out in the Town of Malacky (Sztruhar et al., 1997). Thus the current data base contains eight storm events. Waste water samples were analysed for common constituents and are summarised in Tab. 1

Tab.1 Parameters of pollution analysed in waste water samples

DATA ANALYSIS

The primary reason of data collection and analysis in urban runoff monitoring programs is to provide general data for pollution loads from combined sewer systems. One of the priorities in the current study was set, therefore, to provide event mean concentrations (EMC) for annual pollution load calculations from sewer systems in Slovakia and in other regions in Central and Eastern Europe with similar urban drainage practices (e.g. Czech Republic, Poland, Hungary, Slovenia, etc.).

A. Statistical characterisation of urban runoff quality

Data analysis has been guided by the results of the US EPA Nationwide Urban Runoff Program (NURP) (US EPA, 1983). From the general findings of this project we followed the four main and still valid points:

1. In most cases the total runoff load is more important than individual concentrations of particular parameters. This is due to the relatively short duration of runoff (overflows) followed by relevant mixing in the receiving waters. The behaviour of the receiving waters is then rather a response to the total loads rather than to variable concentrations within each particular event. Therefore, the NURP studies focused on evaluating EMCs defined as the ratio of mass of pollutant contained in the runoff event (kg) and the total volume of flow in the event (m³):

   (1)

   where

   EMC	event mean concentration of a particular pollutant (kg/m3)
   Qi	discharge during time interval i (m3/min)
   ci	concentration of pollutant during time interval i (kg/m3)
   delta ti	length of time interval i (min).

2. The evaluation of the NURP database has revealed that the probability distribution of EMCs for most of the constituents follows a log-normal probability distribution. In other words, the logarithmic transformed series appear to be symmetrical and normally distributed. The NURP study found, that EMCs for TSS, P_tot, TKN, Pb and Zn are extremely well represented by the log-normal distribution. COD and nitrite-nitrate are well represented by this distribution and for other constituents, including BOD, soluble phosphorus and copper the log-normal distribution can not be rejected. The NURP study also showed that EMCs are weakly correlated to runoff volumes.

The mean concentration of the logarithmic transformed series (mg/l) can be calculated after Marsalek (1990) as follows

(2)

where

m	mean concentration of the logarithmic transformed series
s	standard deviation of transformed series.

The log-normal distribution is completely defined by it‘s mean and standard deviation of the transformed logarithmic series.

The establishment of log-normal distribution for major constituents of urban runoff has a number of benefits:

• Concise summaries of variable data can be developed.
• Mean values can be obtained more precisely even from highly variable data bases.
• Comparison of results from different site and/or events can be made.
• The log-normal distribution provides a base for probabilistic modelling of urban runoff quality which allows to calculate concentration return periods and exceedence values for water quality and/or ecological criteria..

3. The NURP program concluded, that in a statistical sense there were no first flush effects observed in the studied sewer systems. It has to be added to the above information, that the NURP study dealt with separate sewer systems while in Europe (and in Slovakia) the prevailing sewer system is the combined system. However, later studies (Ellis, 1986; Saget et al., 1996) confirmed that first flush effects are rare in combined systems, too.

During the evaluation of measured data attention was devoted to the following main areas:

• confirmation of existence or non-existence of log-normal distribution of EMCs, and
• confirmation of existence or non-existence of first flush of organic pollutant in the overflowing water.

EMCs for each event were calculated using Eq. (1). We obtained eight different values of mean concentrations. We ranked the values in ascending order and plotted them on a probability paper. The resulting probability plots of TSS, COD, N_tot and P_tot are in Figs.1, 2, 3 and 4, respectively. Log-normal distribution was found to be relevant for all examined parameters (BOD5, COD, TSS, N_tot, N-NO₃, N-NO₂, N-NH₄, P_tot, P-PO₄). Statistical distribution of heavy metals could not be tested due to their extremely low concentrations (just at the level or below detection limits). EMCs calculated from the logarithmic transformed data are listed in tab.2.

