Council archive · source document
Jellyfish Compliance with SQIDEP · MCU15/0122
Attached to the e-mail MCU15/0122 - Response to Information Request -Stormwater (Jellyfish compliance) Email 3 of 3.
JellyFish® Filter Compliance with SQIDEP [OCR of page images] JellyFish® Filter Compliance with SOIDEP ee
See the original
The page text is free to read. Viewing or downloading an original file needs an account.
Log in to see the originalFull text of the other 27 pages
Page 2
Requirements Field Evidence Criteria Humes Ipswich Test Site PASS/FAIL
Sampling Events
Type of Event Rainfall Events1 Yes PASS
Minimum Number of Events Statistically relevant. Minimum of 10-15 (to be 10 events published and statistically relevant
PASS
confirmed) (see Appendix B).
Minimum Rainfall Depth Total event rainfall depth ≥2mm Only recorded if minimum is 2.6mm (see
PASS
Appendix A)
Minimum/ Maximum Storm Indicative 1 hour – Minimum recorded was 19min storm.
Duration/ Volume Importantly, the minimum storm event should relate Maximum recorded was 20 hour storm.
to the hydrograph and include a consideration of Given the small size of the site, the minimum PASS
catchment characteristics and may need to be event can be accepted as majority of runoff
adjusted based on site selection would have been recorded (see Appendix A)
Minimum Inter-event Time 72 hours. 2* (see discussion points following) 3 days/72 hours (see Appendix A) PASS
Device Size Full Scale Full Scale (see Appendix A) PASS
Runoff Characteristics Target pollutant profile of influent and effluent Influent and Effluent samples were tested not
only for TSS, TP and TN, but a range of heavy PASS
metals and other pollutants (see Appendix A)
Runoff Volume or Peak Flow Runoff at least 3 events should exceed 75% of the The treatment flowrate for a JF1200-2-1 is
design water quality volume/ treatment flow rates of 12.5L/s. Peak runoff through the device was at
the design & 1 event greater than the design flow least 75% or greater of the treatment flowrate
PASS
(9.4L/s) for 6 events. These 6 events also were
greater than the full treatment design flowrate
(see Appendix C)
1
Must not Include Controlled Field Tests. See glossary for the definition of controlled field tests.
2
Includes Controlled Field Test. See glossary for the definition of controlled field tests
3
from 1 site or minimum of 8 per site if more than one site is used
Page 3
Sampling Procedures and Techniques
Automated Sampling Composite samples on a flow weighted basis Two ISCO 6712 Automatic Samplers (one each
PASS
to measure influent and effluent)
Minimum Number of Aliquots 8 per event* (see discussion points in SQIDEP) All events except 3 achieved 8 or more
samples. Number of samples collected was
automatically selected based on the intensity
of the rainfall received. However, It may not
add to the scientific consideration of PASS
performance to omit samples on the basis of
too few aliquots where all other indicators
indicate a qualifying storm event. (see
discussion in SQIDEP, and Appendix A)
Hydrograph coverage Indicative 50% (importantly the rising and falling Peak and median rainfall runoff volumes were
hydrograph components should be included in testing, recorded. Additional data for hydrograph can
PASS
and dependent on catchment and rainfall patterns, be provided on request (refer to Appendix C)
multiple peaks should be accounted for).
Manual Sampling Only for constituents that transforms rapidly, require Sample collection, handling and transport
special preservation or adhere to bottles, or where were undertaken in accordance with AS/NZS
compositing can mask the presence of some 5667.1:1889 (samples were returned to lab PASS
contaminants through dilution. See Section 10 for within 24 hours of collection, etc)
details.
Sampling Location As identified and agreed in the submitted QAPP. Commercial site at Ipswich (see Appendix A) PASS
Analytical Methods Various (refer to agreed EPA reference document) Sample collection, handling and transport
and/or Standard Methods (for organic, inorganic and were undertaken in accordance with AS/NZS
biological analysis as required) 5667.1:1889. Testing was conducted in PASS
accordance with APHA (2012) with various
methods (refer table 1 of Appendix A)
Page 4
Chemical and Physical analyses As identified and agreed in the submitted QAPP. Sample collection, handling and transport
were undertaken in accordance with AS/NZS
5667.1:1889. Testing was conducted in
PASS
accordance with APHA (2012) with various
methods (refer table 1 of Appendix A) by a
NATA registered laboratory
Rainfall Recording
Flow Measurement Location Inlet, Outlet and Bypass*, as applicable. Based on The upstream sampling point is equipped with
relevant accepted measurement protocols for flow a flow measuring device and an automatic
type (e.g. open channel, in pipe) sample collection system while the
PASS
downstream sampling point is only equipped
with an automatic sample collection system.
(refer to experimental set up in Appendix A)
Precipitation Measurement Automatic rain gauge (pluviometer) RIMCO RIM7499 tipping bucket rain gauge
PASS
(solar powered/automatic)
Recording Intervals 5 minutes or less Data Logger set to log every minute after
PASS
qualifying rainfall event detected
Recording Increments No greater than 0.25mm RIMCO RIM7499 tipping bucket rain gauge
(solar powered/automatic) has resolution of PASS
0.20mm (see Appendix A)
Rain Gauge Calibration Twice during verification period Commissioned to site fully calibrated PASS
Data Analysis and Reporting
Performance Indicators Based on the Performance Claim stated in Detailed Report in Appendix A refers to TSS, TP & TN
Evaluation Report. (Can include but not limited to TSS, removal, but also mentions heavy metals and PASS
Metals, TPH, TP & TKN) pH (among other pollutants).
Page 5
Performance Indicators Calculation 1. Percent Concentration Removal (See Section Performance indicators used the Percent
14.24) (Arithmetic average and median. If Concentration Removal (or Concentration
difference is 10% or greater, inspect data set Reduction Efficiently (CRE)) method. Median
closely)
and Mean values were within 10% for TSS &
2. Percent Mass Removal (See Section 14.24) TN, however TP was 11%. Data was inspected,
(Arithmetic average and median. If difference
and was concluded that experimental
is 10% or greater, inspect data set closely)
variability was the cause. (see Appendix A)
3. Relative Achievable Efficiency (See Section
14.24) (Arithmetic average and median. If
difference is 10% or greater, inspect data set
closely
4. Summation of loads (See Section 14.24) PASS
(Arithmetic Average and median. If
difference is greater than 10% inspect
dataset closely)
5. Efficiency Ratio (See Section 14.24)
(Arithmetic Average and median. If
difference is greater than 10% inspect
dataset closely)
6. Flow Based Variability (FBV) (See Section
14.24), including a plot of one of the above
performance measures against the 25, 50,
75, 100 and 125 percent of the treatable flow
rate.
Performance Variability Schematics Box and Whisker Plot Published Paper (see Appendix B) PASS
Statistical Significance Testing Log-transformed inlet and outlet paired samples at Published Paper (see Appendix B)
PASS
90% confidence level.
Sizing Methodology See Section 13.2.4 Generic Node in MUSIC, based on
performance removal claims in report in PASS
Appendix A
Page 6
Given that the objectives table in the draft SQIDEP document are all achieved, the Jellyfish Filter can be approved for use with the removal capabilities described in the testing report.
Page 7
Appendix A: Evaluation of Treatment Performance of a Stormwater Treatment Membrane Filter under Australian Conditions
Page 8
Australian Journal of Water Research
ISSN: 2203-9482 (Print) ISSN: 2203-9490 (Online)
Evaluation of Treatment Performance of a Stormwater
Treatment Membrane Filter under Australian Conditions
Dr Charles Kelly (Corresponding author)
Commercial Manager – Humes Water Solutions
18 Little Cribb Street, Milton 4064, Australia
Tel: 61-7-3364 2957 E-mail: [email protected]
Anton Bardak
Water Solutions Engineer – Humes Water Solutions
18 Little Cribb Street, Milton 4064, Australia
Tel: 61-7-3364 2956 E-mail: [email protected]
The research was conducted by Queensland University of Technology and is financed by Holcim
Australia
Abstract
Stormwater runoff transports a range of pollutants and is a significant non-point source of urban water
pollution. Effective removal of stormwater pollutants requires treatment technologies that are
appropriate to site constraints, as well as rainfall and pollutant load characteristics. A range of
engineered stormwater treatment systems are marketed by various commercial vendors. Assessing
their performance requires quantification of inputs and outputs to receiving environments under diverse
event durations, flows, and loads, Herr and Sansalone (2015).
In line with these recommendations the treatment performance of a device has been assessed with a
field monitoring program centered on a device installed at Ipswich in South East Queensland. The
findings to date of the field monitoring are as follows:
• For the seven qualifying events monitored, median concentration reduction efficiencies for
TSS, TN and TP are 89%, 55% and 65%, respectively.
