United States
Environmental Protection
Agency
Testing Solids Settling
Apparatuses for Design and
Operation of Wet-Weather Flow
Solids-Liquid Separation
Processes
Research Report
-------
EPA/600/R-02/090
October 2002
Testing Solids Settling Apparatuses for Design
and Operation of Wet-Weather Flow
Solids-Liquid Separation Processes
by
Thomas P. O'Connor, David Fischer and Richard Field
Water Supply and Water Resources Division
National Risk Management Research Laboratory
Edison, New Jersey 08837
U.S.A.
John Cigana, BenoTt Gagne and Martin Couture
John Meunier Inc.
Saint-Laurent, Quebec H4S 2B3
Canada
CRADANo. 136-96
Project Officer
Richard Field
Urban Watershed Management Branch
Water Supply & Water Resources Division
National Risk Management Research Laboratory
Edison, NJ 08837
National Risk Management Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
-------
Notice
The information in this document contains information from a project supported by the United States
Environmental Protection Agency under CRADANo. 136-96. Although it has been subjected to the
Agency's peer and administrative review and has been approved for publication as an EPA document,
it does not necessarily reflect the views of the Agency and no official endorsement should be inferred.
Also, the mention of trade names or commercial products does not imply endorsement by the United
States government.
11
-------
Foreword
The U.S. Environmental Protection Agency is charged by Congress with protecting the
Nation's land, air, and water resources. Under a mandate of national environmental laws, the Agency
strives to formulate and implement actions leading to a compatible balance between human activities
and the ability of natural systems to support and nurture life. To meet this mandate, EPA's research
program is providing data and technical support for solving environmental problems today and
building a science knowledge base necessary to manage our ecological resources wisely, understand
how pollutants affect our health, and prevent or reduce environmental risks in the future.
The National Risk Management Research Laboratory is the Agency' s center for investigation
of technological and management approaches for preventing and reducing risks from pollution that
threatens human health and the environment. The focus of the Laboratory's research program is on
methods and their cost-effectiveness for prevention and control of pollution to air, land, water, and
subsurface resources; protection of water quality in public water systems; remediation of contaminated
sites, sediments and ground water; prevention and control of indoor air pollution; and restoration of
ecosystems. NRMRL collaborates with both public and private sector partners to foster technologies
that reduce the cost of compliance and to anticipate emerging problems. NRMRL's research provides
solutions to environmental problems by: developing and promoting technologies that protect and
improve the environment; advancing scientific and engineering information to support regulatory and
policy decisions; and providing the technical support and information transfer to ensure
implementation of environmental regulations and strategies at the national, state, and community
levels.
This publication has been produced as part of the Laboratory's strategic long-term research
plan. It is published and made available by EPA's Office of Research and Development to assist the
user community and to link researchers with their clients.
E. Timothy Oppelt, Director
National Risk Management Research Laboratory
in
-------
Abstract
This study was a side-by-side comparison of two settling column tests: one traditional and one
new. The newer apparatus was developed by the Centre d'Enseignement et de Recherche pour la
Gestion des Ressources Naturelles et de I'Environnement (CERGRENE) of France and uses several
small columns to sequentially measure particle-settling velocities. The new apparatus was compared
with a larger, more traditional column, which has been widely used in the past as a research and
academic tool, but it is difficult to transport and set up in a field location due to its size. The newer
settling testing method was thought to be more amenable to field use because of ease of transport and
sampling and the limited number of samples generated.
The study was conducted in three phases: fabrication and preliminary testing, laboratory
testing, and field testing. Equipment for the two testing methods was fabricated and laboratory tested
and preliminary evaluations were made. Laboratory tests were conducted with two well characterized
settling media, microsand and clay soil in order to measure suspended solid (SS) concentrations and
develop settling distributions of known substances in the columns prior to testing actual WWF which
exhibits variable SS concentrations and settling distributions. Field tests were conducted at a
combined sewer control structure to compare the performance of the two columns when filled with
combined sewage.
A summary of the performance as measured by predicted percent removal of both columns for
15 laboratory tests and 3 field tests is presented, as well as a comparison of the advantages and
disadvantages of the two methods. The newer testing method (CERGRENE) did not perform up to
the anticipated theoretical expectations of the method. The report ends with conclusions and
recommendations regarding the two specific methods and settleability test in general.
