Content uploaded by Tim D Fletcher
Author content
All content in this area was uploaded by Tim D Fletcher on Feb 10, 2015
Content may be subject to copyright.
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
Blecken et al. 1
Heavy metal removal by stormwater biofilters:
can it withstand alternative wetting and drying conditions?
Godecke-Tobias Blecken1*, Yaron Zinger2, Ana Deletić2, Tim Fletcher2, Maria Viklander1
1 Urban Water, Department of Civil, Mining and Environmental Engineering,
Luleå University of Technology, 971 87 Luleå, SWEDEN
2 Facility for Advancing Water Biofiltration, Department of Civil Engineering,
Monash University, Victoria 3800, AUSTRALIA
*Corresponding author, e-mail godble@ltu.se
ABSTRACT
Urban stormwater contains substantial loads of Cu, Pb and Zn, which are considered as key
stormwater contaminants. Stormwater biofiltration is a promising option to treat these
contaminants. Biofilters are exposed to an alternate cycle of drying and wetting, and the
influence of this on pollutant removal performance is as-yet unknown. To investigate the
effect of drying and subsequent rewetting on the retention of heavy metals by stormwater
biofilters, a laboratory study has been conducted using three groups of biofilter columns,
which were dosed with semi-synthetic stormwater according to three different drying and
wetting regimes. Some biofilters were fitted with a submerged zone combined with a carbon
source, at the bottom of the filter. Overall, the biofilters were very effective in heavy metal
removal, provided that they received regular stormwater input. However, after drying
extending to three or four weeks, removal of heavy metals decreased significantly. A
statistically significant correlation between antecedent dry days and metal removal was
shown. Furthermore, a clear effect of the submerged zone was found: after extended dry
periods, biofilters with this feature performed significantly better than those without it. In
particular, the removal of Cu was clearly increased both during wet and dry periods; for Pb
the negative effect of drying was completely eliminated by introducing a submerged zone.
KEYWORDS
Biofiltration, Drying and rewetting, Heavy metals, Stormwater; Submerged zone
INTRODUCTION
Urban stormwater is a major source of water pollution in urban environments (Marsalek et al.,
1999). Biofilter technologies have a high potential to remove pollutants from stormwater
before it is discharged to receiving waters. Thus, stormwater biofilters can contribute to the
development of sustainable urban water management options, involving treatment and
possibly even harvesting of stormwater (Hatt et al., 2007a).
A typical biofilter consists of a vegetated swale or basin, underlain by a filter medium, e.g.
soil or gravel, and a drainage pipe on the bottom (figure 1). Stormwater infiltrates, is filtered
by the vegetation and filter media, and the treated water is collected in the drainage pipe
(Melbourne Water, 2005; Prince George's County, 2002). Several studies have shown that
biofilters have the capacity to treat stormwater effectively; removal of Cu, Pb and Zn in
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
2 Heavy metal removal by stormwater biofilters: effect of alternative wetting and drying
excess of 90% has frequently been shown (Davis et al., 2003; Hatt et al., 2007a; Lau et al.,
2000; Sun et al., 2007).
However, stormwater events are very variable due in their frequency. Stormwater biofilters
are thus exposed to an alternating regime of wetting and drying, for varying durations.
Despite this, most biofilter laboratory studies have provided only results based on stormwater
application at constant intervals, and with constant volumes (Davis et al., 2001; Davis et al.,
2003; Lau et al., 2000; Scholz, 2004). Thus, the effect of altering drying and wetting on the
treatment efficiency has not been taken into account, thus paying no attention to ‘real-world’
conditions. Better knowledge of biofilter performance under more natural conditions is
essential to ensure that biofilters are designed in a way that maximises the efficiency and
reliability of treatment.
Until now, only Hatt et al. (2007b) tested the effect of drying and wetting on pollutant
removal, but only for unvegetated stormwater biofilters. No significant effect of drying and
re-wetting on metal treatment was observed. However, drying processes and associated
oxidation of sediment have a significant effect on the chemical phase distribution of metals in
soils or sediments (Saeki et al., 1993), leading to increased mobilisation of the metals
(Förstner et al., 1998).
In this paper the preliminary results of a laboratory study of vegetated biofilter systems under
three different varying drying and re-wetting regimes are presented. We investigated if there
is an effect of extended drying on the heavy metal removal during subsequent storm events.
Furthermore, we determined if a submerged zone at the bottom of the filter has an effect on
the metal removal, by maintaining a more constant soil moisture regime in the biofilter.