• Comparison of monitored rainfall events suggests that these are within the envelope of typical
rainfall events for Brisbane and are equivalent to events monitored in Florida.
• Monitored stormwater quality in terms of TSS, TN and TP are well within the expected
stormwater treatment quality for the Brisbane region.
• Performance characteristics observed in this field monitoring study are comparable with the
performance reported in the previous Florida study.
Keywords: Performance comparison; Performance evaluation; Stormwater quality; Stormwater
treatment
Australian Society for Commerce Industry & Engineering
www.scie.org.au
1
Page 9
Australian Journal of Water Research
ISSN: 2203-9482 (Print) ISSN: 2203-9490 (Online)
1. Introduction
Stormwater runoff transports a range of pollutants and is a significant non-point source of urban water
pollution. Effective removal of stormwater pollutants requires treatment technologies that are
appropriate to site constraints, as well as rainfall and pollutant load characteristics. A range of
engineered stormwater treatment systems are marketed by various commercial vendors. Assessing
their performance requires quantification of inputs and outputs to receiving environments under diverse
event durations, flows, and loads, Herr and Sansalone (2015).
The Jellyfish® Filter is an engineered stormwater quality treatment technology featuring membrane
filtration in a compact stand-alone treatment system that is designed to remove a wide variety of
stormwater pollutants. The Jellyfish® Filter integrates pre-treatment and filtration with a passive self-
cleaning mechanism. The system utilizes membrane filtration cartridges with high filtration surface
area and flow capacity and designed to operate under relatively low driving head compared to
conventional filter systems. Performance of the Jellyfish® Filter as a stormwater treatment device had
been tested and deemed appropriate for the US market (UoF, 2011).
Extensive field testing of the Jellyfish® Filter treatment device had been undertaken in the US by
University of Florida. However, further performance testing under Australian climatic conditions was
considered essential for the introduction of the Jellyfish® Filter to the local market. Accordingly,
Queensland University of Technology (QUT) was requested by Humes Australia to undertake a
comprehensive field based monitoring study to verify the performance of the treatment device under
South East Queensland (SEQ) climatic conditions.
A two phase approach was adopted by QUT to assess the treatment performance of the Jellyfish® Filter.
The first phase consisted of assessing the compatibility of performance characteristics reported in UoF
(2011) to South East Queensland (SEQ) climatic conditions using advanced statistical approaches. The
second phase consisted of evaluating the treatment performance under SEQ climatic condition by
undertaking a field monitoring program centered on a device installed at Ipswich, South East
Queensland.
Phase 1 of the study has been completed and reported in August 2014 (Egodawatta et al. 2014). This
paper follows on from Phase 1 and presents the findings to date of the field monitoring program
currently being undertaken as Phase 2.
2. Monitoring System
2.1 Site Description
The site is a recently developed commercial facility located at 292 Brisbane Street, West Ipswich 4305,
QLD. The site has a total area of 1678m2 with approximately 550m2 of roof area and 1128m2 of
impervious driveways and parking lots. Figure 1 shows the layout of the development, including the
stormwater drainage network.
(a) (b)
Figure 1. Study Site: (a) Aerial View; (b) Drainage Network
Australian Society for Commerce Industry & Engineering
www.scie.org.au
2
Page 10
Australian Journal of Water Research
ISSN: 2203-9482 (Print) ISSN: 2203-9490 (Online)
The stormwater drainage at the site is handled by a piped, subsurface network. Driveway runoff from
the site enters the drainage network via grated field inlet pits while downpipes from the roof are
directed to rainwater tanks, and the overflow pipe to the closest field inlet. The treatment device is
fitted as a part of the drainage network after the last grated manhole within the site as shown in Figure
2.
Figure 2. Monitoring System Layout
2.2 Instrumentation and Sampling Methodology
Automatic sampling stations are installed at two sampling points as shown in Figure 2. The upstream
sampling point is equipped with a flow measuring device and an automatic sample collection system
while the downstream sampling point is equipped with an automatic sample collection system. A
tipping bucket rain gauge is installed as part of the field monitoring system. A schematic of the field
monitoring system is shown in Figure 3.
Figure 3. Schematic of the Two Sampling Points
Australian Society for Commerce Industry & Engineering
www.scie.org.au
3
Page 11
Australian Journal of Water Research
ISSN: 2203-9482 (Print) ISSN: 2203-9490 (Online)
The flow measuring device installed at the upstream sampling point consists of a high accuracy Control
Logic LMP307 pressure transducer with accuracy of 0.1% and a pipe insert Palmer-Bowlus flume (8
inch). The pressure transducer is mounted on the wall of the field inlet pit just upstream of the flume so
that depth measurements can be transformed to flow using the rating table associated with the flume.
The flume device and the pressure transducer operation were calibrated over a range of flow rates prior
to the commencement of the monitoring program.
Two ISCO 6712 automatic samplers with capacity to hold 24 samples are installed to collect runoff
samples from upstream and downstream sampling points as shown in Figure 3. The samplers are
housed in weatherproof security cabinets close to each sampling point. The suction hoses are kept as
short as possible to reduce head loss. Sampling stations are equipped with data loggers, battery and
solar chargers to ensure independent operation. The RIMCO RIM7499 tipping bucket rain gauge
installed to measure rainfall has a resolution of 0.2mm. The rain gauge is mounted on a three meter
pole as shown in Figure 3 to reduce the impact of the rain shadow created by any neighboring
buildings. A Bureau of Meteorology operated rain gauge is situated approximately 600m away from the
study site.
2.3 Sample Handling and Testing
A carefully formulated methodology was adopted for the selection of appropriate rainfall events for
evaluation, runoff sample collection and sample handling and testing. Only runoff samples originating
from rainfall events with a minimum of three antecedent dry days and more than 2.6mm rainfall depth
were tested. These rainfall events were classified as qualifying events. This was to ensure that there are
representative contributions from both driveways/parking lots and roof runoff as well as to ensure that
there was appreciable accumulation of pollutants on the impervious surfaces. The automatic samplers
were programmed to collect discrete samples.
The samplers were programmed to trigger based on the occurrence of rainfall. Sampling intervals were
programmed to vary based on the intensity of the rainfall received. Runoff samples collected from a
qualifying rainfall event are combined to form a composite sample (Event Mean Concentration –
EMC). Representative aliquots of samples were extracted using a churn sample splitter (Bel-Art
Products) and submitted to a NATA registered laboratory (Advanced Analytical Laboratory) for
analysis. The parameters tested and the test methods adopted are listed in Table 1. Sample collection,
handling and transport were undertaken in accordance with AS/NZS 5667.1:1998.
Table 1. List of Parameters and Test Methods
Parameter Method
pH APHA (2012) Method 4500
Electrical conductivity APHA (2012) Method 2510B
Total metals – Fe , Al, Mn , Cu, Cr, US EPA 200.8/3050/6010B by ICPMS US
Pb, Ni, Zn, Cd EPA (1994), US EPA (1996)
APHA (2012) Method 4500 by discrete
Total nitrogen, NO3, TKN
analyser with persulphate digestion
Total phosphorus, FRP APHA (2012) Method 4500 by discrete
(orthophosphate P) analyser with persulphate digestion
Total suspended solids (103oC) APHA (2012) Method 2540 D
APHA (2012) Method 5310 by TOC
Total organic carbon
analyser
Total petroleum hydrocarbon (C10-
NEPM 2011 (draft)
C40), silica gel
Australian Society for Commerce Industry & Engineering
www.scie.org.au
4
Page 12
Australian Journal of Water Research
ISSN: 2203-9482 (Print) ISSN: 2203-9490 (Online)
3. Results and Discussion
The outcomes from the evaluation of seven qualifying rainfall events that occurred over a 5 month
period from 28th June 2014 to 7th November 2014 are presented in this paper. Rainfall and runoff data
for the monitored events are presented in Table 2.
Table 2. Characteristics of Monitored Events
Date Sampled Rainfall Depth Rainfall Duration Peak Intensity Number of
(mm) (minutes) (mm/hr) Samples
collected
28/06/2014 5.2 19 57 3
16/08/2014 33 1222 27.4 24
27/08/2014 7.2 128 13.7 14
25/09/2014 8.6 214 18.9 4
13/10/2014 7 597 15.4 13
27/10/2014 3 36 10.3 7
3.1 Parameter Concentration Values
The laboratory test results for the monitored events are given in Table 3 and Table 4.