Fig. 1 Cumulative probability distribution TSS

Fig. 2 Cumulative probability distribution COD

Fig. 3 Cumulative probability distribution N_tot

Fig. 4 Cumulative probability distribution P_tot

Tab.2 Event mean concentrations of urban storm runoff parameters (combined sewers) in Slovakia

*CFU - colony forming units

Similar to the NURP study, calculated EMCs were found to be independent from rainfall intensity and duration, geographic location, size and slope of sewer systems and runoff volumes. A comparison with the original NURP data also reveals, that higher COD, BOD₅ and TSS values were obtained during this study. This could be explained by extensive sedimentation in our sewers and by their transport during storm events (see below for detailed analysis).

B. First flush

The existence or non-existence of first flush in urban storm runoff has still not been satisfactorily answered. Ellis (1986) based on a wide literature search stated that first flush is not a consistent feature of either separate or combined sewer systems. Probably each sewer system reacts individually to each storm event. Generally, the occurrence of first flush is more frequent in smaller storm sewer systems with short concentration times and without noticeable amounts of sediments. In combined sewer systems the question of first flush is more complicated. In general, there are three sources of pollution in combined sewer systems. Firstly, dry weather discharges, secondly, elevated storm discharges during wet weather, and finally, scoured bottom sediments. All three sources significantly contribute to the overall pollution of water. Near bed solids are believed to be a major source of continuous organic pollution during storm flows (Arthur and Ashley, 1998). This material has a significant portion of dry weather flow COD and BOD concentrations, whilst it only contains a modest amount of TSS. Many researchers believe that if there are no local data available, any strategy or remedial action based on the assumption of first flush could be misleading. Guidelines ATV A 128 (ATV, 1992) assume, that first flush effects are relevant on small watersheds with times of concentrations shorter than 15 minutes. In larger combined systems first flush effect is levelled out due to long transportation times and continues sediment transport.

If we define first flush with the ratio 80:30 (that is 80% of pollution is transported within the first 30% of runoff volumes), than first flush was not observed in any of the events. A milder criterion of 60:30 would lead to one observed first flush out of the eight measured events. Fig. 5 shows a typical event, where first flush has not been observed.

Fig. 5 Typical diagram of cumulative pollution transport without observed first flush effect

C. Detailed analysis of VSS as an indicator of organic content of TSS

Since many of the parameters in urban runoff are associated with suspended solids, the next step in data analysis was to compare the organic portion of suspended solids (approximately represented by VSS - volatile suspended solids) with the total load of suspended solids (expressed as TSS - total suspended solids). Such comparison could answer the nature of sediment transport during storm flows. Fig. 6 shows typical VSS and TSS concentrations during dry weather periods. As seen from the figure, VSS concentrations are well below 50 mg/L (except one peak), while TSS concentrations vary from low to almost 500 mg/L. During a storm event (Fig. 7), which started after a ca. two weeks dry period, the absolute concentrations reached 20 to 30 times higher values compared to those measured in dry weather flows. It has to be added, that this particular overflow commenced during the morning minimum discharges at 5:30. A next storm event starting the same day shortly after the first one ended did not produce higher concentrations (Fig. 8).

A look at the ratio of VSS to TSS reveals, that it varied between 10 to 20% (Fig. 9) during the dry weather period. This implies that the upper layers of sewer sediments (rich in organic material) are removed either during dry weather periods or during the first flush of storm water towards the waste water treatment plant. Sediments removed during the following periods contained as low as 10% organic materials. During the overflow the organic content started to raise and peaked at the level of 40-50% of TSS concentrations (Fig. 10). This phenomenon could be expressed by mobilisation of the deeper layers of sediments rich in faecal materials where aerobic/anaerobic decomposition processes were in progress. Flush of anaerobic decomposition products were confirmed in the Malacky study earlier (Sztruhar et al., 1997).

What is remarkable in the comparison of the ratio of VSS/TSS is that it had an increasing tendency up to 40 - 45%. This ratio did not increase during the second storm event (Fig. 11). This means, that during this particular event there was a maximum ratio of 40 - 45% of organic content in the suspended solids, which remained relatively constant during the whole storm event. Based on the above analysis we can state that the hypothesis of first flush has not been confirmed in the overflowing water due to continues mobilisation and transport of deeper horizons of sewer sediments.