Contrary to the findings of Hatt et al. (2007b), it was found that extended drying caused
decreased metal removal from stormwater. However, even after drying of up to seven weeks,
between 70% and 90% of metals were removed in the filter. Most importantly, the effect of
drying could be minimised for Cu and even eliminated for Pb by a permanent submerged
zone at the bottom of the filter media.
METHODS
Experimental set-up
Laboratory tests were conducted on 18 biofilter columns (mesocosms) made of PVC
stormwater pipes (inner diameter: 375mm, area: 0.11 m2, height: 900 mm; Figure 1). A
transparent perspex pipe (height: 400 mm) glued on top allowed water to pond, without
affecting the plant’s light availability. The columns were randomly placed in a greenhouse
with open mesh on the sides, but a roof to exclude rainfall (so that a mass balance of inflows
and outflows could be maintained).
The filter media in the columns included four layers (listed from top, Figure 1a):
• Top layer, 400 mm, sandy loam, with vegetation (seven plants per column of Carex
appressa R.Br. (Tall Sedge),
• Bottom layer, 400 mm, fine sand,
• Transition layer, 30 mm, coarse sand,
• Drainage layer, 70 mm, fine gravel, with an embedded drainage pipe at the bottom which
leads to a sampling outlet.
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
Blecken et al. 3
In 12 columns a submerged zone (hereinafter called ‘SZ’) of 450 mm was created by
elevating the outflow port using an external riser pipe (Figure 1). In the SZ, 800 g of a carbon
source (hereinafter called ‘C’) consisting of 1/3 pea straw and 2/3 red gum wood chips, was
added. The primary purpose of this was to provide optimal conditions for denitrification, as
part of a parallel study on the optimisation of nitrogen removal (Zinger et al., 2007).
Vegetation:
C. appressa
Sandy
loam
outflow with SZ
Sand
outflow no SZ
Figure 1. Biofilter columns: (a) configuration and filter media set up, (b) columns in
greenhouse (transparent top, vegetation, riser pipe)
Experimental procedure
Drying / Wetting. The 18 columns were divided into three groups with six columns each,
which were dosed as follows:
• Group A: 1 week wet / 7 weeks dry / 4 weeks wet
• Group B: 1 week wet / 1 week dry / 1 week wet / 2 weeks dry / 2 weeks wet /
3 weeks dry / 2 weeks wet
• Group C: 1 week wet / 4 weeks dry / 2 weeks wet / 3 weeks dry / 2 weeks wet
During wet periods the columns were dosed with 25 L stormwater each twice weekly based
on the following assumptions: biofilter area of 2.5% of catchment area; effective precipitation
(after initial losses) of 590 mm per year, 104 rainfall events per year (i.e. twice weekly).
During dry periods the columns received no inflow at all. Each group consisted of two
replicate standard columns without a submerged zone and without carbon source, and four
replicate advanced columns with both a submerged zone and carbon source.
Semi-synthetic stormwater. Semi-synthetic stormwater was utilised, since natural stormwater
was not available in the required quantity and consistency over the time of the experiment. It
was made by adding real stormwater sediment from a stormwater pond to dechlorinated
(using Sodium thiosulfate) tap water, and topping up with laboratory-grade chemicals as
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
4 Heavy metal removal by stormwater biofilters: effect of alternative wetting and drying
required to replicate typical urban stormwater event mean concentrations (Duncan, 1999) as
given in Table 1 were achieved.
Table 1. Selected mean pollutant inflow concentration over the entire experiment duration
and their sources in the semi-synthetic stormwater
Pollutant Mean stormwater
concentration
Chemicals used for topping up
of existing sediment concentrations
as required
Total susp. solids (TSS) 154 .2 mg/L stormwater pond sediment (≤300µm)
Copper (Cu) 95.1 µg/L copper sulphate (CuSO4)
Lead (Pb) 199.1 µg/L lead nitrate (PbNO3)
Zinc (Zn) 587.3 µg/L zinc chloride (ZnCl)
Sampling. For each stormwater dosing, composite inflow samples (for all 18 columns) were
taken in two replicates to monitor the inflows over the entire duration of the experiment.
Additionally, 1 L outflow composite samples were taken at most events.
Water quality analyses. All samples were taken in polyethylene bottles and then digested with
nitric acid according to standard methods (APHA/AWWA/WPCF, 1998). The samples were
analysed for the total concentrations of Cu, Pb and Zn using an ICP-OES (Varian 720-ES).
The device detection limits were 0.3 µg/L for Cu, 0.6 µg/L for Pb and 0.5 µg/L for Zn.