IV
-------
Contents
Notice ii
Foreward iii
Abstract iv
Contents v
List of Figures vii
List of Tables vii
Abbreviations viii
Acknowledgment ix
1. INTRODUCTION 1-1
Background 1-1
Objectives 1-2
Combined Sewer Overflows 1-3
Settling Columns 1-4
Traditional Long Column 1-4
CERGRENE Columns 1-4
Other Columns 1-5
Theory of Settling Design 1-5
Field Sampling 1-6
Field Site 1-6
Field Sampling Review 1-7
2. MATERIALS AND METHODS 2-1
Conclusion of Phase I: Preliminary Testing 2-1
Column Description and Delivery 2-1
Long Column 2-1
CERGRENE Columns 2-3
Mixing Basins 2-4
Sampling 2-5
Long Column 2-5
CERGRENE Columns 2-6
Quality Assurance Samples 2-6
Solids Analyses 2-6
Identification of Experimental Materials for Phase II 2-8
Experimental Design 2-9
3. RESULTS OF PHASE H 3-1
Quality Control Analysis 3-1
Blanks and Standards 3-1
Completeness 3-2
Mixing Basin - Recycle 3-4
Settleable Solids 3-4
Laboratory Experiments 1-15 3-5
Concentration versus Time 3-5
Long Column Shortcomings - Initial Concentration Gradient 3-5
CERGRENE Shortcomings - Lack of Repeatable Results 3-6
Percent Removal Long Versus CERGRENE 3-8
Matrix Iteration Process for CERGRENE Columns 3-8
Eckenfelder Analysis for Long Column 3-10
Design Removal Comparison 3-10
-------
Initial Concentration Comparison 3-13
Discussion and Recommendations 3-13
4. RESULTS OF ADDITIONAL LABORATORY SAMPLING 4-1
Quality Control Analysis 4-2
Blanks and Standards 4-2
Completeness 4-3
Laboratory Duplicate 4-3
Results of Experiments 1 and 2 4-3
Other Considerations 4-5
Conclusions and Recommendations 4-6
5. RESULTS OF PHASE m FIELD TESTING 5-1
Experimental Design 5-2
Quality Control Analysis 5-4
Blanks and Standards 5-4
Completeness 5-6
Additional QA Concerns: Ratio of Total Solids to Suspended Solids 5-6
Field Experiments 1-3 5-7
Mixing Basin - Background 5-7
nitial Concentration 5-8
CERGRENE Duplicate Analysis 5-9
Concentration versus Time 5-10
Percent Removal Long Versus CERGRENE 5-10
Matrix Iteration Process for CERGRENE Columns 5-11
Design Removal Comparison 5-12
6. CONCLUSIONS AND RECOMMENDATIONS 6-1
General Conclusions 6-1
Experimental Conclusions 6-1
Discussion and Recommendations 6-3
7.REFERENCES 7-7
APPENDICES
A Qmplig Qif Stipuhpxht Binning T>reycJefft (P»4 SftPMfrPX? Mtfifldfl; Phagft IT A-l
B Graphs of Concentration Venus Time for Long and CERGRENE: Phase nB-l
C Graphs of Percent Removal for Long and CERGRENE:
Phase n Experiments 10-15 C-l
D Resuhs of VICTOR Analysis: Phase II D-l
E Graphs of Concentration Versus Time for Long and CERGRENE:
Phase m Experiments 1-3 E-l
F Gi^hs of Percent Removd for Long and CERGRENE: Phase m Experiments 1-3 I l
G Results of VICTOR Analysis: Phase m Experiments 1-3 G-l
H Data Sets Phase H H-l
VI
-------
List of Figures
Figure 1-1 Flowchart of Project 1-2
Figure 2-1. Long Column General Arrangement 2-2
Figure 2-2. CERGRENE® Column Configuration 2-3
Figure 2-3. Configuration for Side-by-Side Phase II Experiments 2-11
Figure 3-1 Comparison of Removals of Microsand 3-12
Figure 3-2 Comparison of Removals of Microsand with Error Included 3-12
Figure 3-3 Comparison of Initial Concentrations 3-13
Figure 4-1 New configuration with additional flask to overflow CERGRENE column 4-2
Figure 5-1 Configuration for Side-by-Side Field Analysis 5-3
Figure 6-1A Suggested Modifications to the CERGRENE Column Design 6-5
Figure 6-IB Suggested Modifications to the CERGRENE Column Design 6-6
List of Tables
Table 2.1 Summary of Standard Methods and Procedures 2-7
Table 2.2 Experimental Design for Phase II Laboratory Experiments 2-9
Table 2.3 Critical Time Measurements for a Microsand Experiment, Phase II 2-10
Table 3.1 Estimated Number of SS Analysis for One Experiment of Phase II 3-1
Table 3.2 QA Objectives for Measurements 3-1
Table 3.3 Completeness of Suspended Solids Analysis 3-3
Table 3.4 Duplicate Analysis for Recycle Concentration and CERGRENE Columns 3-7
Table 3.5 Measured Overflow Rate and Predicted Percent Removal 3-11
Table 4.1 Completeness of Suspended Solids Analysis 4-3
Table 4.2 Experiment 1 Suspended Solids Concentrations of Neshaminy Clay Soil 4-4
Table 4.3 Experiment 2 Suspended Solids Concentrations of Microsand 4-4
Table 5.1 Analysis of Blanks 5-4
