HAL Id: hal-00508188
https://hal.archives-ouvertes.fr/hal-00508188
Submitted on 2 Aug 2010
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
behaviour and limits due to hydraulic overloading on first stage filters
Pascal Molle, A. Lienard, A. Grasmick, A. Iwema
To cite this version:
Pascal Molle, A. Lienard, A. Grasmick, A. Iwema. French vertical flow constructed wetlands: reed bed behaviour and limits due to hydraulic overloading on first stage filters. 9ème conférence sur les marais artificiels et 6ème conférence sur le lagunage naturel, Sep 2004, Avignon, France. 5 p. �hal-00508188�
French vertical flow constructed wetlands: reed bed behaviour and limits due to hydraulic overloading on first stage filters P. Molle*, A. Liénard*, A. Grasmick**, A. Iwema***
* Cemagref, Research unit: Water quality and pollution prevention, 3bis, quai Chauveau - CP 220, 69336 Lyon Cedex 09 – France (E-mail: [email protected]; [email protected])
** Laboratoire de Génie des Procédés et Sciences des Aliments – CC024 Université de Montpellier II – Place E. Bataillon - 34095 Montpellier Cedex 05 – France (E-mail: [email protected])
*** Agence de l'Eau Rhône Méditerranée Corse, 2-4 allée de Lodz, 69363 Lyon Cedex 07 – France (E-mail: [email protected])
Abstract:
Study of the hydraulic limits of reed beds, based on the knowledge of hydrodynamics in unsaturated porous media, shows the ability of the system to accept flow overloads. Measuring different parameters (flow, pollutant removal, infiltration rate (IR), pressure head profiles) in pilot and full-scale studies, we suggest new hydraulic limits with accompanying sizing rules and operational recommendations according to the level of deposit on the filter surface. Overloads of ten times the dry weather flow are possible whilst still complying with the European standards.
Keywords:
Vertical flow constructed wetlands; hydraulic overload; hydraulic behaviour.
INTRODUCTION
Vertical flow constructed wetlands (VFCWs) have been very successful in France over the last five years. This fact is due to aspects such as the good performances obtained for SS, COD and nitrification (Molle, 2004) and the acceptance of raw water on the first stage filters leading to easier sludge management. If we add the low operation costs, it is easy to understand this choice by communities of less than 3000 p.e. The sizing of VFCWs is still on roughly based organic load acceptance, (in terms of active area per people equivalent [p.e.] with slight clear water intrusion into the sewerage system). Nevertheless these assumptions are often wrong in the context of many small French communities where infiltration or storm water can considerably increase the organic load.
This can possibly alter the biological activity through a reduction of oxygenation of the media. If hydraulic overloads are often observed in such systems it is difficult to estimate the exact limits in order to secure durable filter operation. To specify the hydraulic limits would be of great help to Water Authorities in deciding at what point is it preferable to renovate the sewerage network rather than adapt the wastewater treatment plant. This would also permit precise sizing for storm water treatment.
With this in mind a study using pilot and full-scale experiments was carried out to follow hydraulic fluxes and their impact on pollutant removal. A 200 p.e. plant, in operation since 1994, was studied over two years. Knowing that filters are accumulative systems, different sewage plants were studied to observe the impact of the age of the plant (and thus biomass and deposit accumulation levels) on hydraulic capacity and removal efficiency under hydraulic overload conditions.
MATERIALS AND METHODS
Pilot experiments (samples from full scale reed beds) were used to define the hydraulic and biological mechanisms under the influence of hydraulic overloads more precisely than would be possible from full-scale studies. Full-scale studies were used to define the deposit layer, biomass development and effect of reeds on hydraulic capacities. Intermittent batch feeding was used on each plant and pilot filter.
Pilot Study
An investigation was carried out on a 20cm diameter column consisting of a core sample from the first stage filter of the Colomieu plant (8 years in operation). From the top to the bottom there was 4cm of deposit layer (DM: 30.2%; OM: 37.6% of the DM); 37cm of gravel (d10: 1.94mm; d60: 3.65mm) and 13cm of 5/10 mm gravel to drain the column. Two periods of two months were used to evaluate the behaviour of the media. Feeding was done with tap water for the first period and settled wastewater for the second. Hydraulic loads applied ranged between 1 and 1.4m.d-1 with organic loads of 250 ± 70g.COD.m-2.d-1 for settled wastewater. Three feeding rhythms were tested (2.4cm of water per 30 min., 4.8cm per hour and 9.5cm per 2 hours) to observe the impact on operational mode, hydraulic acceptance and treatment efficiency.