Soil moisture content. In each group, the moisture content θ (%) of one column with and one
without SZ and C was measured using ML2 ThetaProbes (Delta-T Devices Ltd. Cambridge,
UK), placed 225 mm under the soil surface in the top layer.
Data analyses
Removal efficiencies. In this paper Cu, Pb and Zn removal data were analysed because these
metals are regarded as key stormwater pollutants. The removal was calculated as a % of the
inflow concentrations: removal (%) = (1 - out(µg/L) / in(µg/L)) · 100%.
Treatment during wet periods. To assess the overall performance of the filters (regardless of
drying) and to compare it with other studies (with regular stormwater applications), analysis
of only sampling events during wet periods with a maximum of three antecedent dry days
(ADD) was undertaken; based on residence time of water in the filter, it is known that after
this time, the filter media will still retain relatively high levels of soil moisture.
Effect of drying and SZ on metal removal. To investigate the effect of drying on metal
removal, a regression analysis between removal (%) and the length of drying before a storm
event (ADD) was performed separately for columns with and without SZ. Accordingly, fitted
line plots of removal vs. ADD were compiled. Furthermore, the effect of SZ on the treatment
performance after drying was tested.
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
Blecken et al. 5
RESULTS AND DISCUSSION
Metal treatment regardless of drying
During wet periods (with a maximum of three ADD) Pb and Zn removal in the biofilter
columns is consistently very high (Figure 2) for all columns, both with and without C and SZ.
Cu removal is very good as well, however with significantly worse treatment in columns
without SZ and C (Figure 2). The very effective treatment corresponds to the findings of other
studies of biofilters (Davis et al., 2003; Lau et al., 2000; Sun et al., 2007). The significant
influence of C and SZ on Cu removal confirms in general the results of a previous study
undertaken with the same columns used for these experiments (Blecken et al., submitted).
The positive effect of C on Cu treatment is explained by adsorption of Cu by C (formation of
Cu-organic matter complexes, Reuter et al., 1977). In contrast to Cu, other metals have less
affinity to organic matter (Yin et al., 2002). With introduction of SZ, pH in the outflow water
increases (data not reported in this paper) and, thus, metal solubility decreases. Furthermore,
it is assumed that the submerged zone contains pockets of both aerobic and anaerobic
conditions, facilitating both nitrification and denitrification (as shown by Zinger et al. (2007)
the latter would not occur in oxic conditions). Higher metal sorption of sediments under
anoxic conditions (Bradl, 2004) might explain the improved treatment of metals in the SZ.
However, the effect of pH is considered to be stronger because pH is the main factor
influencing metal sorption (Bradl, 2004).
80
85
90
95
100
1
Removal (%)
Cu without SZ+C Cu with SZ+C Pb Zn
Figure 2. Overall removal of key metals during wet periods; influence of C and SZ on Cu
removal.
Effect of drying on metal treatment
Metal removal (%) is generally reduced after longer periods of drying. While there is no
obvious effect after one and two weeks of drying, a clear influence of drying in excess of
three to four weeks can be seen. Only Pb removal in columns with SZ and C shows no
dependency on drying (Table 2).
After drying the biofilters recover quite quickly and often outflow concentrations reach their
initial extend already one or two dosings after long drying. However, in climates where there
are prolonged dry periods, interspersed with isolated rainfall events, the influence of drying
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
6 Heavy metal removal by stormwater biofilters: effect of alternative wetting and drying
on the overall metal removal will be significant. Importantly, where achievement of receiving
water concentration targets is a primary consideration, the influence of drying may be critical.
Table 2. Mean metal removal during wet periods and after drying
Mean metal removal (%) ± Standard deviation
after drying
Metal and
column design wet (1) 1-2 weeks 3 weeks 4 weeks 7 weeks
Cu with SAZ + C 98.8 ± 0.2 98.8 ± 0.3 93.0 ± 4.7 91.3 ± 5.7 85.2 ± 2.3
Cu without SAZ + C 92.4 ± 2.1 90.1 ± 1.0 80.9 ± 4.1 82.1 ± 0.0 72.1 ± 0.4
Pb with SAZ + C 98.9 ± 0.5 99.0 ± 1.0 99.8 ± 0.1 99.2 ± 0.3 99.0 ± 0.1
Pb without SAZ + C 99.0 ± 0.4 97.7 ± 2.4 97.1 ± 0.3 94.2 ± 0.0 90.8 ± 1.5
Zn with SAZ + C 98.5 ± 0.4 98.7 ± 0.3 97.7 ± 0.8 96.2 ± 1.4 87.6 ± 2.4
Zn without SAZ + C 98.2 ± 0.7 97.4 ± 0.2 96.7 ± 1.3 96.6 ± 0.0 92.6 ± 0.4
(1) only results of the first two stormwater applications (first week of the run-time)
Regression analysis between removal (%) and length of drying before a storm event
(antecedent dry days, ADD) shows a strong linear (or quadratic) relationship. Where
significant (ie. p<0.05), regression coefficients R2 (adjusted) were all in excess of 73%
(Figure 3). However, these results must also be placed in the context of their practical
implications; even after 7 weeks of drying (and in the worst case, i.e. filters without SZ and
C) still around 70% of Cu, 90% of Pb and 85 to 90% of Zn were removed (Figure 3).
