11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008

Quantifying the Performance of Storm Tanks Dr Will Shepherd*, Professor Adrian Saul and Dr Joby Boxall Pennine Water Group, Department of Civil and Structural Engineering, The University of Sheffield, Sheffield, S1 3JD, England, UK. *Corresponding author, e-mail [email protected]

ABSTRACT Storm tanks at Wastewater Treatment Works (WwTW) provide storage and sedimentation for excess flows entering the WwTW as a result of storm events, as such they are an essential but often neglected component of the sewerage system. Historically in the UK, flows entering the WwTW are limited to approximately 6 times the mean daily dry weather flow (DWF) through the use of combined sewer overflows in the system and an emergency overflow at the entrance to the works. Approximately half of the works inflow (i.e. 3 × DWF) is passed to full treatment and the remainder is discharged to the storm tanks. Once the tanks are full the excess flows are spilled to the nearest watercourse or ocean. After the storm event has subsided and the DWF returns to normal, the storm tanks are emptied, with the effluents being given full treatment. The pollution retention performance of storm tanks has generally been omitted from regulatory guidelines, the tanks being designed solely on the stored effluent volume. As a result, certainly in the UK, the design philosophy for storm tanks has been static, and has not moved forward in the same way as other developments associated with integrated quantitative and qualitative modelling of sewer systems. The Water Framework Directive, will require all water utility companies (within the EU) to better understand the magnitude, volume and quality of all intermittent discharges that issue from sewerage systems into receiving waters. This approach is recommended at the integrated catchment scale and hence the contribution to receiving waters from storm tanks may form one of the major elements of pollution. This paper is based on UKWIR research project WW22B (UKWIR, 2007) and includes a review of current practice and presents results of a field and laboratory based experimental programme which improves understanding of the hydraulic and pollutant retention performance of rectangular storm tanks.

KEYWORDS Storm tanks, design, hydraulic performance, residence time, pollution retention performance, fieldwork, laboratory work.

INTRODUCTION The treatment efficiency of Wastewater Treatment Works (WwTW) is influenced by the volume and strength of sewage passing through them. As such, flow from combined sewer systems into the treatment process is limited to approximately 3 times the mean Dry Weather Shepherd et al.

1

11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008 Flow (DWF). Flows between 3 and 6 times the DWF are passed into storm tanks, which have been an integral part of the sewage treatment process since the early 20th century and the basic functions of storm tanks include: 1. acting as temporary storage tanks when excess flows enter the WwTW and to retain as much storm sewage as possible for later treatment; 2. reducing the number of occasions on which excess flow would have discharged to the receiving water; 3. allowing settlement to reduce the strength of the storm sewage that is spilled from tanks; 4. increasing the time of concentration of the sewerage system such that the spill flow occurs at a later time which should coincide with increased flows in the receiving water, thereby providing additional dilution; 5. retaining the more highly polluted first flush of sewage that is frequently observed during the early part of the increase in sewer flow caused by rainfall. To achieve acceptable performance it is desirable that the tank has sufficient depth to allow the settlement of solids, that scour and re-suspension of settled solids does not occur and that the washout of solids to the receiving water is avoided. In the UK, the design of storm tanks is generally based on the Environment Agency recommended guidelines of 2 hours retention at 3PG + I + 3E (where P is population, G is number of litres per head per day, I is infiltration inflow and E is industrial effluent) or 68 litres per head of population. These guidelines result in a design tank volume and it is common practice to use tanks which are either rectangular and circular on plan. There are however no guidelines on the length to breadth to depth ratio for rectangular tanks or of the diameter to depth ratio for circular tanks. Neither guideline specifically considers the need to retain pollutants in the storm flow (although any first flush may be anticipated to occur within the early part of a storm) or the pollutant retention processes that occur in the tank. Tank geometry will influence the hydraulic performance of the tank and this in turn will influence the retention of the pollutants that enter the tank within the inflow. However, no link is made, at the time of tank design, between the hydraulic performance and the selected geometry and dimensions of a tank, such that the pollution retention performance, for a known inflow (quantity and quality) is optimised. This is primarily due to the fact that the current regulatory drivers do not require that the quality of the spilled flow from tanks be known. However, European legislation, in the form of the Water Framework Directive, will require all member states to better understand the pollutant impact of all intermittent discharges to receiving waters, including all CSO structures and storage tanks within the sewerage systems and storm tanks at WwTW. This paper presents a brief review of previous research and results from a study to better understand the behaviour of rectangular storm tanks through testing in both field and laboratory.

