1

AN ANALYSIS OF A CONSTRUCTED WETLAND FOR TREATING ROAD RUNOFF IN IRELAND

BY NEIL HIGGINS

DUBLIN CITY COUNCIL NEIL.HIGGINS@DUBLINCITY.IE

Summary Constructed wetlands have been widely used and examined for the treatment of sewage, and for urban and agricultural runoff. Moreover, they have recently become a more common feature on road networks as a treatment facility for highway runoff. Despite the increase in popularity there is still relatively little data available on their treatment performance and its relationship with the design parameters. In the autumn of 2004 the first constructed wetland system to exclusively store and treat runoff from a major highway in Ireland was established. The site was located adjacent to the M7 motorway, which links the towns of Kildare and Portlaoise and has an average daily traffic flow of approx 30,000 vehicles. The wetland was designed and constructed as a free-surface-flow system with a mix of two plant species, Phragmites australis and Typha latifolia. The performance efficiency of the system has been evaluated since spring 2005 analysing a mixture of physical and chemical parameters associated with road runoff (such as heavy metals), which have fluctuated notably throughout the monitoring period. The pollutant load removal efficiency for total suspended solids was 95%, 85% for total phosphate, 74% for total organic carbon, 86% for total copper, 95% for total zinc, 86% for total cadmium and 85% for total lead respectively. These results would suggest that the constructed wetland is very efficient at removing pollutant loads in highway runoff. Introduction The construction of motorway grade roads in Ireland has intensified in recent years under the driving influence of the Irish Government’s National Development Plan 2000-2006 and concomitant with the development of the road network has been an increase in traffic densities. With vehicle numbers reaching some 1.8 million in Ireland (NRA, 2005), traffic numbers on many roads are now exceeding the 30,000 vehicles per day threshold limit set for implication of mitigation methods, as laid out in the UK Design Manual for Roads and Bridges (1998) which has been adopted by the National Roads Authority in Ireland. These guidelines recommend that some form of mitigation, whether vegetative or structural should be incorporated into the infrastructure of any road exceeding the threshold traffic flow limit. EU legislation, in particular the Water Framework Directive 2000 is also applying pressure for the control of discharges to any receiving water whether ground or surface (EU, 2000). With the Irish climate and geological features favourable for wetland growth and prosperity, their use as a possible mitigation method to treat highway runoff has now become a topic of wide interest

2

Constructed wetland basins normally comprise of a non-soil substrate and a permanent, but usually shallow volume of water that can be almost entirely covered in aquatic vegetation (Halcrow, 1998). The wetlands provide physical, chemical, and biological water quality treatment of highway runoff (CIRIA, 1998; Halcrow, 1998). Physical treatment occurs as a result of decreasing flow velocities in the wetland, which promotes sedimentation, evaporation, adsorption, and filtration (Halcrow, 1998). Biological processes include decomposition, plant uptake and removal of nutrients, plus biological transformation and degradation (CIRIA 1998, Halcrow, 1998). The overall treatment performance of the constructed wetland is influenced both by its design and the way it has been constructed. Factors such as: local climate, topography and geology; traffic loadings (present and future); road drainage area; land availability; cost size/extent and type of receiving water body; water quality classification and objective; and environmental enhancement value will all dictate whether a constructed wetland is appropriate for a site (Schutes et al., 1999). Few full-scale trials using wetlands to intercept and treat road runoff have been reported to date, but the studies that have been undertaken (Bulc et al., 2003; Revitt et al., 2004) demonstrate that wetlands can be one of the most efficient practices for flood attenuation, reduction of peak discharges and overall enhancement of the water quality. The added vegetation has been shown to promote a significant removal of soluble constituents such as nutrients and dissolved metals as reported in a desk study by Mudge and Ellis (2001). This review of previous in-situ studies on wetlands reported pollutant removal efficiencies of 70% to 95% for total suspended solids, 50% to 85% for hydrocarbons and 40% to 75% for various metals (including up to 40% for the dissolved metal fraction). This paper presents the initial results of the performance of a constructed wetland, in response to a series of storm events that has been built to receive the runoff from a new motorway in Ireland as located in Figure 1.