Table 5.2 QA Objectives for Measurements 5-5
Table 5.3 Completeness of Solids Analysis 5-6
Table 5.4 Comparison of Total Solids to Suspended Solids Ratio 5-7
Table 5.5 Solids Concentrations in Mixing Basin during Field Experiments 5-8
Table 5.6 Comparison of Initial Column Concentrations to Background 5-9
Table 5.7 CERGRENE Duplicate Analysis 5-9
Table 5.8 Comparison Predicted Removal between Long and CERGRENE Columns 5-13
vn
-------
Abbreviations
Centre d'Enseignement et de Recherche pour la Gestion
des Ressources Naturelles et de 1'Environnement CERGRENE
Combined Sewer Overflow CSO
Cooperative Research and Development Agreement CRADA
Dry Weather Flow DWF
Environmental Protection Agency EPA
Quality Assurance QA
Relative Percent Difference RPD
Relative Standard Deviation RSD
Standard Deviation s
Suspended Solids SS
Total Solids TS
Urban Watershed Management Branch UWMB
U.S. Environmental Protection Agency EPA
United States Infrastructure, Inc. USI
Volatile Suspended Solids VSS
Wastewater Treatment Plant WWTP
Wet Weather Flow WWF
Vlll
-------
Acknowledgments
This research was supported in part by an appointment to the Postgraduate Research
Participation Program at the National Risk Management Laboratory, U.S. Environmental Protection
Agency (EPA), and was administered by the Oak Ridge Institute for Science and Education (ORISE)
through an interagency agreement between the U.S. Department of Energy and the EPA.
The efforts of Larry Fradkin of the EPA's Office of Science Policy who assisted with the
establi shment of the Cooperative Research and Development Agreement (CRAD A13 6-96) are greatly
appreciated.
Chi-Yuan (Evan) Fan of the EPA's Urban Watershed Management Branch (UWMB) lent
invaluable support to this project by establishing the CRAD A, reviewing reports, assisting in field
sampling and providing technical input to the project.
Thanks to Robert Gibson, Wastewater Manager of Perth Amboy, New Jersey and Michael
Olexa, formerly Superintendent of Sewers of Perth Amboy for allowing us to sample at the Perth
Amboy Sewage treatment plant for the field testing.
Thanks also to Dr. Asim Ray of UWMB and Jess Pritts of the EPA's Office of Water for
reviewing the Interim report. Others from UWMB, Ron Rovansek, an ORISE Postgraduate
researcher, and Phil Kao, Steve Oliveri, and Sang Yi, Environmental Career Organization interns,
assisted during Phase III sampling. United States Infrastructure, Inc. performed the solids testing
during Phase III.
IX
-------
1. INTRODUCTION
Background
This final report pertains to the laboratory and field evaluations of a side-by-side comparison
of two settling column tests: one traditional and one new. The newer apparatus was developed by
the Centre d'Enseignement et de Recherche pour la Gestion des Ressources Naturelles et de
I'Environnement (CERGRENE) of France and uses several small columns to sequentially measure
particle-settling velocities. This method was adapted for North American application by John
Meunier, Inc. of Montreal, Quebec, Canada. Particle-settling-velocity distribution and pollutant
content measurements made with the apparatus are intended to be used for wet-weather flow (WWF)
treatment process selection and design, and for evaluation of preliminary or existing process
operations that depend on solids-liquid separation.
The U.S. Environmental Protection Agency (EPA) National Risk Management Research
Laboratory's Water Supply and Water Resources Division, Urban Watershed Management Branch
(UWMB) in Edison, New Jersey, and John Meunier Inc. in Montreal, Quebec, Canada, established
a Cooperative Research and Development Agreement (CRADA) (136-96) to develop settling
columns suitable for obtaining particle-settling-velocity distribution data for WWF. This CRADA
compared the new apparatus with a larger, more traditional column, and detailed the advantages and
disadvantages of each method, quality assurance (QA) procedures, expected results, and the
limitations for both settling-velocity-distribution tests for WWF. John Meunier, Inc. and the
UWMB were jointly responsible for the fabrication, testing, and field evaluation of both
technologies compared.