Table 3. Water Quality Parameters: Solids, Nutrients and Organic Carbon
Date
Sampling pH EC TSS Nitrate TKN TN Phosphate TP TOC
sampled
Point Units pH µS/cm mg/L mg/L mg/L mg/L mg/L mg/L mg/L
PQL 1.0 1 2 0.01 0.1 0.1 0.01 0.02 1.0
Inflow 6.4 40 180 0.19 1.1 1.3 <0.01 0.24 5
28/06/2014
Outflow 6.8 120 13 0.11 1.2 1.3 0.03 0.24 5
Inflow 7.3 70 170 0.33 2.4 2.8 0.18 0.41 9
16/08/2014
Outflow 7.1 50 2 0.28 0.2 0.5 0.02 0.03 4
Inflow 7 120 19 0.42 0.6 1.1 0.06 0.093 4
27/08/2014
Outflow 6.9 70 2 0.17 0.3 0.5 0.02 0.05 2
Inflow 6.9 150 26 1.9 1.7 3.1 0.02 0.17 9
25/09/2014
Outflow 6.6 55 4 0.28 0.8 1.1 <0.01 0.06 6
Inflow 7 125 25 0.79 1 1.9 0.05 0.17 7
13/10/2014
Outflow 7 140 4 0.31 1.5 1.9 0.03 0.12 6
Inflow 6.7 145 51 0.73 5.2 6 0.27 0.53 32
27/10/2014
Outflow 6.9 130 10 0.82 1.5 2.3 0.07 0.17 18
Inflow 6.8 105 74 0.58 2.6 3.2 0.36 0.31 11
06/11/2014
Outflow 6.8 80 8 0.61 1 1.6 0.06 0.08 8
Note: PQL – Practical Quantification Limit
Australian Society for Commerce Industry & Engineering
www.scie.org.au
5
Page 13
Australian Journal of Water Research
ISSN: 2203-9482 (Print) ISSN: 2203-9490 (Online)
Table 4. Water Quality Parameters: Total Petroleum Hydrocarbons and Metals
Date TPH TPH TPH TPH
Sampling Al Cd Cr Cu Fe Mg Ni Pb Zn
sampled C6-C9 C10-14 C15-28 C29-36
Point Units µg/L µg/L µg/L µg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L
PQL 25 50 100 100 0.05 0.001 0.002 0.002 0.05 0.005 0.003 0.006 0.005
Inflow <25 <50 <100 <100 3 <0.001 0.012 0.039 3.8 0.12 0.005 0.01 0.3
Outflow 28/06/2014 <25 150 200 <100 0.2 <0.001 0.002 0.015 1.2 0.13 <0.003 <0.006 0.041
Inflow <25 <50 <100 <100 4 <0.001 0.018 0.069 4.6 0.16 0.012 0.012 0.75
Outflow 16/08/2014 <25 <50 <100 <100 0.05 <0.001 0.006 0.012 <0.05 0.012 <0.003 <0.006 0.017
Inflow <25 <50 <100 <100 0.74 <0.001 0.004 0.01 0.68 0.023 <0.003 <0.006 0.11
Outflow 27/08/2014 <25 <50 <100 <100 0.08 <0.001 <0.002 0.005 0.17 0.023 <0.003 <0.006 0.024
Inflow <25 <50 140 <100 0.94 <0.001 0.008 0.026 0.96 0.034 <0.003 <0.006 0.18
Outflow 25/09/2014 <25 <50 120 <100 0.08 <0.001 0.007 0.01 0.11 0.013 <0.003 <0.006 0.027
Inflow <25 <50 160 110 0.75 <0.001 0.004 0.016 0.86 0.037 <0.003 <0.006 0.15
Outflow 13/10/2014 <25 <50 180 110 0.093 <0.001 0.002 0.005 0.95 0.099 <0.003 <0.006 0.032
Inflow <25 100 570 240 1.8 <0.001 0.009 0.038 2.1 0.13 0.005 0.01 0.43
Outflow 27/10/2014 <25 98 450 200 0.19 <0.001 0.002 0.012 1.1 0.087 <0.003 <0.006 0.06
Inflow <25 <50 160 <100 1.9 <0.001 0.007 0.025 2.3 0.11 0.004 0.007 0.29
Outflow 06/11/2014 <25 <50 160 <100 0.19 <0.001 <0.002 0.007 0.34 0.032 <0.003 <0.006 0.035
Note: PQL – Practical Quantification Limit
Australian Society for Commerce Industry & Engineering
www.scie.org.au
6
[OCR of page images]
Australian Journal of Water Research &@
ISSN: 2203-9482 (Print) ISSN: 2203-9490 (Online) a
Table 4. Water Quality Parameters: Total Petroleum Hydrocarbons and Metals
Date TPH | TPH | TPH | TPH ,
Sampling | sampled c6-c9 | c10-14 | c15-28 | c29-36 | “I Cd Cr Cu Fe Mg Ni Pb Zn
Point Units pg/L pg/L pg/L pg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L
PQL 5 50 100 100 0.05 | 0.001 | 0.002 | 0.002 | 0.05 | 0.005 | 0.003 | 0.006 | 0.005
Inflow <25 <50 <100 | <100 3 <0.001 | 0.012 | 0.039 3.8 0.12 | 0.005 | 0.01 03
Outflow | 28/06/2014 |~~Z55 150 200 <100 02 | <0.001 | 0.002 | 0.015 12 0.13 | <0.003 | <0.006 | 0.041
Inflow <25 <50 <100 | <100 4 <0.001 | 0.018 | 0.069 4.6 0.16 | 0.012 | 0.012 | 0.75
Outflow | 16/08/2014 [~~ 255 <50 <100 | <100 | 0.05 | <0.001 | 0.006 | 0.012 | <0.05 | 0.012 | <0.003 | <0.006 | 0.017
Inflow <25 <50 <i00 | <100 | 0.74 | <0.001 | 0.004 | 0.01 0.68 | 0.023 | <0.003 | <0.006 | 0.11
Outflow | 27/08/2014 |~Z55 <50 <i00 | <100 | 0.08 | <0.001 | <0.002 | 0.005 | 0.17 | 0.023 | <0.003 | <0.006 | 0.024
Inflow <25 <50 140 <i00 | 0.94 | <0.001 | 0.008 | 0.026 | 0.96 | 0.034 | <0.003 | <0.006 | 0.18
Outflow | 29/09/2014 |~Z55 <50 120 <100 0.08 | <0.001 | 0.007 | 0.01 0.11 0.013 | <0.003 | <0.006 | 0.027
Inflow <25 <50 160 110 0.75 | <0.001 | 0.004 | 0.016 | 0.86 | 0.037 | <0.003 | <0.006 | 0.15
Outflow | 13/10/2014 |~Z55 <50 180 110 0.093 | <0.001 | 0.002 | 0.005 | 0.95 | 0.099 | <0.003 | <0.006 | 0.032
Inflow <25 100 570 240 18 | <0.001 | 0.009 | 0.038 21 0.13 | 0.005 | 0.01 0.43
Outflow | 27/10/2014 | 255 98 450 200 0.19 | <0.001 | 0.002 | 0.012 11 0.087 | <0.003 | <0.006 | 0.06
Inflow <25 <50 160 <100 19 | <0.001 | 0.007 | 0.025 23 0.11 | 0.004 | 0.007 | 0.29
Outflow | 99/11/2014 [~~ 255 <50 160 <100 | 0.19 | <0.001 | <0.002 | 0.007 | 0.34 | 0.032 | <0.003 | <0.006 | 0.035
Note: PQL — Practical Quantification Limit
Australian Society for Commerce Industry & Engineering
www.scie.org.au
Page 14
Australian Journal of Water Research
ISSN: 2203-9482 (Print) ISSN: 2203-9490 (Online)
3.2 Treatment Performance
Measured concentrations representing inflows to the treatment device and outflows from the device
was used for determining the treatment performance. Treatment performance was evaluated for TSS,
TN and TP as these are the most common stormwater quality parameters. The estimated treatment
performance is shown in Table 5.
Table 5. Concentration Reduction Efficiency and Efficiency Ratio
Date TSS (mg/L) TN (mg/L) TP (mg/L)
sampled
Inflow Outflow CRE Inflow Outflow CRE Inflow Outflow CRE
28/06/2014 180 13 92.8 1.3 1.3 0.0 0.24 0.24 0.0
16/08/2014 170 2 98.8 2.8 0.5 82.1 0.41 0.03 92.7
27/08/2014 19 2 89.5 1.1 0.5 54.5 0.093 0.05 46.2
25/09/2014 26 4 84.6 3.1 1.1 64.5 0.17 0.06 64.7
13/10/2014 25 4 84.0 1.9 1.9 0.0 0.17 0.12 29.4
27/10/2014 51 10 80.4 6 2.3 61.7 0.53 0.17 67.9
06/11/2014 74 8 89.2 3.2 1.6 50.0 0.31 0.08 74.2
Average 78 6 2.8 1.3 0.27 0.11
Average CRE 88.5 44.7 53.6
Median CRE 89.2 54.5 64.7
Note:
1. CRE – Concentration reduction efficiency. CRE is the percentage reduction in concentration
with respect to inflow concentration for individual events.