Fig. 1: Location of Wetland Site

3

METHODOLOGY Constructed Wetland Design The site selected for the construction of the wetland was located adjacent to the new motorway linking the towns of Kildare and Portlaoise in the East of Ireland. This site was preferable as the road had a predicted traffic count of at least 30,000 vehicles per day and also allowed the research team safe and easy access to construct the wetland before the road went ‘live’ to commuting vehicles. The overall design of the constructed wetland was determined from guidelines contained in two interim manuals: “Treatment of highway runoff using constructed wetlands” (Halcrow, 1998); and “Review of the design and management of constructed wetlands” (CIRIA, 1998). The hydrological data and pollutant loading rates from two other roads in Ireland that were being monitored as part of a wider overall research project in to highway runoff were also used to formulate the design. The first step was to design the pipe network into the wetland in order to avoid any blockages or backflow onto the road. The critical inflow Qc into the system for a one-year return period was calculated as 58 l/s using the Rational Method. The required pipe diameter into the wetland was determined to be a 375 mm pipe onto which four 0.1 m diameter pipe inlets were attached via saddles at equal spacings across the width of the wetland to ensure even flow distribution. These inlets discharged onto stone gabions to prevent localised scouring and channelling. The outlet pipes were a similar setup with mobile T-pieces used as weirs to control the water level within the wetland. The dimensions of the system were calculated on the basis of modelling the wetland as a simple reservoir system. As the area in which to construct the wetland was limited, a minimum retention period of 1 hour was chosen, contrary to the recommended minimum period of 5 – 10 hours (Halcrow, 1998). The dimensions were based around a maximum 1-hour rainfall event with a 1-year return period recorded as 7.8 mm per hour for the area. Other critical factors used in the design were that the depth of the wetland was no greater than 0.4 metres at any point and the cross sectional slope was between 0.5 and 1% of the total longitudinal length. The final dimensions of the wetland worked out to be 14m wide by 19.5m length with a cross sectional slope of 1%. When the excavation was completed the clay base was compacted several times to produce a relatively impermeable layer. The permeability of the clay liner was subsequently tested using the Double Ring Infiltrometer, which yielded a value of K < 1x10-9m/sec. The importance of this liner was to minimise loss of the runoff down through the base, which in turn could impact upon the underlying groundwater. Once the base was completed a 120mm layer of topsoil was added and the section was divided into two subsections, cells A and B as seen in Figure 2. Cell A was planted with 500 Phragmites australis and cell B with 500 Typha latifolia - approximately four reeds per square metre.

4

Fig. 2: Diagram of wetland system

Fig.3: Elements of Wetland System

5

MONITORING The site was installed with an ISCO 674 0.1mm tipping bucket rain gauge, which recorded the rainfall on a one-minute time series. The flow into the wetland was measured with a low profile ISCO 750 area velocity flow module which was placed within the inlet pipe to measure the average velocity (using the Doppler effect principle) and depth (hydrostatic pressure) of the flow in the pipe. On the contrary the flow out of the wetland system was deemed to have a much smaller flow rate than the inlet so a v-notch weir system with an ISCO 730 bubble module to measure the depth was installed and a relationship between depth and flow over the v-notch was established. Probes that measured the temperature, pH, dissolved oxygen and conductivity were also installed in both the inlet and outlet. All these devices led into the central ‘mother’ system, an ISCO 6712 automatic sampler (Fig. 4), which acted as a storage cell but also took samples of the runoff via a suction tube deposited into a series of 300 ml bottles. Once all monitoring equipment was correctly in place, the sampler was programmed to activate when a certain criteria was met, such as the recording of a depth of rainfall. The pacing at which the samples were taken was regulated by the flow in the pipe, for instance a sample of runoff would be taken after every 3 m3 of runoff had passed the flow meter. This setup worked well in capturing the entirety of the main storm events. In the time period from summer to autumn 2005 six major storm events were captured and fully sampled. The samples of runoff were collected at both the inlet and outlet and transported immediately back to the laboratory where they were analysed for a number of parameters commonly found in highway runoff, including total suspended solids, total organic carbon, total phosphorus and four heavy metals, zinc, cadmium, copper and lead respectively. The water quality analysis was carried out in the laboratory in accordance with the Standard Methods (APHA, 1998).