The traditional column has been widely used in the past as a research and academic tool, but
it is difficult to transport and set up in a field location due to its size. The newer settling testing
method was thought to be more amenable to field use because of ease of transport and sampling and
the limited number of samples generated. The comparison attempted to predict whether these tests
can capture the solids in WWF, particularly the rapidly settling particles, and whether both systems
provide similar design information. Measurements of suspended solids (SS) for several settling times
were used to compare the methods. A summary of each column's performance as measured by
percent removal for 15 laboratory bench-scale and 3 field experiments is presented.
This study was performed in three phases (a flow chart Figure 1-1 also shows project
development):
• Phase I: Preliminary screening - In phase I, different types of particles were tested to select
the best media to be used in the benchtop laboratory studies and to assist in determining
sampling and analysis procedures, experimental parameters, and number of samples
required. Aspects of this phase were performed independently by both parties (UWMB and
John Meunier, Inc.). Procedures were then adjusted to allow for any difficulties
encountered.
• Phase II: Laboratory bench-scale experiments - The official QA-approved experimental
runs of the side-by-side analysis of the two settling characterization methods were conducted
in the laboratory of John Meunier, Inc. by both parties.
• Phase III: Field study - The side-by-side comparative study was continued at an offsite
location (Perth Amboy Sewage treatment plant, Perth Amboy, New Jersey) predominantly
by UWMB.
Chapters 3 and 5 cover most of the experimental results for Phases II and III, respectively.
1 - 1
-------
Chapters 1 and 2 provide background information, Phase I findings, and experimental design for
Phase II and III. In addition to the distinct phases of the experiment, there was some additional
laboratory testing performed between phases II and III which is covered in Chapter 4. Chapter 6
discusses conclusions and specific recommendations.
This side-by-side comparison was intended to determine the limitations and advantages (e.g.,
cost, setup requirements, correlation to actual settling [in a primary sedimentation tank]) of each
approach. Onsite settling column sampling better represents settling velocities than offsite
(laboratory) testing because sample storage and transport prior to the settling test may change the
naturally occurring matrix of settleable material. The field settling samples were delivered
immediately to the UWMB laboratories and analyzed for various solids parameters. The Perth
Amboy Sewage treatment plant was approved by both parties for the sampling of combined sewer
overflow (CSO).
Development of experimental procedures,
Phase I shakedown of equippment, testing of
various settling media
i x
l\
15 Experiments, 5 experiments for each type of media
(microsand, Neshaminy soil and mixture)
pu|| BOX Experimental Design with Duplicate of Experiment
Including 1 duplicate run Experiments performed in random order
\/
PFRnRFNF 2 Experiments - one experiment each with
CYDCDIIMTC microsand and Neshaminy soil - testing
EXPERIMENTb overflowing procedure
Planned 4 Field Experiments -
Phase 111 Completed 3 for CERGRENE Column and two for LONG Column
4 data sets for CERGRENE and 3 data sets for LONG
Figure 1-1 Flowchart of Project
Objectives
The monitoring and analysis needed for proper selection, application, assessment, design,
and evaluation of WWF treatment are expensive, time consuming, and complex; however, reliable
data collection may save even more costly construction costs by eliminating unnecessary facilities
and/or additional controls. The particle-settling-velocity distributions of WWF samples as related
to total solids (TS), SS, and associated pollutant content are essential for proper assessment of
high-rate settling and solids-liquid separation technologies.
The objective of this study was to compare sampling and analytical procedures of two
settling column techniques used to characterize the settling velocity of SS in WWF. The results
1 -2
-------
were intended to aid engineers in obtaining pertinent WWF pollution-abatement facility selection
and design data by analyzing particle-settling-velocity distribution and the settleable solids and SS
fractions.
Low cost, expedient methods to obtain facility-design data or settling-velocity distributions
are necessary because WWF characteristics are highly site specific. In order to test the viability of
the newly developed CERGRENE columns, a comparison to a settling method with a precedent was
needed. Thus, a traditional column settling method was used.
Combined Sewer Overflows
CSO are a mixture of storm drainage and municipal-industrial wastewater discharged from
combined sewers typically when the flow capacity of a sewer system is exceeded during rainstorms.