2. Concentrations less than Practical Quantification Limit (PQL) was replaced by 50% of
detection limit for performance calculations.
It is commonly known that the average of a small sample set can be far from the true mean of a
population. In this respect, median values could be a true representative of performance indicators. This
is particularly true when a data set is log-normally distributed rather than normally distributed. As
evident in Table 5, median of the percentage concentration reduction efficiencies for TSS, TN and TP
are 89.2%, 54.5% and 64.7%, respectively. Average values (mean) for the same parameters are less
than 10% difference from the median except for TP. This indicates consistency in treatment
performance of the Jellyfish® Filter for all the events investigated.
4. Comparison of Treatment Performance with Previous Studies
Demonstrating the validity of the data obtained and estimated performance is critical in field
monitoring. Accordingly, the data obtained and estimated performance was compared with treatment
performance reported in previous studies.
4.1 Comparison of Rainfall Data
It was considered important to demonstrate that the range of rainfall events monitored is within the
range of rainfall typical to local conditions. At the same time, it was also important to compare the
events monitored with the events monitored by the UoF (2011) study. Accordingly, the events
monitored in this study are plotted with typical Brisbane events from 2004 (selected as a representative
year) and events monitored in UoF (2011) study as shown in Figure 4. Phase 1 of the study undertaken
confirmed that 2004 represents a rainfall year with average annual rainfall depth and hence can be
considered as a representative year (Egodawatta et al. 2014). As evident in Figure 4, events monitored
in this study are well within the envelope created by the other two sets of rainfall data, thus confirming
the representativeness of the monitored events in this study.
Australian Society for Commerce Industry & Engineering
www.scie.org.au
7
Page 15
Australian Journal of Water Research
ISSN: 2203-9482 (Print) ISSN: 2203-9490 (Online)
Figure 4. Comparison of Rainfall Events
4.2 Comparison of Inflow Water Quality Data
Additionally, the stormwater inflow concentrations were evaluated with respect to typical values from
urban areas in SEQ and Florida as shown in Figure 5 and Figure 6. Typical inflow stormwater quality
was considered equivalent to MUSIC concentration parameter values for the Brisbane region. Two
monitoring programs undertaken by QUT in the SEQ region are also included in the comparison.
Inflow concentrations for Florida were obtained from UoF (2011).
MUSIC parameters are typically presented as mean and standard deviation of a long-term data set.
Since the requirement of this comparison was to justify the fact that measured data are within those
statistical limits, mean and 1.96 times the standard deviation is presented as positive and negative error
bars in Figures 5 and 6. It is typically considered that 1.96 times the standard deviation as positive and
negative error bars represent 95% of the data ranges of the assessed population. Means and 95% error
bars are presented for urban residential, industrial, commercial and rural residential areas for
comparison. Similar statistical measures of Florida and Ipswich water quality data are presented for
direct comparison. Original Florida and Ipswich water quality data are also presented as scattered data
sets (x axis is not relevant) in Figure 5 and Figure 6.
As evident from Figure 5 and Figure 6, stormwater quality at the monitored site is well within the
envelope of expected Brisbane water quality characteristics. Particularly, TP concentrations are within
the typical concentrations for Brisbane, though they show significant differences with the Florida data.
It is hypothesised that specific soil characteristics and landscape management practices would have
influenced phosphorus levels reported in UoF (2011) study.
Australian Society for Commerce Industry & Engineering
www.scie.org.au
8
Page 16
Australian Journal of Water Research
ISSN: 2203-9482 (Print) ISSN: 2203-9490 (Online)
3.5
3
2.5
Log 10 TSS (mg/L)
2
1.5
1
Brisbane MUSIC Parameters
Florida Data
0.5 Florida Statistics
Ipswich Data
Ipswich Statistics
0 0 1 2 3 4 5 6
Industrial
Urban
Residential
Commercial
Ipswich
Statistics
Florida
Statistics
Rural
Residential
TSS
Figure 5. Comparison of Ipswich Inflow TSS Concentrations with Typical Brisbane and Florida Data
1.5 2.5
Brisbane MUSIC Parameters
1 Florida Data
2 Florida Statistics
0.5 Ipswich Data
Ipswich Statistics
1.5
0 0 1 2 3 4 5 6
Log 10 TN (mg/L)
Log 10 TP (mg/L)
-0.5 1
-1
0.5
-1.5 Brisbane MUSIC Parameters
Florida Data
Florida Statistics 0 0 1 2 3 4 5 6
-2
Ipswich Data
Ipswich Statictics
-2.5 -0.5
Urban
Urban
Residential
Residential
Commercial
Industrial
Industrial
Rural
Residential
Ipswich
Statistics
Commercial
Rural
Residential
Ipswich
Statistics
Florida
Florida
Statistics
Statistics
TP TN
Figure 6. Comparison of the Ipswich Inflow TP and TN Concentrations with Brisbane and Florida Data
Australian Society for Commerce Industry & Engineering
www.scie.org.au
9
Page 17
Australian Journal of Water Research
ISSN: 2203-9482 (Print) ISSN: 2203-9490 (Online)
4.3 Comparison of Treatment Performance
Event based performance of a treatment device can be influenced by a range of factors including inflow
concentration, flow rate and the characteristics of the events that the device has treated. This paper
compares the field performance of the Jellyfish® Filter under Southeast Queensland climatic
conditions with the performance shown in the Florida study. As the study is still ongoing, it is expected
that further detailed analysis of treatment performance will be undertaken in the future.
Table 6. Comparison of Pollutant Removal Performance
Median Removal Efficiency (%)
Monitoring Program TSS TN TP
Florida 89 51 59
Ipswich (SEQ) 89 55 65
Table 6 presents the median pollutant removal efficiencies in relation to TSS, TN and TP. It is evident
that all three removal efficiencies are of a similar order for both monitoring programs. TSS and TN
removal efficiencies are closely comparable and can be primarily attributed to having similar inflow
quality as shown in Figures 5 and 6. TP removal efficiency does not show a significant difference even
though inflow concentrations of phosphorous in Florida are significantly higher compared to the
Ipswich values.
5. Summary
The treatment performance under SEQ climatic conditions are being assessed by undertaking a field
monitoring program centered on a device installed at Ipswich in South East Queensland. The findings
to date of the field monitoring are as follows:
• For the seven qualifying events monitored, median concentration reduction efficiencies for
TSS, TN and TP are 89%, 55% and 65%, respectively.
• Comparison of monitored rainfall events suggests that these are within the envelope of typical
rainfall events for Brisbane and are equivalent to events monitored in Florida where a
comprehensive field monitoring program was undertaken.
• Monitored stormwater quality in terms of TSS, TN and TP are well within the expected
stormwater treatment quality for the Brisbane region.
• Performance characteristics observed in this monitoring study are comparable with the
performance reported in the previous Florida study.
MUSIC nodes have been created for the Jellyfish Filter treatment device based on all of the field test
data, which can be found at the following location.
http://www.humes.com.au/precast-solutions/stormwater/stormwater-treatment/tertiary.html
Australian Society for Commerce Industry & Engineering
www.scie.org.au
10
Page 18
Australian Journal of Water Research
ISSN: 2203-9482 (Print) ISSN: 2203-9490 (Online)
6. References
APHA, 2012, Standard methods for the examination of water & wastewater, Washington DC, American
Public Health Association.
AR&R, 1998, Australian rainfall and runoff - A Guide to Flood Estimation, Vol 1, Engineers, Australia,
Barton, ACT.
AS/NZS 1998, Australian New Zealand Standards, Water quality Sampling, Part 1: Guideline of the
Design of Sampling Programmes, Sampling Techniques and the Preservations and Handling of
Samples.
Egodawatta, P. and McGree, J., Wijesiri, B., Goonetilleke, A., 2014, Compatibility of Stormwater
Treatment Performance Data between Different Geographical Areas, Water, Vol 41, No 5
Herr, C. and Sansalone, J. 2015, In Situ Volumetric Filtration Physical Model to Separate Particulate
Matter from Stormwater. J. Environ. Eng., 10.1061/(ASCE)EE.1943-7870.0000946, 04015017.
Published on line 24/02/2015
UoF, 2011, TARP field test performance monitoring of a Jellyfish® Filter JF4-2-1, Engineering School
of Sustainable Infrastructure and Environment (ESSIE), University of Florida, Gainesville, Florida.
US EPA, 1994, Method 200.8: Trace elements in waters and wastes by inductively coupled plasma-
mass spectrometry, U.S. Environmental Protection Agency, Ohio.
US EPA, 1999, Method 3050B: Acid digestion of sediments, sludges, and soils, U.S. Environmental
Protection Agency, Ohio.