Fig.4: Automatic sampler, rain gauge and flow monitoring equipment

6

RESULTS AND DISCUSSION Hydrological Analysis The hydrological characteristics of the 6 storm events sampled for untreated runoff into the wetland and treated runoff out of the wetland during the summer and autumn seasons of 2005 are summarised in Table 1 with the hydrographs shown in Figure 2. The data presented covers intensive monitoring over a three-month episode during which a number of short duration intense storm events were captured. The constructed wetland performed well hydraulically, buffering the inflow to the system from the highway and substantially reducing the peak flow by as much as 96%. The total volume entering the wetland was efficiently stored within the system, particularly during the month of August, which had frequent intense storm events with associated high pollutant loads. The runoff was detained by the wetland for a number of days and then was discharged by the following storm events. The hydraulic retention times (HRT) for the wetland during these periods varied from between 1 and 12 days. Outflow from the wetland was a steady low flow with little potential for erosion of the receiving river with a mean flow as much as 94% lower than the inflow. In all, the wetland provided both sufficient storage and flow attenuation so that a number of processes such as sedimentation, filtration and biodegradation could be effectively harnessed for the removal of pollutants.

Volume Peak Flow

Mean Flow

Date Inlet Outlet Reduction Inlet Outlet Reduction Inlet Outlet Reduction

(m3) (m3) (%) (l/s) (l/s) (%) (l/s) (l/s) (%)

18Aug05 39.52 3.52 91.1 10.74 0.53 95.1 3.87 0.25 93.5

23Aug05 10.62 1.02 90.4 6.69 0.23 96.6 1.36 0.12 91.2

09Sep05 65.86 13.36 79.7 12.02 1.31 89.2 3.41 0.55 83.9

26Sep05 56.6 47.80 15.6 16.43 10.55 35.8 4.72 1.66 64.8

28Sep05 33.61 13.88 58.7 20.07 1.86 90.7 5.09 0.54 89.4

10Oct05 98.99 52.70 46.8 21.37 3.91 81.8 3.23 1.39 57.0

MEAN 50.86 22.05 63.7 14.55 3.06 81.5 3.61 0.75 80.0

Table 1: Summary of hydrological data for events sampled during 2005

7

Inflow vs Outflow for Wetland 9 Sept 05

0

3

6

9

12

15

18

09:00 10:20 11:40 13:00 14:20 15:40 17:00 18:20 19:40 21:00 22:20 23:40

Time

Flow

(l/s

)0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

Rai

nfal

l (m

m)

Rainfall Inflow Inflow Sample Outf low Outf low Sample

Inflow vs Outflow for Wetland 28 Sept 05

0

5

10

15

20

25

30

13:40 14:40 15:40 16:40 17:40 18:40 19:40

Time

Flow

(l/s

)

0

0.5

1

1.5

2

2.5

Rain

fall

(mm

)

Rainfall Inf low Inflow Sample Outf low Outf low Sample

Fig. 5: Hydrographs of two storm events sampled during autumn 2005 Hydrochemical Analysis Water quality analysis was carried out on each of the samples to determine the concentrations of total suspended solids, total organic carbon, total phosphate and four heavy metals, zinc, lead, cadmium and copper. However, it was clear from a visual comparison of the influent and effluent samples that the wetland was certainly functioning well in terms of removing the sediment from the runoff.