Figure 3. Biofilter performance affected by length of drying: metal removal after drying vs.
length of drying period (antecedent dry days). Linear or quadratic regression line provided
where significant. P-values and R2(adj) of the regression analysis.
70
75
80
85
90
95
100
0 7 14 21 28 35 42 49
Antecedent dry days
Cu removal (%)
85
90
95
100
0 7 14 21 28 35 42 49
Antecedent dry days
Pb removal (%)
85
90
95
100
0 7 14 21 28 35 42 49
Antecedent dry days
Zn removal (%)
Cu +anox. zone Cu -anox. zone
Pb +anox. zone Pb -anox. zone
Zn +anox. zone Zn -anox. zone
p=0.508
R
2(adj)=0.0%
p=0.000
R
2(adj)=73.4%
p=0.000
R
2(adj)=91.7%
p=0.000
R
2(adj)=86.6%
p=0.000
R
2(adj)=85.1%
p=0.000
R
2(adj)=93.4%
(quadratic regression)
Cu with SZ + C
Pb with SZ + C
Zn with SZ + C
Cu without SZ + C
Pb without SZ + C
Zn without SZ + C
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
Blecken et al. 7
The strong negative correlation between treatment performance of the biofilter and the length
of antecedent drying might be explained by the effects of the decreasing moisture content in
the filter media during drying. As a result of this process, accumulated metals in the filter are
likely to be oxidised and, thus, washed out at the next stormwater event. Oxidation of
(initially anoxic) sediments is estimated to be the most efficient way to mobilise metals into
the environment (Förstner et al., 1998). The behaviour could also be explained in part by
physical changes in the media after drying. It is likely that long periods of drying result in
small fissures appearing in the media, thus creating preferential flow paths and diminishing
the trapping efficiency of particulate and dissolved material (Hatt et al., 2007b). Only Pb
removal in biofilters with SZ was not correlated to antecedent drying at all (very high removal
rates over 98% were always achieved) while without SZ Pb treatment was strongly correlated
to ADD, decreasing as the dry period became longer (Figure 3).
Enhanced treatment after drying by introducing SZ
Introduction of a submerged zone (in these experiments combined with C) maintains optimal
Cu and Pb treatment, even after drying periods in excess of three weeks (Figure 3). Soil
moisture content, measured in the top layer (thus above the actual SZ) was around 5% higher
in columns with SZ, both during wet and dry periods compared to columns without these
features (Figure 4). Thus, less oxidation of the filter media during dry periods might occur in
biofilters with SZ even above the actual SZ. This might explain less leaching of already
accumulated metals after drying from these biofilters (Tack et al., 1998; Tack et al., 2006).
Because the reported values were measured relatively close to the top, it is, thus, assumed that
the significant moisture content decrease due to drying (Figure 4) flattens by depth.
Figure 4. Soil moisture content vs. length of drying period (antecedent dry days). Quadratic
regression line, p-values and R2(adj) of the regression analysis. Influence of SZ (columns
without SZ and C: dashed line, white markers; columns with SZ and C: continuous line, black
markers).
0
5
10
15
20
25
0 7 14 21 28 35 42 49
Antecedent dry days
Soil moisture content (%)
p=0.002
R2(adj)=72.5%
p=0.003
R2(adj)=84.4%
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
8 Heavy metal removal by stormwater biofilters: effect of alternative wetting and drying
CONCLUSION
The analysis of the preliminary results of this study show that:
• similar to several similar previous studies, metal treatment in stormwater biofilters during
wet periods is excellent, with removal exceeding 95% in most cases,
• there is a significant effect of drying (in excess of 3 to 4 weeks) on the filter performance,
resulting in significantly decreased metal removal,
• introduction of a submerged zone combined with a carbon source can reduce (Cu removal)
or even eliminate (Pb removal) this negative effect of drying.
The results of this study suggest that design of biofiltration systems for metals removal should
take into account the criticality of variations in concentrations in the outflow. Where small
variations in performance are permissible, ‘standard’ biofilters will perform acceptably well.