! Hydraulic measurements: The column was equipped with an electronic balance (manufactured by Thermonobel) in order to follow differences in mass due to humectation/drainage and biomass variations. Inflow and outflow measurements were also made via electronic balances (figure 1). The balance measurements are collected through data acquisition modules connected to a computer via a RS-485 plug. The accuracy for the column mass monitoring is about +/-5g.
Infiltration rates (IR) were followed by measuring the level of temporary excess surface water.
Micro tensiometers (SDEC 220) were used to measure the pressure potential each 10 cm. Pulsed periodic flow tracer experiments (Fig 2) were carried out using NaCl and conductivity measurements at the outlet of the column.
Outlet
Data acquisition
3°C 7°C
Wastewater tank 4°C
measure of the level of temporary excess surface water
Outlet
Data acquisition
3°C 7°C
Wastewater tank 4°C
measure of the level of temporary excess surface water
Tracer injection
Time
Inflow rate
Batch feeding Tracer injection
Time
Inflow rate
Batch feeding
Figure 1: Column experimental scheme Figure 2: Pulsed periodic tracer scheme
! Chemical measurements: Biological behaviour of the column was followed by classical chemical analysis (COD, dissolved COD, SS, TKN, N-NO3, N-NH4). All analyses were done by French standard methods in the Cemagref laboratories. Measurements were done on 12h samples, but efforts were also made to monitor treatment efficiency variation due to the frequency of batches.
Full scale experiments
Three plants were studied to confirm the hydraulic capacity of VFCWs (Gensac la Pallue, 1700 p.e., 14 years of operation; Colomieu, 200 p.e., 8 years of operation; Evieu, 200 p.e., first year of operation). With the exception of Gensac la Pallue (8 VFCWs in parallel at 1st stage + 3 ponds in series) the plants consist of two stages VFCWs designed according to Cemagref recommendations (Molle et al., 2004).
The Colomieu plant was followed over a two-year period. Due to the combined sewerage system the hydraulic overloads were of great amplitude and, furthermore, could be created and/or increased with river water. There were two periods: the first to study episodic overloads (1.5year), and the second continuous overloads (5months). The first period was used to study the plant in natural conditions i.e. no clean water was added to the influent. Only natural storm events were studied.
During the second period, river water was continuously added to the batch-pumping unit (5 times the dry weather flow, about 1.8m.d-1 on the first stage filter in operation) with the aim of observing the hydraulic acceptance limit and/or decrease in aerobic biological activity. Batch frequency was differed between the three parallel filters by varying the functioning time of the pump used to add
river water to the network. The pump was used with three functioning/rest periods: 0.5/0.5; 1/1; 2/2 hours.
Evieu plant was followed over 6 months (March to August) with dry weather and added river water periods to observe the hydraulic behaviour during the first year of operation when there is very few deposit layer on the filter and biomass development is poor.
Gensac la Pallue plant was studied for one week in March to measure the IR in a filter with a deposit layer of 25 cm.
! Hydraulic measurements: For the column, inlet and outlet flow were continuously measured by acquisition of pump functioning time and venturi weirs. IRs were followed by measuring the level of the temporary excess surface water level with ultrasound probes. Tensiometers (SKT 850 from SDEC) were used to measure the pressure potential under the deposit layer at depths of 10cm and 20cm on a first stage filter in Colomieu.
! Chemical measurements: Treatment efficiency was measured by 24-hour flow composite sampling at different times of the year. Each stage of the treatment plant was evaluated for COD, BOD5, SS, TKN, N-NH4, N-NO3, TP and PO4-P according to French standard methods.