REVIEW OF STORM TANK DESIGN AND PERFORMANCE. A literature review highlighted that little work has been completed to specifically examine the performance of storm tanks at WwTW. Most of the literature addresses the performance of conventional on-line and off-line storage tanks and of primary and secondary settlement tanks and clarifiers. In respect of the latter, many models of tank performance have been developed. These include theoretically based formulae, regression based methods, and, more

2

Quantifying the Performance of Storm Tanks

11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008 recently, mathematical simulation of the dynamic processes. The measure of a settling tank’s performance to retain pollution has been termed its removal efficiency, η , and this has been defined as the proportion of the inflow load that is retained within the tank over the duration that the tank is in operation. Historically, the efficiency of tanks has been based on Surface Load Clarification Theory for an Ideal Rectangular Tank. For example, in Germany a surface loading rate of 10m/hr and a tank length to width ratio of at least two are specified whilst, in America the specified loading rates are much lower, ranging from 0.5m/hr for small populations (US Army Corps of Engineers) to 5m/hr (Metcalf and Eddy, (1991)). The theory was originally developed by Hazen (1904), and expanded by Camp (1946), who defined an ideal rectangular continuous flow settling basin as having the following characteristics: 1. the direction of flow is horizontal and the velocity is the same in all parts of the settling zone (hence, each particle of water is assumed to remain in the settling zone for a time equal to the detention period – namely, the volume of the settling zone divided by the discharge rate); 2. the concentration of suspended particles of each size is the same at all points in the vertical cross section at the inlet end of the settling zone; 3. a particle is removed from suspension when it reaches the bottom of the settling zone; 4. for any given discharge, the removal is a function of the surface area and is independent of the depth of the basin, or, the removal is a function of the overflow rate, and, for a given discharge is independent of the detention period; 5. the concentration of suspended matter at any cross section in the settling zone increases with the depth below the surface, and decreases with the proximity of the cross section to the outlet end of the basin. Any particle settling in a moving liquid will move in a particular direction and at a particular velocity, which is the vector sum of its own settling velocity and the velocity of the surrounding liquid. In an ideal rectangular tank the paths of all discrete particles will be straight lines, and all particles with the same settling velocity will move in parallel paths (as a function of the position that they enter the tank). In the design of settling tanks, the usual procedure is to select a particle with a terminal velocity Vc and to design the tank so that all particles that have a terminal velocity equal to or greater than the design terminal velocity will be removed. The rate, Q, at which clarified water is produced is then: Q = AVc

where A is the surface area of the basin. Effectively the selection of Vc defines the surface overflow rate or surface loading rate, expressed in cubic metres per square metre surface area. Particles that have a fall velocity of less than Vc will not be removed during the time provided for settling. If an average settling velocity (Vs ) is assumed for the suspension, then according to surface load theory the removal efficiency (η ) for quiescent conditions is given by: η=

Vs Q/ A

It is stressed that this expression is only valid under uniform, steady and laminar flow conditions. Furthermore, Metcalf and Eddy (1991) identified that particles may settle as discrete particles, as flocculants, by hindered settling or due to compression. Of these processes discrete particle settling and flocculant settling are most commonly observed in storm tanks. Hence particle retention is a function of the nature and type of the particles and Shepherd et al.

3

11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008 of the surface overflow rate. Clements (1966), studied the velocity variations through a rectangular storm tank, and he observed that sometimes there were significant velocity variations across the width of the tank. These changes in velocity would change the surface overflow rate, and, based on the work of Hazen (1904), he introduced the concept of an effective velocity – based on the effective settling length of the tank taking due regard of the velocity variations across the width of the tank. When velocity variations were high, the effective mean velocity through the tank was also high, with a consequent reduction in the settling efficiency when compared to that calculated from the actual mean velocity. New design guidelines for storm tanks were proposed based on the concept of an effective velocity to allow the particles of sewage to settle out. Consideration has also to be given to the pollution characteristics of the influents that enter into a tank. A typical influent contains a distribution of particle sizes and hence a range of particle terminal velocities, Chebbo and Bachoc (1992) and Michelbach and Weiss (1996) whilst Madaras and Jarrett (2000) described the spatial and temporal distribution of sediment concentration and particle size distribution in a full size sedimentation basin. It is usual for the distribution to be split into a number of discrete terminal velocity bands with the efficiency calculated for each band. The total removal efficiency is calculated by summing the respective efficiency for each terminal velocity band. Lessard and Beck (1991) presented a conceptual model for the simulation of the dynamic performance of storm tanks. Four modes of behaviour were considered: the filling and emptying process and the effects of dynamic sedimentation and quiescent settling. The model was applied to a storm event at Norwich WwTW and the simulated results showed that the storm tank had a relatively small effect on the spilled load of suspended solids. A sensitivity analysis showed that this finding was a function of the characteristics and distribution of the sediments that entered the tank. Similar conclusions were reached by Baumer et al. (1996) and Lindeborg (1996) who carried out a series of dynamic performance tests in final settling tanks and sedimentation tanks respectively. The US EPA (1986) proposed the following methodology to estimate sediment removal under dynamic conditions: ⎛