The EPA National Combined Sewer Overflow Control Policy (59 Federal Register 18688) (CSO
Policy) guidance "Combined Sewer Overflow - Guidance for Nine Minimum Controls" (EPA, 1995)
requires:
• Maximization of flow to the publically owned treatment works (POTW) for treatment
• Control of solid and floatable materials in CSOs
and "Combined Sewer Overflow - Guidance for Long-Term Control Plan" (EPA, 1995) further
requires:
• Characterization, monitoring, and modeling activities as the basis for selection and design
of effective CSO controls
• Evaluation of alternatives that will enable the permittee, in consultation with the National
Pollutant Discharge Elimination System (NPDES) permitting authority, water quality
standard (WQS) authority, and the public, to select CSO controls that will meet Clean Water
Act (CWA) requirements
• Cost/performance considerations to demonstrate the relationships among a comprehensive
set of reasonable control alternatives
• Maximization of treatment at the existing POTW for WWFs
The CSO Policy recommends control/treatment without defining the need for analysis of the
flow characteristics and constituents to obtain design information. Determining certain flow
characteristics and constituents will optimize the selection and design of unit processes for various
degrees of established physical treatment technologies, e.g., vortex separation, screening,
sedimentation, flocculation-clarification, dissolved air flotation, and filtration, and assist in the
assessment of newer technologies, e.g., microcarrier coagulation-sedimentation processes. Site
specific, storm-event evaluations are needed for designing CSO treatment facilities, as CSO differs
from dry-weather flow (DWF). Combined sewer overflow settleable solids build-up and
characteristics in the sewer system are a function of the length of the antecedent dry-weather period,
sewer slope and cross-sectional area, drainage area (catchment) size, flowrate, and drainage area soil
characteristics, etc., whereas DWF SS characteristics (barring industrial sources) are similar from
place to place. Furthermore, settleable solids and SS concentrations can vary with time during storm
events and from storm to storm.
Studies have identified urban stormwater runoff as a major contributor to the degradation
of many urban lakes, streams, and rivers. Industrial and commercial parking lots, material storage
areas, and vehicular service stations are the most significant contributors of a variety of pollutants
to WWF. Chebbo et al. (1990) found that the fine particles which make up the majority of SS are
also the principal vector of pollution in stormwater during wet weather. Fine particles (< 50 • m)
found in stormwater can achieve settling velocities of 2.5 m/h (0.07 cm/s) or more (Chebbo et al.,
1 -3
-------
1990); 70 to 80% will deposit within 15 min and more than 97% after 1 h.
Settling Columns
The traditional method for determining settling-velocity distributions is a batch test that uses
a large column equipped with vertically spaced side ports or syphon tubes to obtain samples for
solids analysis. Camp (1945) published settling curves using Stoke's Law based on particle settling.
Eckenfelder (1966) described the use of a large column as a design aid for sedimentation processes
and for analysis of flocculation. There is substantial variability associated with this method (referred
to as the Long column in the remainder of the document).
Currently, only one method for measuring gravity separation is accepted by Standard
Methods (SM 2540.F.b; 19th Edition), called "settleable solids". A sample is pipetted from the
center of a column of at least 20 cm in depth after 1 h of quiescent batch settling to directly
determine the nonsettleable solids. The settleable solids are then determined by subtracting the
nonsettleable solids concentration from the initial SS concentration. However, this method neither
determines particle-settling velocity nor enables calculations for settling-velocity-distribution
curves. This gravimetric method only measures the initial and final SS concentration after 1 h.
There are no quality control limits or QA data for the method.
Traditional Long Column
The typical traditional Long column is a relatively large apparatus (Camp, 1945;
Eckenfelder, 1966; Dalrymple et al., 1975; in addition to being described elsewhere), standing 1.8
to 2.5 m (6 to 8 ft) high with a diameter of 20 to 30 cm (8 to 12 in.) and side withdrawal ports evenly
spaced along the column depth. The height of the column simulates the effective settling depth
which occurs in a sedimentation tank typically having constructed depths exceeding 2.5 m (8 ft).
The Long column requires an extensive laboratory layout and is not readily adapted for use in field
situations. Depending on specific dimensions, between 40 and 80 L (10 and 20 gal) are required to
fill the column. The initial water height in the column is measured. The samples, withdrawn from
the side ports sequentially from top to bottom at predetermined time intervals, require further SS
analysis. After each set of samples is collected, the depth of the water in the column is measured.