Australian Society for Commerce Industry & Engineering
www.scie.org.au
11
Page 19
Appendix B: A Statistical Approach to Assess Stormwater Treatment Device Performance Data [OCR of page images] Appendix B: A Statistical Approach to Assess Stormwater Treatment Device Performance Data
Page 20
1
Technical Papers
A STATISTICAL APPROACH TO
ASSESS STORMWATER TREATMENT
DEVICE PERFORMANCE DATA
A demonstration of how statistical methods can be
used to gauge whether or not presented data sets
are sufficient to draw reliable conclusions
C Kelly, A Bardak
ABSTRACT There is a range of engineered analysis to be conducted, such as the
stormwater treatment systems to paired t-test, and calculations involving
It has long been established that the pre-
consider. Assessing the performance confidence intervals, and the number of
treatment of stormwater runoff prior to
of hard engineered systems requires paired samples required to achieve a given
reaching its final destination is essential
quantification of inputs and outputs to confidence level. Table 1 displays a data set
for river and catchment health. There is a
receiving environments under a diverse of 10 qualifying events (Kelly and Bardak,
plethora of stormwater runoff treatment
range of event durations, flow rates and 2015), which is used to demonstrate this
devices either available or about to
loads (Herr and Sansalone, 2015). statistical approach for three pollutants:
enter the market. Stormwater Australia
Total Suspended Solids (TSS), Total
has released a consultation draft of its Stormwater Australia has released
Phosphorus (TP) and Total Nitrogen (TN).
Stormwater Quality Improvement Device a consultation draft of its Stormwater
Evaluation Protocol (SQIDEP), which Quality Improvement Device Evaluation The Event Mean Concentration (EMC) is
includes a requirement to undertake Protocol, or SQIDEP (2014), which a flow-weighted concentration parameter
statistical assessment as part of the includes a requirement for statistical derived for each storm. It can be calculated
validation of performance data. assessment to validate the performance as follows (Geosyntec Consultants &
data. There remains an important step Wright Water Engineers, 2009):
This paper uses data from a field- in the further development of SQIDEP,
testing case study of an engineered which is to engage with academia and
treatment technology to demonstrate regulators to reach a consensus on the
how statistical methods can be used to robustness of a statistical approach.
where:
determine whether or not the data sets
being presented are actually sufficient Field-testing data are used in this
paper to illustrate a statistical assessment
to be able to draw reliable conclusions.
process to determine whether or not the
The assessment is conducted for Total
data sets being presented are actually The Concentration Removal Efficiency
Suspended Solids, Total Phosphorus and
sufficient for statistically valid conclusions (CRE) method was used to calculate the
Total Nitrogen data, and shows that the
to be drawn. The paper presents a per cent removal in Kelly and Bardak
required number of paired samples is
worked example of the proposed analysis (2015). It is also used in this paper to
different for each of these parameters
technique using a data set reported in demonstrate the statistical methodology.
and depends on the variability of the
Kelly and Bardak (2015) for a proprietory CRE% was calculated for each event as
data sets. It is necessary to engage
stormwater treatment device (the follows (Geosyntec Consultants & Wright
with academia and regulators to reach
Jellyfish Filter). Details on experimental Water Engineers, 2009):
a consensus on the robustness of a set-up, sampling methods used and
statistical approach, and this paper definitions of qualifying events are
provides an example. reported in Kelly and Bardak (2015).
INTRODUCTION METHOD DETAIL It should be noted that there are
Stormwater runoff transports a range AND RESULTS several recommended methods that can
of pollutants and is a significant non- Initially, potential outliers are identified, be used to calculate the contaminant
point source of urban water pollution. examined and removed (if necessary). removal efficacy, such as CRE, Efficiency
It has long been established that the The data is then tested for normality to Ratio, Mass Reduction Efficiency, etc.
pre-treatment of stormwater runoff identify which statistical methods can Selection of a relevant percentage
prior to reaching its final destination is be used on that set. For cases where removal calculation method is discussed
essential for river and catchment health the raw data is not normally distributed, in SQIDEP (Stormwater Australia, 2014)
(Herr and Sansalone, 2015). Effective but appears to be positively skewed, and Geosyntec Consultants and Wright
treatment requires technologies that are a lognormal transformation may be Water Engineers (2009). The following
appropriate to site constraints, as well as applicable (Singh et al., 1997). Normally statistical assessment approach can be
rainfall and pollutant load characteristics. distributed data allows standard statistical applied to all of these methods.
MAY 2015 WATER
Page 21
2
Technical Papers
Table 1. Influent and effluent data for TSS, TP and TN.
Event TSS EMCi TSS EMCe TSS CRE TP EMCi TP EMCe TN EMCi TN EMCe
TP CRE (%) TN CRE (%)
Number (mg/L) (mg/L) (%) (mg/L) (mg/L) (mg/L) (mg/L)
1 180.0 13.0 93 0.2 0.2 0 1.3 1.3 0
2 170.0 2.0 99 0.4 0.0 93 2.8 0.5 82
3 19.0 2.0 89 0.1 0.1 44 1.1 0.5 55
4 26.0 4.0 85 0.2 0.1 65 3.1 1.1 65
5 25.0 4.0 84 0.2 0.1 29 1.9 1.9 0
6 51.0 10.0 80 0.5 0.2 68 6.0 2.3 62
7 74.0 8.0 89 0.3 0.1 74 3.2 1.6 50
8 21.0 5.0 76 0.5 0.1 80 5.7 2.5 56
9 13.0 1.0 92 0.2 0.1 63 2.0 1.3 35
10 18.0 3.0 83 0.14 0.09 36 1.3 1.0 23
Notes: EMCi: Event mean concentration of influent sample; EMCe: Event mean concentration of effluent sample; CRE: Concentration removal efficiency.
OUTLIER IDENTIFICATION Outlier
An outlier is defined as a data point that Max Observation Below
is distinctly separate from and inconsistent Upper Inner Fence
with the majority of the data points
(Walfish, 2006). Outliers can have negative
Q3
effects on data analyses as they skew and
distort the data set or, conversely, can
provide useful information about abnormal
characteristics of the data (Seo, 2006). Median
Therefore, it is a critical step to identify
Q1
and detect outliers.
A basic approach for identifying outliers Min Observation Above
is outlined in Tukey (1977), and involves Lower Inner Fence
simple numerical and graphical methods Figure 1. Definition of features in a Box
that are used when constructing box plots and Whisker Plot.
of the data (Walfish, 2006). It identifies a As seen in the Box and Whisker Plots in
potential outlier as any data point more Figure 2, a number of influent and effluent
than 1.5 interquartile ranges (IQRs) below
data points fell outside the inner fences,
the first quartile, or above the third
but were within the outer fences (see
quartile; that range is referred to as the
Tukey (1977) for calculation method of the
inner fences. An extreme outlier is a value
inner and outer fences). However, the data
that is outside the outer fences, and can be
between the inner and outer fences all
defined as three interquartile ranges below
fell within referenced parameters, which
the first quartile or above the third quartile
suggest typical concentration occurrences
(Tukey, 1977) and should be thoroughly
in Australia (Duncan, 1999); therefore, there
investigated. The box and whisker plots
is no justifiable reason to exclude any data
graphically display outliers, and examples
points from this data set. The CRE (%) data
can be seen in Figure 2 for the three
was also analysed for outliers, but all CRE
parameters of interest. Figure 1 describes
percentages fell within the inner fences.
the features of the box and whisker plot,
Therefore, all data points have been used.
including all relevant statistical measures.
NORMAL DISTRIBUTION
There is no rigid mathematical
VALIDATION
definition of what constitutes an outlier in
the literature; it is a subjective decision. An Following outlier evaluation, the next
analysis should be conducted to determine step in this approach is to test for data
why the outlier may have occurred, and if a normality using graphical methods
justifiable reason can be identified then the and parametric statistical analysis.
data point may be legitimately excluded The Stormwater Australia protocol calls
(Walfish, 2006). Most importantly, data for log-transformed inlet and outlet-
that have been excluded as outliers must paired samples (Stormwater Australia,
still be reported along with the reasons 2014), however if the baseline data
Figure 2. Box and Whisker Plots for TSS,
for their exclusion, as they still form part is already normally distributed then TP & TN, with outliers labelled with their
of the evaluation process. there is no requirement to conduct a event number.
WATER MAY 2015
Page 22
3
Technical Papers
Figure 3. Q-Q plots for the Concentration removal efficiency (CRE) % distribution for TSS, TP and TN against Z-Score
(distance from the mean).