CONCLUSIONS

• A long-term urban drainage monitoring study has been carried out in 1996-1999 in Slovakia. As a result of the joint effort of the Water Research Institute and the Department of Sanitary Engineering of STU Bratislava over 300 CSO chambers have been visited and evaluated. A catalogue of CSOs has been produced containing all relevant structural data about the chambers.

• Furthermore, waste water samples have been taken and analysed and EMCs were produced from eight monitored storm events carried out in four combined sewer systems. These values represent a first ever data base produced in Slovakia and are recommended to use in any further urban drainage studies and load calculations. Having in mind similarities between sewer systems and urban drainage practices in the neighbouring countries, the presented database could be of use in other Central European states, such as Poland, Hungary or the Czech Republic.

• EMCs of organic pollution and suspended solids are higher compared to other, previously published data. Other constituents were comparable with literature data.

• Based on a statistical analysis of data we can state that the majority of EMCs follow a log-normal probability distribution.

• First flush of organic material in combined sewer overflows has not been confirmed in any of the eight measured storm events. It is assumed that organic material is continuously removed by dry weather discharges and is transported towards the waste water treatment plant without having effect on the receiving water.

• Furthermore, storm discharges release and transport the mobilised sediments during the whole storm event; thus, any possible first flush effects are levelled out. The organic content of suspended solids tends to increase due to mobilisation of deeper horizons and the ratio of VSS/TSS reaches values of ca 40 to 45% towards the end of the storm event.

• The presented study strives to improve knowledge about the transport of pollution in urban drainage networks within the possibilities and financial constraints of the research project.

Fig.6 Typical VSS and TSS concentrations during dry weather periods

Fig.7 Typical concentrations of VSS and TSS during overflow after a two weeks
dry weather period

Fig.8 Typical concentrations of VSS andTSS during overflow in the second storm event

Fig.9 Typical ratio of VSS/TSS during dry weather periods

Fig.10 Typical ratio of VSS/TSS during overflow after a two weeks weather period

Fig.11 Typical ratio of VSS/TSS during overflow in the second storm event

ACKNOWLEDGEMENT

Financial support of this project by the Ministry of Land Management of the Slovak Republic is gratefully acknowledged. Financial support of part of this research (Malacky) in the form of a NATO Linkage Grant is gratefully acknowledged.

REFERENCES

Arthur, S. and Ashley, R. M. (1998) The influence of near bed solids transport on first foul flush in combined sewers. Wat. Sci. Tech., 37/1, 131-138.

ATV A 128 (1992) Richtlinien für die Bemessung und Gestaltung von Regenentlastungen in Mischwasserkanälen.

Ellis, J. B. (1986) Pollutional aspects of urban runoff. In: Urban runoff pollution, NATO ASI Series G: Ecological sciences, vol. 10. Torno, H.C., Marsalek, J., Desbordes, M. (Eds.), Springer-Verlag Berlin, 1-38.

Markovic, A. (1999) Determination of technical criteria for CSO design and operation. Final Report of Part 04.01, Project 95/5145/442, Water Research Institute Bratislava.

Marsalek, J. (1990) Evaluation of pollutant loads from urban non-point sources. Wat. Sci. Tech., 22/10-11, 23-30.

Rusnak D. and Stanko S. (2000) Comments on the use of limiting rainfall. Proc. Conf. Actual Problems in Urban Drainage, R. Teplice, 22-23.2.2000, 102-112.

Saget, A., Chebbo, G. and Bertrand-Krajewski, J. L. (1996) The first flush in sewer systems. Wat. Sci. Tech., 33/9, 101-108.

US EPA (1983) Results of the National Urban Runoff Program. Vol. 1, Final Report, US EPA Water Planning Division, Washington, D.C.

Sokac, M. and Markovic, A. (2000) Mathematical modelling of pollution transport in combined sewer systems. Proc. Conf. Actual Problems in Urban Drainage, R.Teplice, 22-23.2.2000, 113-119.

Sztruhar, D., Sokac, M., Hyanek, L., Frankova, E., Rusnak, D., Namer, J., Stanko, S. and Marsalek, J. (1997) Conjunctive monitoring of a sewer system and receiving waters in a medium sized community. Wat. Sci. Tech., 36/8-9, 271-276.