The most notable difficulty with the Long column method is the inability to develop a
homogeneous initial SS concentration at the initial sampling time, t0, due to the heavy particles in
WWF. This is partially caused by keeping the large volume in the Long column well mixed prior
to sampling and the time required to withdraw samples from all ports sequentially. Various methods
have been used to pre-mix the sample before the column test begins, e.g., plunger plates. Pisano et
al. (1984) went to the extent of mounting the Long column on a device that allowed horizontal and
vertical axial rotation in an attempt to achieve a better estimate of SS concentration at t^ It is almost
impossible to have a homogeneously mixed sample using the Long column for WWF, which may
result in predictions of lower than actual SS concentrations.
CERGRENE Columns
CERGRENE (Chadirat et al., 1997) developed anew design that uses several small columns
to analyze the particle-settling velocities. Instead of sampling various fractions with a single
sampling device, the CERGRENE protocol uses different settling columns. The CERGRENE
settling columns, like the Long column, are designed to estimate SS concentrations of the WWF in
the original sewage matrix.
The new CERGRENE columns have a shorter time to fill (< 10 s) which should provide an
aliquot that better represents the completely mixed contents at the initial sample time, t0. It is
thought that the CERGRENE column may account for a wider range of settling solids which may
result in establishing better design parameters for WWF. The CERGRENE column was designed
1 -4
-------
for field as well as laboratory use. Settling-velocity-distribution samples taken in the field should
give a truer representation of the settling velocities of the combined sewage. This should also
provide a better representation for design purposes than samples transported to the lab or stored in
the lab for longer periods of time, due to less time allowed for agglomeration.
Other Columns
Other settling-velocity-distribution methods developed in Europe are:
• Brombach or German (Michelbach and Wohrle ,1993; Pisano and Brombach, 1996);
• Norwegian Institute for Water Research (NIVA) (Lygren and Damhaug, 1986; Walker et al.,
1993); and
University of Aston, U.K. (Tyack et al., 1993)
These methods were specifically designed for the relatively high concentration of heavier particles
in storm-generated flows, and accordingly, may offer several benefits over the Long column for field
analysis. They require fewer analyses, yielding one sample per time measurement withdrawn from
the bottom as opposed to several simultaneous samples from the multiple-side ports. These devices
use testing volumes of approximately 4 to 12 L (1 to 4 gal). The Brombach and the NIVA columns
(less than 1 m deep and 5 cm wide) are more amenable to field use. These methods also provide
a truer representation of high settling-velocity SS because the concentrated sample is situated above
the settling column and dropped into it at t0. The Aston column stands at least 2.2 m tall and
requires more assembly than the other two, as it rotates about the center of the column.
Unlike the Long and CERGRENE settling column designs which sample the WWF mixture,
the Brombach and NIVA methods separate, dry, and then reintroduce the SS into clean water. The
Aston column was previously tested directly against various forms of the CERGRENE column
(Aiguier et al., 1995). Results of these tests suggested that the derived settling-velocity curves from
the various innovative methods tend to give different results. For this reason, only the Long and the
CERGRENE columns were analyzed for the purposes of this project.
Theory of Settling Design
Several factors are used in the design of a settling basin including design flow, required
detention time, and desired percent removal. In order to design for percent removal, the
characteristics of the SS in the WWF must be taken into account. For settling tanks (Tchobanoglous
and Burton, 1991), the design velocity, Vc (m/s), can be related to the liquid depth, D (m), in the tank
and the detention time, td (s), as follows:
Vc = D/td (1-1)
Given a certain flowrate through the settling tank, Q (m3/s), and the plan area of the tank, A (m2),
Vc (m/s) can be related to the overflow rate, q (m3/m2/s or m/s), in the following manner:
Vc = q = Q/A (1-2)
This assumes that all SS with a settling velocity > Vc or q will be removed with some fraction of all
other particles with settling velocities < Vc also being removed. For the purposes of this project, q
will be used as a surrogate for a design settling velocity, instead of Vc, which inherently implies a
single design settling velocity for a particle instead of a settling-velocity distribution.
Settling can also be broken down into four types: discrete, flocculant, hindered, and
compression (Tchobanoglous and Burton, 1991). The settling velocities of discrete and flocculant
particles are of most concern with respect to WWF. The hindered and compression zones of settling
are issues of high concentration waste streams, which typically occur at a wastewater treatment plant
(WWTP) handling sanitary sewage in secondary-settling tanks and sludge-handling devices or
1 -5
-------
industrial applications.