RESULTS AND DISCUSSION
IRs evolve with the feeding of the filter. A decrease in IR is observed during the first days of feeding until stabilisation is reached (Fig 3). This decrease is due to the moisture saturation of the filter, which changes the pressure gradient, and to the deposit and biomass accumulation in the upper layers of the filter. As a result of pressure head measurements, it can be stated that the filter always stays in an unsaturated condition even when batches are passing through the filter. Surface deposits on the first stage hinder infiltration. Despite coarser particle size on the first stage (d10 = 1.94mm; d60 = 3.65mm) than the second stage (d10 = 0.22mm; d60 = 1.13mm), the first stage is more limiting due to SS deposit and biomass development. Reed development over the year induces an increase in infiltration capacity (Fig 3). We can observe a relationship between the excess surface water level and the IR (Fig 4) that progresses with hydraulic load and days of feeding. This relationship is a IR = a.exp(b.h) type where a and b are constants and h is the surface water level.
For constant hydraulic loads only the a constant changes until equilibrium. Increasing the hydraulic load will change the b coefficient (the slope in fig 4).
1,E-05 1,E-04 1,E-03
0 2000 4000 6000 8000 10000 12000
time (min)
infiltration rate (m/s)
02-02 03-02 06-02 08-02 09-02 10-02
Febru March June Augu Sept Oct
Q = 2.3E-06e0.0411h R2 = 0.90 Q = 8,8E-07e0.0615h
R2 = 0.90 y = 1.2E-06e0.065h
R2 = 0.88
1,E-05 1,E-04 1,E-03
0 20 40 60 80 100 120 140
Temporary excess surface water level (mm)
infiltration rate (m.s-1)
Day 3 Day 4 Day 6 Day 9 storm event
Figure 3: IR changes in a first stage filter for the Colomieu plant during the year 2002
Figure 4: Relationship of excess water level to IR in the Colomieu plant - Sept 2002 with a hydraulic overload of
1.44m.d-1 during a storm.
Reed development has an impact on the evolution of the a and b parameters which is explained by Molle (2003). Reed development has the opposite effect on IR. From approximately one meter of height, reeds allow water to pass down the stems via the tubular spaces formed around them by their oscillation. In this way, they reduce the braking role of the deposit layer. On the other hand, water can moisten the lower layer more and so reduce the pressure gradient, resulting in a slowing down
of IR. So when reeds are absent (harvested) the deposit layer is the major element controlling the IR. A high hydraulic overload induces an excess of surface water while IR is controlled by the hydraulic conductivity of the deposit layer. Reeds increase this conductivity and allow a better IR but under high hydraulic overload moisture in the media is greater and penetrates deeper and so the pressure gradient is reduced which slows the IR. It appears that rest period and batch frequency also have an impact on the IR.
0 0,2 0,4 0,6 0,8 1 1,2 1,4
0 1000 2000 3000 4000
Time (min) mass variation of the column (kg)
3 L/ 2 h 1,5 L/ 1 h 0,75 L/ 30 min
20 30 40 50 60 70 80
0,4 0,6 0,8 1 1,2
Water retention in column at the begining of tracer experiments (litre) water proportion of the tracer batch at the outet for the first batch [%]
0.75 L / 30 min 0.75 L / 30 min 1.5 L / 1 h 1.5 L / 1 h 3 L / 2 h 3 L / 2 h
Evolution with biomass development
Figure 5: Residual water content during feeding for different
batch frequencies on column experiments. Figure 6: Dilution of the tracer batch for batch frequencies for the column experiments.
In fact greater spacing out the feeding with the same hydraulic load allows better drainage of the media and residual water content is less (Figure 5). Consequently we observed better IRs because pressure gradients are greater (greater submersion and drying of the media due to longer rest periods between two feedings). Nevertheless, spacing out feeding, and using a bigger volume of water for each batch, can lead to poorer pollutant removal because of a large volume and short contact time with the biomass. Water exchanges in the column are poorer with high volume batches (figure 6).In fact, as measured in pilot and full-scale studies, pollutant removal behaves differently for COD removal or nitrification. Table 1 is a summary of filter behaviour and pollutant removal as a consequence of batch frequency.