Figure 4. Q-Q Plots of log-transformed inlet and outlet concentration paired samples for TSS, TP and TN. The linear plots
demonstrate the data is now normally distributed.
log-transformation on the data. If the “bell” shape if it is normally distributed. Table 2 presents the results of the
baseline data is not normally distributed, It should be noted that skewed data parametric tests on the CRE% removal
a log-transformation may be required, and data with large kurtosis (high and data from Table 1. The test statistic, W,
which will result in some minor variations wide peak), even though it may have is calculated using the following formula:
to the statistical processes, as described the appearance of a “bell” shape, is
in Olsson (2005) regarding confidence not considered normally distributed.
levels, and Singh et al. (1997) regarding The Q-Q plot is the most common and
transformation back to original data. effective method to graphically identify
where:
if data is normally distributed (Razali and
LOG TRANSFORMATION OF DATA
Wah, 2011). Using this method, normally
Lognormal distributions tend to be distributed data will plot as a straight
common in environmental data sets, as The per cent removal data is
line, with minimal variance.
results from contaminant concentration normally distributed as the p-value
data sets tend to follow a skewed Figures 3 and 4 display how the data from the Shapiro-Wilk Test in Table 2
probability distribution. In the case that in Table 1 are distributed. The per cent is greater than 0.05. This numeric
the raw data is not normally distributed, removals of TSS, TP and TN can be seen result supports the graphical analysis
but appears to be positively skewed, to be normally distributed according to of Figure 2. The influent and effluent
then a lognormal transformation may the linear Q-Q plot (Figure 3). The inlet results were also tested for normality
be applicable (Singh et al., 1997). and outlet Q-Q plots in Figure 4 were after being log transformed, and the
The log-transformed data is then run constructed using a log transformation results shown in Table 3.
through parametric tests as per normally of the inlet and outlet data. These results
The p values in Table 3 show that
distributed data, albeit with slight indicate that the percentage removal is
the influent and effluent data for all three
variations (Olsson, 2005; Singh et al., normally distributed, while the inlet and water quality parameters are lognormally
1997). However, it should be noted that, outlet EMCs are log normally distributed. distributed. This is a critical step in
if this log-transformed dataset still fails Visual inspection alone, however, usually preparation for the subsequent paired
the normality test (graphical method and does not provide conclusive evidence t-test, as a valid test requires that data
Shapiro-Wilk Test (1965)), then a lognormal that the data is normally distributed, and be normally distributed (Shapiro and
distribution is not suitable and another a numerical parametric analysis should be Wilk, 1965).
transformation should be adopted conducted (Razali and Wah, 2011).
(Limpert et al., 2001). It should be noted that, while it
PARAMETRIC ANALYSIS is crucial to prove that the influent
GRAPHICAL METHODS TO OF NORMALITY and effluent data are lognormally
IDENTIFY NORMALITY A parametric statistical analysis involves distributed in preparation for the
Graphical methods involve plotting data using numerical statistical methods to paired t-test, it is also important to
frequency on a histogram, a box-plot, a test data against the null hypothesis prove that the percentage removal
stem-and-leaf plot, or a quantile-quantile that the data is normally distributed. rates are also normally distributed.
plot (Q-Q plot), then visually inspecting There are many such statistical methods This is critical, as the confidence interval
it in order to conclude if the data is available, however the Shapiro-Wilk is constructed around the percentage
normally distributed. For histogram Test (1965) is considered to be the removal, which requires the data to
plots, the data will display a uniform most powerful (Razali and Wah, 2011). be normally distributed.
MAY 2015 WATER
Page 23
4
Technical Papers
Table 2. Results of the Shapiro-Wilk Table 3. Results of the Shapiro-Wilk Parametric Test for log-transformed influent
Parametric Test for % removal to test and effluent data.
for normality of the data distribution.
TSS TP TN
%
% TP % TN Influent Effluent Influent Effluent Influent Effluent
TSS
0.039 0.511 0.449 7.012 5.574 2.893 3.215 3.208 2.589
0.040 0.551 0.516 8.054 5.724 3.032 3.234 2.979 2.835
W (test statistic) 0.982 0.928 0.872 W (test statistic) 0.871 0.974 0.954 0.994 0.929 0.913
p-value (>0.05) 0.974 0.441 0.116 p-value (>0.05) 0.111 0.708 0.693 0.996 0.448 0.357
NON-PARAMETRIC TESTING 3. Calculate the mean of the differences CONFIDENCE INTERVAL
If there are no transformations that can ( ) After the data has been proven to be
be utilised to transform the raw data 4. Calculate the standard deviation of the statistically significant, a 95% confidence
into a normal distribution, then non- differences ( ) interval can be constructed. The 95%
parametric tests need to be used to confidence interval is defined in Neyman
test for differences in populations. An 5. Calculate the Standard Error (SE) of the
(1937) as follows:
example is the Wilcoxon Signed-Rank Test mean difference:
instead of the paired t-test (McDonald, “There is a 95% probability that
2014). Confidence intervals can also be the calculated confidence interval
calculated non-parametrically, as described encompasses the true value of the
in Geyer (2003), or utilising methods such population mean.”
6. Calculate the t-statistic, using a
as bootstrapping (Wang, 2001). However, t-distribution with n – 1 degrees of This involves calculating the standard
these methods can sometimes be complex freedom: error, then adding/subtracting the
in nature and require specialised statistical standard error to/from the mean to create
software to conduct statistical modelling. the upper and lower bounds. There is a
A less mathematically intensive method 7. t-distribution tables can be referenced
95% probability that the true population
would be to increase the sample size to to find the corresponding p-value mean exists within these bounds (Tan
larger than 30, as the Central Limit Theorem (probability) with the t-statistic (T) and Tan, 2010; Neyman, 1937).
states that, as sample size increases, the and n – 1 degrees of freedom.
distribution of sample means approaches
a normal distribution, regardless of initial For a 95% confidence level, if the p-value
distribution (Bonacci, 2012). is less than 0.05, the null hypothesis (there where:
is no difference between the sample set
STATISTICAL SIGNIFICANCE – means) is rejected, and the data sets are
PAIRED T-TEST considered statistically different.
The paired t-test is used to assess the
The paired t-test was conducted on the
statistical difference (or significance)
log transform of the influent and effluent
between two normally distributed data
test data, and results are presented in For the 95% confidence level, the Z
sets. In this case, it is used to compare the
Table 4. The paired t-test results for TSS, score equates to 1.96 for a two-tailed test.
influent and effluent data sets to establish
TP and TN are significantly beyond the 95%
if their population means are statistically In this case, as the CRE percentage
confidence level (p<0.05), therefore we can
different. If it can be confirmed that the defines the removal efficiency of the
reject the null hypothesis and conclude that
two sets of data do not have the same stormwater treatment device, confidence
the influent and effluent concentration data
population mean, then it can be concluded intervals were constructed around the
sets are statistically different.
that there is a statistical significance
between them (Cramér, 1946). Table 4. Paired t-test results for the field test data reported in Table 1. The
log-transformed influent and effluent concentrations are being compared for
The following steps describe the statistical difference.
methodology used in conducting a paired
H0 = there is no difference between the
t-test (Shier, 2004):
concentration means of the influent & TSS TP TN
1. Construct a null hypothesis, i.e. effluent populations
that there is no statistical difference
Mean of Differences (µd) Loge(mg/L) 2.267 1.011 0.672
between the population means.
Std. dev. of Differences(σd) 0.858 0.746 0.525
2. Calculate the difference for each pair.
This is done by subtracting the effluent Std Error (SE) 0.271 0.746 0.166
concentration from the influent
T-Statistic 8.357 4.288 4.042
concentration for each event (i.e.
for each Degrees of Freedom (n-1) 9 9 9
event pair of data, where i varies from
Paired t-test Result (Must be < 0.05) 1.6x10 -5
0.002 0.0029
1 to n).
WATER MAY 2015
Page 24
5
Technical Papers
mean CRE percentage for each pollutant,
Table 5. Confidence Levels for TSS, TP Table 6. Minimum number of paired
rather than the influent and effluent and TN about CRE % Removal. samples required to establish CRE%
concentration data. This is due to the at the 95% Confidence Level.
TSS TP TN
fact that justification is required by the TSS TP TN
approving authority for the percentage Sample Set Mean (µ) 91% 63% 51%
removal of the device, rather than the Z-Score (Zα/2) 1.96 1.96 1.96
Z-Score (Zα/2) 1.96 1.96 1.96
justification of influent and effluent COV 0.07 0.44 0.54
concentrations at the testing site, which Std. dev. (σ) 0.07 0.28 0.28
% Removal 91% 63% 51%
can be justified by aligning with referenced Number of Samples (n) 10 10 10
parameters for typical concentration Number of Paired
occurrences in Australia as described Std Error of the Mean Samples Required
4% 17% 17% 0.03 3.46 9.87
in Duncan (1999). Table 5 presents (SE) to Meet 95%
the confidence intervals for the data. Lower Confidence Confidence Interval
87% 46% 34%
Bound
It can be observed that TSS has a much POPULATION VARIABILITY
lower range than TP and TN. This can be Mean 91% 63% 51%
& SAMPLE NUMBER
attributed to the fact that the TSS data is Upper Confidence
more closely correlated and robust than 95% 80% 68% It is clear from Table 6 that the number
Bound
the other two data sets, as a narrower of paired samples required to achieve
confidence interval will indicate a more where the 95% confidence level can vary greatly
precise estimate, while a wider confidence depending on the target pollutant and
interval indicates a less precise estimate the variability of the particular data set.
of the true population mean (Tan and Tan, Next, setting the lower boundary Also, it might appear strange that there
2010). From Table 5 it can be concluded of the influent confidence interval to is less than one paired sample required
that we are 95% confident that the true the upper boundary of the effluent to satisfy the 95% confidence level for
population mean of the per cent removal confidence interval gives: the mean TSS removal figure (91%).
is between the lower and upper limit for Revisiting the TSS baseline data in
each pollutant type. Table 1, however, shows a very consistent
per cent removal rate for each event
MINIMUM NUMBER
regardless of influent concentration.