8

Total Suspended Solids (TSS) Inflow values of TSS ranged between 43 and 437 mg/l. These values were comparable with earlier findings from the wider research project of direct highway runoff and were found to be dependent on a number of factors including rainfall intensity, antecedent conditions and traffic densities. The outflow values ranged between 10 and 32 mg/l respectively. The removal efficiencies of TSS fluctuated from 87.6 to 92%, with an average of 88.3%, which is relatively high and confirmed, as indicated by the visual inspection, that the wetland was behaving as an excellent sedimentation system. Total Organic Carbon (TOC) Inflow values of TOC were quite variable and ranged between 2.23 and 47.04 mg/l. These values are considered to be high which possibly can be explained by the fact that the highway was new and had just been surfaced with a thick layer of bitumen. The outflow values were also surprisingly variable ranging from 2.74 to 13.4 mg/l. The removal efficiency of the wetland fluctuated from a negative value of -22 to 40.4%, with an average of 13.7%. It is suggested that this relatively poor performance is primarily due to the wetland not effectively removing the organic fraction in the form of oil since no oil interceptor had been installed upstream of the wetland process. It is therefore recommended that an oil interceptor should always be incorporated as an inherent part of this treatment technology. Total Phosphate (TP) Inflow values of TP ranged between 0.096 and 0.614 mg/l. The outflow values ranged between 0.06 and 0.188 mg/l. The removal efficiency varied between 52 to 75%, with an average of 62%. The dominant removal process of the phosphates removal was assumed to be sedimentation. Heavy Metals: Zinc, Copper, Cadmium and Lead Inflow values of zinc ranged between 0.081 and 0.426 mg/l. Outflow values were much lower and ranged between 0.012 and 0.045 mg/l. The corresponding removal efficiencies varied between 75 to 90.6%, with a high average of 85%. Inflow values of copper ranged between 0.019 and 0.095 mg/l. Outflow values ranged between 0.008 and 0.039 mg/l. Removal efficiencies varied between 50 to 79.7%, with an average of 68.4%. Inflow values of lead ranged between 0.027 and 0.092 mg/l. Outflow values ranged between 0.016 and 0.045 mg/l. Removal efficiencies varied between 47.1 to 70%, with a high average of 61.6%. Inflow values of cadmium ranged between ND and 0.008 mg/l. Outflow values ranged between ND and 0.002 mg/l. Removal efficiencies varied between 50 to 71.4 %, with a high average of 61.9%. Heavy metal analysis is ongoing and in particular is focusing on the dissolved fraction, which is the most toxic state. With both the hydraulic and water quality data available, the total pollutant loading rates into and out of the wetland have been calculated for each individual month of monitoring as summarised in Table 2. It can be seen that removal rates of pollutant loads were very high particularly during the month of August when the wetland was growing steadily.

9

LOAD(g) Aug-05 Sep-05 Oct-05 Mean LOAD(g) Aug-05 Sep-05 Oct-05 MeanTSS TZn

INLET 7,650 24,100 23,800 INLET 10.2 33.3 18.8

OUTLET 85.4 1,660 1,260 OUTLET 0.13 2.31 1.08

%RED* 98.9 93.1 94.7 95.6 %RED 98.8 93.1 94.3 95.4TCd TP INLET 0.27 0.70 0.51 INLET 12.1 40.3 32.0

OUTLET 0.01 0.11 0.11 OUTLET 0.17 10.6 6.02

%RED 95.7 83.7 79.2 86.2 %RED 98.6 73.8 81.2 84.5TCu TOC INLET 2.77 6.76 5.91 INLET 48.7 1,620 580

OUTLET 0.056 1.26 0.69 OUTLET 4.18 708 157

%RED 98.0 81.4 88.4 86.2 %RED 91.4 56.3 72.8 73.5TPb INLET 4.07 9.51 7.01

OUTLET 0.17 1.79 1.45

%RED 95.9 81.1 79.4 85.5 *%RED: Percentage reduction in load

Table 2: Loading rates in grams into and out of the constructed wetland

Difference between Plant Species A semi partition was constructed between the wetland cells A and B, which allowed runoff to flow freely throughout the system. Research is now focusing on the effluent from each cell A and B; however, during the monitoring period, grab samples of runoff were collected from specific locations within each cell. Water quality analysis for TSS gives an initial indication that cell A which is planted with Phragmites australis was performing up to twice as efficiently with respect to TSS removal compared to cell B planted with the Typha latifolia. It should be noted that TSS removal would be expected to correlate closely with the heavy metal and total phosphate fraction. A plausible reason could be due to the more rapid propagation of the Phragmites australis compared to the Typha latifolia, which would promote a more even distribution of flow across the width of the wetland and thus more optimal conditions for solid settlement. Specific Conductivity, pH, Dissolved Oxygen, and Temperature The average inflow concentration of DO was 6.9 mg/l and was noted to be highly variable compared to the more stable DO levels at the outlet, which averaged 7.7 mg/l. This rise in DO was assumed to be due to the high air-water interface of the shallow surface water wetland and relatively low organic content. Inflow temperatures ranged from 2 to 22ºC with an average of 13.3ºC. Average outflow temperature was 14.3ºC, ranging from 10 to 20ºC. The average increase of 1ºC suggests the wetland maybe acting as a heat sink and is discharging slightly warmer water to receiving water body. The