Table 1: Relationship between hydraulic and biological behaviour
Batch frequency Hydraulic behaviour Biological behaviour
More spaced out Less water content, better IR, shorter contact time
Better oxygenation COD removal is poorer Nitrification is greater due to NH4+
adsorption and nitrification during rest period
Closer together Greater water content and effective volume
of reaction, IR slower Less oxygenation which disadvantages nitrification. COD removal good
30 40 50 60 70 80 90 100
0 20 40 60 80
Time (h)
Removal (%)
0.75/0.5h 1.5/1h 3/2h
C DGO = 214 +- 15 g.m-2.j-1
y = -21.462x + 83.911 R2 = 0.9866 0
20 40 60 80 100
0 0,5 1 1,5 2
Ch (m.j-1)
R%
Figure 7: Nitrification efficiency on column experiments under hydraulic overload (1.2m.d-1, N-NH4
inlet=27±4mg.l-1) for different batch frequencies.
Figure 8: Global Oxygen Demand (dissolved COD + 4.57*TKN) removal on Colomieu filter for identical
organic load and different hydraulic overloads.
This batch frequency effect on oxygenation can be observed in Fig 7. Nitrification decreases with time, for identical hydraulic overloads when batches are less spaced out. Of course increasing the hydraulic overload will decrease oxygenation potential and therefore efficiency of pollutant oxidation (Fig 8).
Despite the decrease in removal efficiency, outlet concentrations measured on full-scale plants under hydraulic overload are still low due to the influent dilution effect. For all experiments, the filters maintained good output concentrations (COD < 50 mg.L-1, SS < 5 mg.L-1, NK < 3.5 mg.L-1).
CONCLUSION
The Colomieu study shows that episodic overloading of up to 4 meters per day (> 10 times the dry weather hydraulic flow) and continuous overloads over 5 months of 1.8 meter per day (5 times the dry weather hydraulic flow) can pass through the filters, be treated and still respect the European quality objectives. The first stage of treatment is the hydraulic limiting factor due to the high deposit layer and biomass development. We demonstrate the role of batch frequencies on infiltration capacity and pollutant removal. For hydraulic acceptance, March is the worst period because of the deposit accumulation during winter with poor mineralisation and the absence of reeds. The IRs observed on Evieu (1cm of deposit level, > 0.6.10-4m.s-1), Colomieu (7cm of deposit level,
> 0.6.10-4m.s-1) and Gensac la Pallue (25 cm of deposit level, > 0.3.10-4m.s-1) show that the deposit height has a relative effect on infiltration capacities. In fact it is the new deposit layer, poorly mineralised, which is hydraulically limiting, and not the total height of the deposit. We propose new hydraulic limits for vertical flow reed beds (French configuration) based on the following considerations, IRs observed during critical periods, deposit height on the first stage and the need to ensure an oxygenation period for the surface filter at least half of the time in order to maintain aerobic conditions (table 2).
Table 2: Proposed new hydraulic limits Deposit layer
(cm) 0-10 cm 10-25 cm
Hydraulic
overload Once a week Once a month Once a week Once a month
m.d-1 1.80 3.50 0.90 1.80
m.h-1 0.25 0.25 0.11 0.11
These proposed hydraulic limits deal with diluted influent due to clearwater intrusion or combined sewers. These limits have not been established for organic overloads which could damage the microbial activity over a long period. Acceptance of such hydraulic overloads must be accompanied by design recommendations such as a sufficient border height to allow for temporary flooding of the filter surfaces.
These recommendations are first suggestions, to try to reassure people that reed bed filters will function under hydraulic overloads. More precise design criteria will be defined after more studies on different sewage plants.
ACKNOWLEDGEMENTS:
The authors would like to thank the "Rhône Méditerranée & Corse" water authorities for their financial support, and Helen Burnett for help provided.
REFERENCES:
Molle P., Liénard A., Boutin C., Merlin G., Iwema A. (2004) How to treat raw sewage with constructed wetlands:
An overview of the French systems. In 9th International Conference on wetlands Systems for water Pollution Control, Sept 27-30, 2004 Avignon (France).
Molle P., (2003). Subsurface flow constructed wetlands: Phosphorus retention and hydraulic limit of vertical subsurface flow CWs. (In French) PhD Thesis, engineering processes speciality, University of Montpellier, 267 p.