η =1 − ⎜⎜1 + ⎝

1 Vs ⎞ ⎟ n Q / A ⎟⎠

−n

where n is a turbulence or short circuiting constant that is used to indicate the settling performance of the pond: n = 1, for poor performance; n = 3, good performance; n > 5, very good performance; and n = ∞ , ideal performance. The value of n is thus somewhat subjective, hence there is a need to more accurately define the short circuiting constant.

DETERMINATION OF RESIDENCE TIME IN STORM TANKS As part of the UKWIR study, testing has been carried out in both full scale storm tanks and laboratory scale models. This testing was developed to investigate the deviation of measured residence times from theoretical residence times, which assume idealised plug flow conditions. Field testing The rectangular storm tank used in the study was located at a WwTW in the North West of England and was of a typical design with a full width weir distributing inflow to the tank and a similar weir to over which excess flows spill from the tank. The tank tested was 18.5 m

4

Quantifying the Performance of Storm Tanks

11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008 long, 6.3 m wide and had a mean depth of 1.8 m, giving a volume of 214 m3, this places it around the mean dimensions of the storm tanks surveyed during the project. Scumboards were temporarily fitted inside the tank adjacent to the inlet and spill weirs as per the original tank design, these were then removed to assess their effect on residence times. A photograph of the tank as tested is shown in Figure 1 and a schematic of the test layout in Figure 2. Full details of the site and field testing set-up are included in the UKWIR (2007) report.

Figure 1: Photograph of field tank. FFT Penstocks

Pump

Inlet Channel Tracer Injection

Tank 1

Tank 2

Tank 3

Scufa Sand bag

Figure 2: Schematic drawing of field testing arrangement. Testing was carried out at a range of constant flow rates, which were provided by pumping from the works inlet over the abandoned 3 DWF weir. To estimate the true residence time Rhodamine WT fluorescent tracer was introduced into the inflow to the tank at a location upstream of the tank. ‘SCUFA’ fluorometers were used to measure the tracer concentration on entry to the tank, adjacent to the inlet weir and on exit from the tank at the spill weir.

Shepherd et al.

5

11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008 Inlet Spill First arrival 34 minutes 5 percentile 38 minutes 50 percentile 105 minutes 95 percentile 247 minutes Centroid 121 minutes Peak 51 minutes Theoretical 136 minutes

2.0E-06

1.5E-06

1.0E-06

1.0E-07

8.0E-08

6.0E-08

4.0E-08

2.0E-08

5.0E-07

0.0E+00 0

50

100

150 200 Time (minutes)

250

300

Spill Concentration (l/l)

Inlet Concentration (l/l)

2.5E-06

0.0E+00 350

Figure 3: Example solute trace showing different definitions of residence time. Typical results from field tracer tests are shown in Figure 3, it can clearly be seen that due to mixing processes within the tank, the spill trace is much more lagged and attenuated than the inlet trace. There are a number of different techniques that may be used to define the residence time from the tracer data:- the minimum value is the difference in first arrival times of the two tracer distributions (in this example 34 minutes), the modal average is the time between peaks of the two traces (51 minutes), whilst the mean residence time may be described as the time difference between the centroid of the traces (121 minutes). Alternatively it is feasible to describe the time when a certain percentile of the tracer has passed through the tank, for example 5% (38 minutes), 50% (105 minutes) or 95% (247 minutes). These compare with a theoretical residence time, calculated by dividing the tank volume by the flow rate, of 136 minutes. In this paper the 50 percentile definition was used as this splits the pollutant load equally, half will have a greater residence time and half the pollutant load will have a shorter residence time.