10

average specific conductivity was 10µS/cm at the inflow and 125µS/cm at the outflow respectively. The specific conductivity was important in calculating the optimal retention time of the wetland. The inflow values of pH ranged from 4 to 8 and the outflow values ranged from 7.5 to 9. The average value of the inflow was 7.2, and 8.4 at the outflow respectively. It is well documented that wetlands act as buffer zones neutralizing the acidic nature of the inflow. The pH and net acidity/alkalinity of the water are particularly important because pH influences a number of reactions within the wetland and hence a number of the treatment processes (Kadlec and Knight, 1996). CONCLUSIONS The results of this study show that the constructed wetland performed exceptionally well in pollutant removal efficiency for a number of parameters particularly suspended solids and heavy metals. The dominant process involved in the removal of pollutants appeared to be due to sedimentation of the solid fraction. Research is still ongoing as the wetland matures and other treatment processes such biological interactions develop. The influence of the plant species was not examined thoroughly but early indications suggest that the faster propagation of the Phragmites australis plant compared to the Typha latifolia has provided a greater surface area to inhibit short-circuiting through the wetland and thus promote more optimal hydraulic conditions for solid settlement. Further investigation will be carried out on the capability of the wetland to remove the dissolved fraction of the pollutants especially the heavy metals in addition to the overall pollutant removal efficiency as the wetland matures and flourishes. The final goal of the future research will be to formulate design criteria so that constructed wetlands can be used successfully to treat runoff from highways under Irish conditions. ACKNOWLEDGEMENTS The work was funded by Ireland’s National Development Plan under the ERTDI, managed by the Environmental Protection Agency and co-funded by the National Roads Authority. The project team included Paul Johnston, Neil Higgins, and Laurence Gill from Trinity College and Dr Michael Bruen and Mesfin Desta from University College Dublin. REFERENCES APHA (1998). Standard methods for the examination of water and waste water. Technical report, American Public Health Association, Washington DC, 20th. edition. Bulc. T. and Slak A.S. Performance of a constructed wetland for highway runoff treatment. Wat. Sci. Tech., 48: 315-322. CIRIA (1998). Review of the Design and Management of Constructed Wetlands. Technical Report, CIRIA, Report 180

11

DMRB-UK (1998). Design manual for roads and bridges: Water quality and drainage, volume 11, sec 3, part 10. Technical report, Highways Agency, UK. EU (2000) Water Policy Framework Directive 2000/60/EC, European Commission, December 2000 Halcrow Group Ltd & Middlesex University (1998). Treatment of highway runoff using constructed wetlands: An interim manual. Halcrow Group Ltd & Middlesex University, Urban Pollution Research Centre, Environment Agency Thames Region, Reading. Kadlec, R.H. and Knight, R.L. (1996). Treatment Wetlands. Lewis Publishers, CRC Press, Inc. Boca Raton, Florida, USA Mudge, G. and Ellis J.B, 2001. Guidelines for the Environmental Management of Highways, Chapter 4, 67-102, the Institution of Highways and Transportation, London, UK. NRA (2005) National Roads and Traffic Flow 2004, National Roads Authority Preliminary Report, March 2005 Revitt, D.M. and Schutes, R.B.E. (2004). The performances of vegetative treatment systems for highway runoff during dry and wet conditions. Sci. Total Envir., 335: 261-270. Schutes, R.B.E. and Revitt, D.M. (1999). The design of vegetative constructed wetlands for the treatment of highway runoff. Sci. Total Envir., 235: 189-197.