Various studies used discrete settleable solids and various column devices or settling
methods to determine settling velocities. In Stoke's Law (Equation 1-3), the velocity of an ideal
sphere is proportional to the square of the particle diameter and is represented as follows:
(1-3)
where:
vs = velocity of sphere, m/s
g = acceleration due to gravity, 981 m/s2
••= density of the particle, kg/m3
• •= density of the fluid
d = diameter of sphere, m
• •= dynamic viscosity, N*s/m2
As previously mentioned, settling columns have been used to observe and analyze flocculant
settling. Settling of WWF is often a combination of discrete and flocculant settling.
While direct measurement of sedimentation efficiency can be made on WWF storage and
treatment facilities after installation by taking grab samples at the influent and effluent of the
facilities, the settling column and its predicted removals can assist the engineer in the selection of
design parameters before installation for WWF facilities. Settling columns can help determine the
settling-velocity distributions for local conditions, e.g., WWF and waste streams containing silty
particles may require larger facilities for a desired percent removal, while WWF waste streams with
gritty particles could achieve the same percent removals with much smaller facilities. Observation
of the actual settling-velocity distribution of SS in WWF is a better basis for design than the
representation of ideal settling velocity derived from Stoke's Law, which only relies on the ideal
settling of discrete monolithic particles.
Field Sampling
In order to collect a representative sample of WWF, sampling devices must be able to capture
the heavier SS or settleable solids and not manifest biased results. For an automatic sampling
device, this means that port intake velocities must be greater than the mainstream velocity and
should be placed at multiple levels in order to avoid stratification and capture both the lighter SS in
the top of the water column and the heavier particles near the channel invert.
The importance of the in-field settling test is to minimize any changes in settling properties
of SS due to storage, transport, and any other processes. In a comparison of two tests, Dalrymple
et al. (1975) showed that two distinct Long column tests had different results on two consecutive
days, even though both were run on the same sample. The difference in the test was attributed to
the storage of the sample for 24 h prior to the second test. This difference in stored samples was also
confirmed by CERGRENE (Aiguier et al., 1995) when the settling velocity of four samples, all
collected at the same time, yielded four different settling-velocity distributions. A fresh sample had
the settling test performed in the field and was compared to the settling velocity of three samples
stored for 24 h at different temperatures (room, refrigerated, and frozen).
Field Site
In identifying field sites for Phase III, the UWMB and John Meunier, Inc. looked for
municipalities ready to share technical information regarding location and configuration of
combined sewers and overflow sites. The municipalities needed to supply information on drainage
area (preferably residential to minimize influences due to industrial sources), number and volume
1 -6
-------
of overflows per year, SS concentration of overflows, and frequency distribution of overflow events.
Ideally, candidate sites would not yet have identified or installed treatment options for their CSOs.
Additionally, the municipalities would have to be willing to permit the project team access to their
facilities during CSO events in the summer of 1999 and to publish results based on data collected.
The City of Perth Amboy, NJ operates a combined sewer system and wastewater transfer
pumping station that collects combined sanitary sewage, industrial wastewater, and storm runoff
from an approximately 7 km2 drainage area and pumps it to a regional WWTP owned and operated
by the Middlesex County Utility Authority. The wastewater transfer pumping facility is located at
the junction of Water Street and Sadowski Parkway in Perth Amboy. A CSO regulator is located
about 6 m (20 ft) below the Sadowski Parkway with an overflow weir and a 2-m (7-ft) diameter CSO
tide-gated outfall to the mouth of the Raritan River.
The pumping station inflow from the interceptor passed through mechanical coarse bar
screens which removed large debris and into one of two wet-wells prior to being pumped to the
regional WWTP. Samples were collected from the wet well, since the inflow is a part of the CSO
and has the same characteristics at the outfall point during storm events. The wet-well is
approximately 9 m (30 ft) deep.
Field Sampling Review
As a background for this proj ect, John Meunier Inc. reviewed literature and wrote an internal
report (Champigny et al., 1997) on state-of-the-art field-sampling practices. This field of expertise
is often overlooked in studies and generally considered as a secondary subject. It was a weak point
in many recent characterization studies. The obj ective of this assessment review was to evaluate the
importance of the variability of solids found in sewer systems and to identify the most reliable
method to obtain representative samples from a combined sewer. While many of the methods
analyzed in the assessment were not developed for the study of WWF, the following general
conclusions and recommendations from the complete internal report were made:
• In dry weather conditions, the vertical concentration gradient of SS can be related to the flow
velocity pattern in the pipe or channel.
• A first flush phenomenon where the concentration of SS is higher closer to the beginning of
the storm has been observed by some researchers.
• Sediments found at the bottom of the channels interact with the SS and should be included
in the sampling.