OF PAIRED SAMPLES The Coefficient of Variation (COV) is
Ensuring sufficient samples are collected then substituted for in the above Contrast this, however, with a different
for any type of field research should be equation using the standard relationship: TSS data set, using more paired samples
considered imperative to ensure that than shown in Table 1, where the results
statistically relevant conclusions can be will be very different. Table 7 is a set of
derived from the results. However, in It is more accurate to make the TSS data from a different site, using a
most stormwater data sets, no statistical assumption that as it is different treatment device to that used
processes are utilised to guide the highly likely that . Substituting for Table 1 data.
minimum sample size, and Burton and and rearranging the above equations and
Pitt (2002) suggest that most follow Table 7. Alternative example
solving for ‘n’ eventually yields: influent and effluent data for TSS for
subjective professional judgement. a different site and different treatment
device to that used in Table 1.
In order to numerically and statistically
show that there are sufficient data Event TSS EMCi TSS EMCe TSS
collected to ensure the required confidence Number (mg/L) (mg/L) CRE (%)
This formula will then calculate the
level, Burton and Pitt (2002) proposed minimum required number of paired 1 247 28 89
the use of a variety of basic equations, samples for a specific confidence level. 2 300 280 7
based on coefficients of variation, Z-scores, The results of this equation for the CRE
sample set means, and standard deviations. 3 233 113 52
test data can be seen in Table 6.
4 21 16 24
The Urban Stormwater BMP It is shown in Table 6 that TSS requires
Performance Monitoring Manual 5 32 23 28
less than one paired sample, TP requires
(Geosyntech, 2009), however, describes just over three paired samples, and TN 6 19 14 26
an equation specifically for sampling requires almost 10 paired samples to 7 55 40 27
stormwater treatment removal rates ensure a 95% confidence level of the
8 27 8 70
around a stormwater treatment device, mean % reductions in concentrations
based on similar methods described from Table 5. Given that there are 9 67 10 85
in Burton and Pitt (2002). The BMP 10 samples in the data set, and the 10 28 20 29
Performance Monitoring Manual minimum number of paired samples
11 70 35 50
(2009) begins with the formula defining required for each pollutant is all less
percentage removal (as a fraction than 10, we conclude this data set will 12 45 14 69
rather than a per cent): achieve statistical significance with a 95% 13 208 156 25
confidence level. Table 6 is a calculation 14 121 114 6
used to inform required sample numbers,
15 38 28 26
rather than an a posteriori calculation.
MAY 2015 WATER
Page 25
6
Technical Papers
because both false negatives and false Z score, should be chosen Z scores
Table 8. Minimum number of paired
influent/effluent samples required positives are important (see section on from the one-tailed distribution Z-scores
to establish the mean CRE% for TSS errors). In many field monitoring programs, (Teixeira et al., 2009). Given that,
removal at the 95% Confidence Level. power issues (such as false negative confidence level and statistical power
TSS problems) may actually be more critical should be balanced to ensure further
than confidence (Burton and Pitt, 2002). confidence in the results, but without
Z-Score (Zα/2) 1.96
TYPE I AND TYPE II ERRORS making the test too exclusive. Therefore,
COV 0.645 the confidence level can be reduced to
A Type I error can be defined as the
% Removal 41% 90% and the power increased to 80%.
probability of rejecting a null hypothesis
Z-scores for these values are:
Number of paired samples that is actually true (a false positive) and
required to meet 95% 24.32 is typically associated with the confidence
Confidence Interval level, or α value. The confidence level chosen
is the largest probability of committing an
allowable Type I error and still rejecting the Table 10 presents the results for the
It can be seen in Table 7 that there is
null hypothesis (Park, 2008). number of paired samples required if the
much greater variability around the TSS
treatment removal performance, with per Type II error is included with an 80% power
A Type II error can be defined as the level and balanced with a 90% confidence
cent removal varying from 6% to 89%.
probability of incorrectly retaining the null level requirement (using Table 1 data).
It is important to note that the SQIDEP
hypothesis (a false negative), and is typically
protocol (2014) calls for 15 qualifying When compared to Table 6, it can be
associated with statistical power, or β value.
events, a criterion that this data would seen in Table 10 that when the power is
The power of the test will determine the
satisfy. However, if we apply the minimum included, the number of paired samples
likelihood of selecting an incorrect null
number of paired samples calculations as required almost doubles, even though the
hypothesis, and is often used to determine
per the above, we can see in Table 8 that confidence level is reduced from 95% to
the required sample size (Park, 2008).
the data set falls well short of the required 90%. The importance of setting the power
number, at 24 paired samples required to In many environmental programs, the
to 80% cannot be underestimated, and
establish the mean CRE% for TSS (41%) importance of a Type II error (β) can be very
can be a critical element to increasing the
at the 95% confidence level. high. This would be equivalent to having
statistical quality of data presented.
a null hypothesis of “the water sample
The importance of calculating a
is not contaminated”, then retaining the LIMIT ON MINIMUM NUMBER
minimum number of paired samples
null hypothesis, when it was actually false. OF PAIRED SAMPLES
has been clearly established and it will be
This could lead to environmental damage,
necessary to explore this in great detail for This approach for assessing proprietory
health risks to the public, and possibly
the statistical validation of performance device performance claims recommends
fines imposed by regulatory agencies, as
data under SQIDEP. Burton and Pitt (2002), that field-testing is continued until the
a contaminated discharge of water would
however, take this calculation one step required number of paired samples is
occur (Burton and Pitt, 2002). A higher
further as we discuss next. met as per the criteria shown for Table
statistical power (1-β) can be used to
6 or 10. However, SQIDEP (Stormwater
CONFIDENCE LEVEL AND minimise this type of error, and should be
Australia, 2014) calls for a minimum
STATISTICAL POWER (α carefully selected. Typically, a power of
number of monitoring events that are
AND β VALUES) 80% (β = 0.2) is recommended, whereas
if the Type II error is to be ignored, then a statistically relevant.
Most statistical tests convey a certain
power of 50% (β = 0.5) is adopted. Given the variable nature of field-
required level of confidence (α). This
indicates and sets the criteria for a Therefore, if statistical power were to testing stormwater treatment devices, the
decision about the null hypothesis for the be included in the above equation used to number of paired samples required can
test (Privitera, 2011). Statistical power (β), calculate the minimum number of paired be very high if the data is not particularly
on the other hand, is frequently ignored, samples required, the equation becomes: consistent, as shown by the TSS removal
even though it can sometimes be more data in Table 8. One solution could be
important than confidence level (Burton that, for the full performance claims to be
and Pitt, 2002). verified, the 90% confidence interval and
where 80% power requirements must be met.
Confidence levels are defined as:
However, if the required number of paired
samples has not been achieved under
these statistical criteria, then perhaps the
Whereas statistical power can be
It should be noted that while overall performance of the device could
defined as:
should be chosen from the two-tailed be reduced by a scaling factor. Further
Table 9. Summary of the available choices with Type I and Type II Errors when
When evaluating data using a statistical assessing the null hypothesis.
test, power is the sensitivity of the test
Do Not Reject H0 Reject H0
for rejecting the hypothesis. It is important
that power and confidence be balanced Correct Decision Type I Error
H0 is True
for an effective monitoring program. Most 1 – α: Confidence Level α: Size of Test (Significance Level)
studies ignore power, while providing Type II Error Correct Decision
a high value (typically 95%) for the level H0 is False
β: Size of Test (Power Level) 1 – β: Power of Test
of confidence. This is a flawed approach
WATER MAY 2015
Page 26
7
Technical Papers
Table 10. Number of paired samples THE AUTHORS McDonald J (2014): Paired t-test, Log Normal
required to establish CRE% at the Transformations and Wilcoxon Ranked Sign Test,
Charles Kelly (email: charles.
90% Confidence level and 80% Handbook of Biological Statistics, 3rd ed., Sparky
Power level. [email protected]) has over House Publishing, Baltimore, Maryland.