However, where consistent reductions of at least 95% of inflow concentration are required,
the presence of a saturated zone at the base of the filter is recommended.
REFERENCES
APHA/AWWA/WPCF. (1998). Standard Methods for the Examination of Water and Wastewater, 20th edn.
American Public Health Association / American Water Works Association / Water Pollution Control
Federation, Washington, DC, USA.
Blecken, G.-T., Zinger, Y., Deletic, A., Fletcher, T. D. & Viklander, M. (submitted). Impact of an anoxic zone
and a carbon source on heavy metal removal in stormwater biofilters. Submitted to Ecological
Engineering.
Bradl, H. B., 2004. Adsorption of heavy metal ions on soil and soil constituents. Journal of Colloid Interface
Science, 277, 1-18.
Davis, A. P., Shokouhian, M., Sharma, H., & Minami, C. (2001). Laboratory Study of Biological Retention for
Urban Stormwater Management. Water Environment Research, 73(1), 5-14.
Davis, A. P., Shokouhian, M., Sharma, H., Minami, C. & Winogradoff, D. (2003). Water quality improvement
through bioretention: Lead, copper, and zinc removal. Water Environment Research, 75(1), 73.
Duncan, H. P. (1999). Urban Stormwater Quality: A Statistical Overview (No. Report 99/3). Melbourne,
Australia: Cooperative Research Centre for Catchment Hydrology.
Förstner, U., Ahlf, W. & Calmano, W. (1998). Studies on the Transfer of Heavy Metals between Sedimentary
Phases with a Multi-Chamber Device: Combined Effects of Salinity and Redox Variation. Marine
Chemistry, 28, 145-158.
Hatt, B. E., Deletic, A. & Fletcher, T. D. (2007a). Stormwater reuse: designing biofiltration systems for reliable
treatment. Water Science and Technology, 55(4), 201-209.
Hatt, B. E., Fletcher, T. D. & Deletic, A. (2007b). Hydraulic and pollutant removal performance of stormwater
filters under variable wetting and drying regimes. Water Science and Technology, 56(12), 11-19.
Lau, Y. L., Marsalek, J. & Rochfort, Q. (2000). Use of a Biofilter for Treatment of Heavy Metals in Highway
Runoff. Water Quality Research Journal of Canada, 35(3), 563-580.
Marsalek, J., Rochfort, Q., Brownlee, B., Mayer, T. & Servos, M. (1999). An exploratory study of urban runoff
toxicity. Water Science and Technology, 39(12), 33-39.
Melbourne Water (2005). WSUD Engineering Procedures: Stormwater. CSIRO Publishing.
Prince George's County (2002). Bioretention Manual. Department of Environmental Resources, Programs &
Planning Division, Prince George's County, Maryland, USA.
Reuter, J. H., & Perdue, E. M., 1977. Importance of heavy metal-organic matter interactions in natural water.
Geochimica Cosmochimica Acta, 41, 325-334.
Saeki, K., Okazaki, M. & Matsumoto, S. (1993). The chemical phase changes in heavy metals with drying and
oxidation of the lake sediments. Water Research, 27(7), 1243-1251.
Scholz, M. (2004). Treatment of gully pot effluent containing nickel and copper with constructed wetlands in a
cold climate. Journal of Chemical Technology and Biotechnology, 79, 153-162.
Sun, X. & Davis, A. P. (2007). Heavy metal fates in laboratory bioretention systems. Chemosphere, 66(9), 1601-
1609.
Tack, F. M. G., Singh, S. P. & Verloo, M. G. (1998). Heavy metal concentrations in consecutive saturation
extracts of dredged sediment derived surface soils. Environmental Pollution, 103, 109-115.
11th International Conference on Urban Drainage, Edinburgh, Scotland, UK, 2008
Blecken et al. 9
Tack, F. M. G., Van Ranst, E., Lievens, C. & Vanderberghe, R. E. (2006). Soil solution Cd, Cu and Zn
concentrations as affected by short-time drying or wetting: The role of hydrous oxides of Fe and Mn.
Geoderma, 137, 83-89.
Yin, Y., Impellitteri, C. A., You, S.-J., & Allen, H. E., 2002. The importance of soil organic matter distribution
and extract soil: solution ratio on the desorption of heavy metals from soils. The Science of the Total
Environment, 287, 107-119.
Zinger, Y., Fletcher, T. D., Deletic, A., Blecken, G.-T., & Viklander, M. (2007). Optimisation of the nitrogen
retention capacity of stormwater biofiltration systems. Proceedings of NOVATECH 2007, Lyon,
France.