Measured fifty percentile residence time (field scale minutes)

Figure 4 plots the mean fifty percentile residence times derived from the field data. It can be seen that that these fifty percentile residence times are significantly shorter than the theoretical residence times. The single test at a low flow-rate (high theoretical residence time) appears to show enhanced short-circuiting, whilst removing the scumboards was shown to increase the fifty percentile residence time (at the tested flow rate). The error bars in Figure 4, plotted to show ± 1 standard deviation from the mean, show that where repeat traces have been carried out, values of the fifty percentile residence time exhibit little scatter, enhancing confidence in the results. 200

Theoretical residence time

180

With scumboards

160

Without scumboards

140 120 100 80 60 40 20 0 0

5

10

15

20 Flow (l/s)

25

30

35

40

Figure 4: Windermere Fifty percentile residence times.

6

Quantifying the Performance of Storm Tanks

11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008 Laboratory testing Fieldwork testing was limited due to limited time on site, constraints on flow magnitudes and the fact that only 1 geometrical configuration was available. A series of laboratory tests was therefore completed to enhance the scope of the study by testing a Froude number scaled tank within a controlled and repeatable environment. The design of the laboratory tank, Figure 5, was based upon a dataset of storm tanks commonly used in the UK and was of modular construction such that a range of different geometry tanks could be tested. These geometrical arrangements included a scale model of the storm tank tested in the field and results from this laboratory tank are reported here.

Figure 5: Schematic drawing and photograph of laboratory tank. The laboratory test methodology was the same as that in the field, whereby a slug of fluorescent tracer was monitored as it passed through the tank at a range of constant flow rates. Cyclops fluorometers were used to measure the tracer concentration in the laboratory tests (these are smaller and more applicable to the laboratory situation than the SCUFAs used in the field testing). The primary advantages of laboratory testing were that clean water was used and that the hydraulic conditions were more accurately controlled. This resulted in variations in the background concentration of the injected tracer and in the magnitude of the flow rate over the duration of each test being negligible. Figure 6 shows how the shape of the tracer pulse varies with flow rate and highlighting the magnitude of the attenuation and dispersion processes affecting time varying pollutant concentrations. It can also be seen that there is less noise in the measurements than was seen in the field, meaning the start and end of the tracer pulse can be more accurately determined, and therefore the residence time more accurately calculated. 0.060 l/s 0.243 l/s

18% Concentration (% of inlet peak)

Concentration (% of inlet peak)

100%

80%

60%

40%

20%

0.060 l/s 0.243 l/s

16% 14% 12% 10% 8% 6% 4% 2%

0%

0% 0

100

200

300 Time (s)

400

500

600

0

2000

4000

6000

8000

Time (s)

a) Inlet traces b) Spill traces Figure 6: Comparison of tracer distributions in laboratory tank with scum-boards at high and low flow-rate.

Shepherd et al.

7

11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008

Fifty percentile residence time (s)

Figure 7 shows the mean 50 percentile residence times with error bars of ± 1 standard deviation for the repeat injections, as a function of flow rate. The results show the expected power law trend with discharge, however the measured residence times are always significantly less than the idealised theoretical residence times. 4500

Measured with scum-board

4000

Measured without scum-board Theoretical

3500 3000 2500 2000 1500 1000 500 0 0

0.05

0.1

0.15

0.2

0.25

0.3

Flow rate (l/s)

Figure 7:Variation in laboratory residence times with flow rate, measured and theoretical. In addition to the tracer tests, an experiment to directly quantify Total Suspended Solids (TSS) retention efficiency was conducted. This entailed introducing a constant concentration of crushed olive stone and collecting discrete samples from the tank inlet and spill. These samples were then analysed to determine TSS concentrations and allowed retention efficiency to be estimated based on the difference between inlet and spill concentrations. Crushed olive stone was used as a surrogate for sewer solids as this sediment, when mixed with water, has been shown to accurately represent live sewage particles in appropriately scaled laboratory tests (Ellis 1992). Figure 8 shows the results of the laboratory TSS experiment together with predicted retention efficiencies. These were calculated from the measured 50 percentile residence times and the measured settling velocity distribution of the crushed olive stone. From this it can be seen that when the residence time is correctly estimated, classical sedimentation theory can estimate retention efficiency with good accuracy. 100%

Retention efficiency (%)

90% 80% 70% 60% 50% 40% 30% 20%

Predicted

10%

Measured

0% 0

500

1000

1500

2000

2500

3000

Laboratory residence time (s)

Figure 8: Variation in retention efficiency with laboratory residence time, measured and predicted.