16 years’ experience in the
TSS TP TN construction, consulting and Neyman J (1937): Outline of a Theory of Statistical
Estimation Based on the Classical Theory of
Z-Score (Zα/2) 1.64 1.64 1.64 manufacturing industries.
Probability, University College, London.
He is currently Commercial
Z-Score (Zβ) 0.84 0.84 0.84
Manager – Humes Water Solutions and is Olsson U (2005): Confidence Intervals for the Mean
COV 0.07 0.44 0.54 undertaking research and development of of a Log-Normal Distribution, Swedish University
existing and new stormwater treatment of Agricultural Sciences, Journal of Statistics
% Removal 91% 63% 51% Education, 13, No 1.
products for the Australian market.
Number of paired Charles holds a Bachelor’s degree in Park H (2008): Hypothesis Testing and Statistical
samples required Environmental Engineering, and a PhD in
0.05 5.57 15.9 Power of a Test, The University Information
to meet 90%
Confidence Interval
Mechanical/Environmental Engineering. Technology Services (UITS) Center for Statistical
and Mathematical Computing, Indiana
Anton Bardak is a Civil University, US.
engagement and consultation is required
Engineer with three years’
to determine the most appropriate Privitera G (2011): Chapter 8: Introduction to
experience in the water
approach for calculating the minimum Hypothesis Testing, Statistics for the Behavioral
engineering industry.
number of required paired samples. Sciences, 225, Sage Publications.
After two years with KBR
CONCLUSION as a graduate water and Razali M & Wah Y (2011): Power Comparisons of
wastewater engineer, he joined Humes Shapiro-Wilk, Kolmogorov-Smirnov, Lilliefors and
The draft Stormwater Australia
as a Water Solutions Engineer and now Anderson-Darling Tests, Journal of Statistical
Stormwater Quality Improvement Device Modelling and Analytics, 2, pp 21–33.
Evaluation Protocol (SQIDEP) requires focuses solely in the area of stormwater
inter alia, statistical assessment of the management. Anton provides technical Seo S (2002): A Review and Comparison of
data presented. It is necessary to engage support and designs for the national Methods for Detecting Outliers in Univariate
with academia and regulators to reach Humes business and is also heavily involved Data Sets, Graduate School of Public Health,
in ongoing research and development. Pittsburgh University, US.
a consensus on the robustness of a
statistical approach, and this paper is Shapiro S & Wilk M (1965): An Analysis of
REFERENCES
offered as an exemplar approach to Variance Test for Normality (Complete
Bonacci K (2012): Central Limit Theorem, Indiana
help inform the discussion. Samples), Biometrika, 52, p 591, Great Britain.
University Southeast, viewed at homepages.
ius.edu/kbonacci/HOMEPAGE%20K300_files/ Shier R (2004): Statistics: Paired t-tests,
The following statistical procedures K300%206-5.pdf Mathematics Learning Support Centre, www.
have been proposed:
Burton A & Pitt R (2002): Stormwater Effects statstutor.ac.uk/resources/uploaded/paired-t-
1. Outlier Identification in raw data Handbook: A Toolbox for Watershed Managers, test.pdf
and removal if justified Scientists, and Engineers, Lewis Publishers.
Singh A, Singh A & Engelhardt M (1997): The
Cramér H (1946): Mathematical Methods of Lognormal Distribution in Environmental
2. Normal distribution validation of raw
Statistics, Princeton University Press, Asia Applications, United States Environmental
data (data transformation conducted Publishing House.
Protection Agency, EPA/600/S-97/006.
if required)
Duncan H (1999): Urban Stormwater Quality:
Stormwater Australia (2014): Evaluation
3. Statistical significance of influent and A Statistical Overview (Report 99/3), Monash
University, Melbourne, Australia. Protocol (SQIDEP) for Stormwater Quality
effluent data (paired t-test, or a non- Treatment Devices, Consultation Release by
parametric alternative if not normally Geosyntec Consultants & Wright Water Engineers
Stormwater Australia viewed at stormwater.
distributed) (2009): Urban Stormwater BMP Performance
asn.au/images/SQID/SQIDEP_Release_version_
Monitoring, Water Environment Research
Foundation, viewed at www.bmpdatabase. December_2014.pdf
4. Confidence Interval calculation
of removal efficiency metric org/Docs/2009%20Stormwater%20BMP%20 Tan S & Tan S (2010): The Correct Interpretation of
Monitoring%20Manual%20(Interim%20Ch%20 Confidence Intervals, Proceedings of Singapore
5. Minimum number of paired samples 1-6,%20App).pdf
Healthcare, 19, 3.
calculated. Geyer C (2003): Nonparametric Tests and
Teixeira A, Rosa A & Calapez T (2009): Statistical
Confidence Intervals, School of Statistics,
Some discussion is still required University of Minnesota. Power Analysis with Microsoft Excel: Normal
around point 5, but if all the steps Tests for One or Two Means as a Prelude to
Herr C & Sansalone J (2015): In Situ Volumetric
above are completed and the statistical Using Non-Central Distributions to Calculate
Filtration Physical Model to Separate
tests are satisfied, then consultants and Power, Journal of Statistics Education, 17, 1.
Particulate Matter from Stormwater. Journal
authorities will have a sound scientific of Environmental Engineering, 10.1061/(ASCE) Tukey J (1977): Exploratory Data Analysis, Addison-
basis for accepting treatment efficacy EE.1943-7870.0000946, 04015017. Wesley.
claims of the device. Kelly C & Bardak A (2015): Evaluation of Treatment Walfish S (2006): A Review of Statistical Outlier
Performance of a Stormwater Treatment
Just as importantly, this process Methods. Pharmaceutical Technology, viewed
Membrane Filter under Australian Conditions,
will establish objective performance at www.pharmtech.com/review-statistical-outlier-
Australian Journal of Water Research, 2, 2,
evaluations, reduce contestable decisions; pp 2203–9490. methods.
result in benefits for the community and Limpert E, Stahel W & Abbt M (2001): Log-Normal Wang F (2001): Confidence Interval for the Mean
authorities; and allow for a clear path to Distributions Across the Sciences: Keys and of Non-Normal Data, Quality and Reliability
market for device proprietors. Clues, BioScience, 51, 5. Engineering International, 17, pp 257–267.
MAY 2015 WATER
Page 27
Appendix C: Supplementary Data
Rainfall Runoff
Number
Rainfall Rainfall Peak Peak Median
Date Runoff of
Depth Duration Intensity Runoff Runoff
Sampled Volume (L) Samples
(mm) (minutes) (mm/hr) (L/s) (L/s)
Taken
2014-06-28 5.2 19 57 163 34 7000 3
2014-08-16 33 1222 27.4 178 0.7 51951 24
2014-08-27 7.2 128 13.7 83.5 1.8 10445 14
2014-09-25 8.6 214 18.9 124 13.3 12698 4
2014-10-13 7 597 15.4 20.5 0.3 2420 13
2014-10-27 3 36 10.3 5.3 0.4 646 7
2014-11-06 5.2 36 20.6 69.3 4.1 6209 8
2014-11-28 4.8 182 15.1 6.6 0.5 1672 11
2014-12-06 11.2 277 20.6 8.7 1.3 23479 24
2015-03-19 10.2 204 41.1 6.9 0.5 11449 15
[OCR of page images]
Appendix C: Supplementary Data
Rainfall Runoff
Rainfall | Rainfall | Peak | Peak | Median Number
Date . : Runoff of
Depth Duration | Intensity | Runoff | Runoff
Sampled . Volume (L) Samples
(mm) (minutes) | (mm/hr) | (L/s) (L/s)
Taken
2014-06-28 5.2 19 57 163 34 7000 3
2014-08-16 33 1222 27.4 178 0.7 51951 24
2014-08-27 7.2 128 13.7 83.5 1.8 10445 14
2014-09-25 8.6 214 18.9 124 13.3 12698 4
2014-10-13 7 597 15.4 20.5 0.3 2420 13
2014-10-27 3 36 10.3 5.3 0.4 646 7
2014-11-06 5.2 36 20.6 69.3 4.1 6209 8
2014-11-28 4.8 182 15.1 6.6 0.5 1672 11
2014-12-06 11.2 277 20.6 8.7 1.3 23479 24
2015-03-19 10.2 204 41.1 6.9 0.5 11449 15
Page 28
National sales 1300 361 601 humes.com.au [email protected] A Division of Holcim Australia [OCR of page images] National sales 1300 361 601 humes.com.au [email protected] PST mate rec)
Used in these records
Jellyfish Compliance with SQIDEP.pdfRelated documentApplication case desk ↗
About this file
- File size
- 2.7 MB (large file)
- File type
No other editions are linked here.
Original file
Log in to download the original file
The download may take up to a minute.