8

Quantifying the Performance of Storm Tanks

11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008 Verification of laboratory results In order to verify that the laboratory data is representative of the full scale tanks, a comparison between the data collected in the laboratory and in the field was carried out. To do this, it was necessary to apply scale factors to the laboratory results following Froude scale laws. A comparison of fieldwork and scaled-up laboratory tracer data is shown in Figure 9. Some differences between the two residence time distributions can be attributed to variations in the flow rate and background noise in the field tests. However the general shape of the recession limb is similar in both the laboratory and field. 10%

Field Laboratory

Concentration (% of inlet peak)

9% 8% 7% 6% 5% 4% 3% 2% 1% 0% 0

50

100

150 200 Time (field minutes)

250

300

350

Figure 9: Comparison of field and laboratory tracer data. The field and laboratory fifty percentile residence times for a range of flow-rates are compared in Figure 10. Comparison between the laboratory and fieldwork results show reasonable agreement in the fifty percentile residence times, particularly when flow variations present in the field tests are considered. It is immediately apparent that, in both the full-scale tank and in the scale model, the fifty percentile residence times are always less than the theoretical residence times, at the flows tested. It is suggested that the ratio of the theoretical to measured residence times could be used as a correction factor, and hence, in conjunction with a TSS settling velocity distribution, reliable estimates of solids retention efficiencies may be made. Such ratios would need to be quantified for a range of tank geometries at various flow rates in order to be used in storm tank design. Measured fifty percentile residence time (field scale minutes)

Theoretical residence time 140

Field Laboratory

120 100 80 60 40 20 0 0

10

20

30 40 50 Flow (field scale l/s)

60

70

Figure 10: Comparison of field and laboratory residence times.

Shepherd et al.

9

11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008

CONCLUSIONS Rectangular storm tank 50 percentile residence times measured in the field have been shown to agree well with those measured in a laboratory scale model. This provides confidence that residence times in a range of tank geometries can be successfully investigated using Froude scaled models. The 50 percentile residence times measured in the laboratory and the field have both been shown to be less than the theoretical value obtained assuming idealised flow conditions, thus storm tanks designed without taking this into consideration will over predict pollution retention. It is suggested that a ratio of measured to theoretical residence times could be used to more accurately predict solids retention efficiency, when used with a known solids settling velocity distribution. This is supported by agreement between the results of TSS retention measurements made in the laboratory and TSS retention calculations based on classical sediment transport theory, made using measured residence times and settling velocity distributions.

ACKNOWLEDGEMENT The authors are indebted to UKWIR for financing the project and allowing the publication of the research. The opinions expressed are those of the authors and do not necessarily reflect those of UKWIR or the companies of the authors.

REFERENCES Camp T. R. (1946). Sedimentation and the design of settling tanks, Trans. ASCE, Vol. 111, 895-936 Chebbo, G and Bachoc, A. (1992). Characterization of suspended solids in urban wet weather discharges. Wat. Sci. Tech., Vol. 25, No. 8, 171-180 Clements, M. S., (1966), Velocity variations in rectangular sedimentation tanks, Institution of Civil Engineering Proceedings, Vol. 34, p171-200 Ellis, D.R., 1992. The design of storm drainage storage tanks for self cleansing operation. Thesis, (PhD). University of Manchester. Hazen A. (1904). On sedimentation. Trans., ASCE, Vol. 53, 45 – 71 Lessard, P. and Beck, M.B., (1991), Dynamic simulation of storm tanks, Water Research, Vol. 25, No. 4, 375391 Madaras, J.S., and Jarrett, A.R., 2000. Spatial and temporal distribution of sediment concentration and particle size distribution in a field scale sedimentation basin. Transactions of the American Society of Agricultural Engineers, 43 (4), 897-902. Metcalfe and Eddy, Inc. (1991). Wastewater Engineering, Treatment, Disposal and Reuse, 3rd Edition, McGraw – Hill, Singapore, ISBN 0 – 07-100824 – 1. Michelbach, S., and Weiss, G.J., (1996). Settleable sewer solids at stormwater tanks with clarifier for combined sewage. Water Science and Technology, 33(9), 261-267. UKWIR (2007) Performance of Storm Tanks and Potential for Improvements in Overall Storm Management – Phase 2. (07/WW/22/5) ISBN: 1 84057 469 0

10

Quantifying the Performance of Storm Tanks