Two separate sampling systems were recommended by this report for insewer design:
1. Sampling a complete cross section of the flow from bottom to top, and
2. Placing sampling port intakes at two points.
This second method would mount one sampling point just above the level of the DWF, near the pipe
walls. The second sampling point would be maintained at 60% of the total water level throughout
a WWF event.
1 -7
-------
2. MATERIALS AND METHODS
The design and fabrication of the columns and Phase I evaluations were conducted by
UWMB in the EPA UWMB facility in Edison, NJ, and by John Meunier in the offices of John
Meunier, Inc. in Montreal, Quebec, Canada. In Phase II, a set of 15 experiments was conducted
by the same parties from June 9, 1998 through June 17, 1998. Additional testing was conducted
on the CERGRENE columns at the EPA UWMB laboratory in November of 1999. In Phase III,
three combined sewage field events were performed by UWMB in Perth Amboy, NJ, from
March till May 2000.
Conclusion of Phase I: Preliminary Testing
The preliminary results (Phase I) determined the expected homogeneity of the mixing
basins and the initial performance of the respective columns. Phase I testing showed that
adequate mixing was provided in the mixing basins, that SS were transferred to the settling
columns for further testing, and that the microsand and Neshaminy clay particles were
recoverable in the columns. Other materials were tested, e.g., glass beads, but were found to be
unsuitable. The Long column had insufficient head to sample rapidly from the top two ports,
and drawdown would have eliminated these ports as testing progressed. Only ports 3, 5, 7, and 8
were used during the evaluation for determination of settling methods. At the conclusion of
Phase I testing, analysis had been performed independently by the EPA and John Meunier, Inc.
Column Description and Delivery
Long Column
An accepted but non-standardized settling-velocity-distribution determination method in
the United States employs a 1.8- to 2.5-m (6- to 8-ft) column to study settling characteristics of
solids in wastewater. The EPA designed a 2.5-m (8-ft) modular column (Figure 2-1) consisting
of four separate modules: a base section, a 1.2-m (4-ft) section (which must be attached to the
base), and two 0.6-m (2-ft) sections. The column was constructed from cast acrylic tubing with a
203-mm (8-in.) outer diameter, a wall of 6.5 mm (0.25 in.), and a 190-mm (7.5-in.) inner
diameter. The volume of the column was approximately 70 L (18 gal). The modules were
attached by acrylic flanges with foam gaskets to eliminate leakage. A cap was also available to
prevent foreign material from entering the top of the column, in the case of outdoor sampling .
Sampling ports (125 mm [0.5 in] NPT thread) equipped with quick-disconnect fittings
with flow size diameters of 9.5 mm (0.375 in.) were spaced vertically 30 cm (1 ft) on either side
of the column for a total of 16 ports at eight depths. Sampling from both sides of the column
was meant to yield a more representative sample of the contents, minimizing "wall effects" and
increasing the sampling area. It was originally thought to be more important to sample from
both sides toward the bottom of the Long column where withdrawal velocities were greatest than
at the top where withdrawal velocities were slower due to decreased head. The base section
included a 1-in. diameter drain which was connected to a three-way valve. This valve was used
for filling and draining the column. A conical plastic piece (funnel) was installed above the
drain inside the column to facilitate resuspension of solids during the filling process and aid
clean-out between experiments. A wooden baffle screwed into the cone dispersed the flow and
kept the influent well-mixed.
Filling was accomplished by pumping from a mixing basin through the bottom valve.
Before filling, the pump was primed and the bottom valve turned to "fill." Prior to and during
filling, the mixing basin (described in Mixing Basin) stirred the solids to keep them suspended.
After filling, the pump was turned off, the bottom valve was turned to the middle position, and
sampling from the side ports began.
2- 1
-------
Column Configuration
Sampling Ports
(every 1')
Portl
Sample ..-?
Collection
(Quick Disconnect Fittings.
Y-Connector, Clamp)
Fill
Figure 2-1. Long Column General Arrangement
Sampling tubes were attached to the male quick-disconnect fittings at each sampling port.
For two-sided sampling, two tubes that were attached to the male quick-disconnect fittings lead
to a T-connector which was attached to a short tube. This short output tube brought the sampling
streams together. Sampling from the end of this output tube was controlled by attaching another
set of quick-disconnect couplings.
Sampling was initiated at the upper ports of the column and progressed downward for
each time interval. Sampling tubes were purged before sampling. The output tube was placed in
the plastic bottle marked "Purge," and a male pipe adapter was attached to an elbow hose barb.
The "Purge" bottle was filled to a measured marking (at a minimum), and then in midstream the
tube was quickly moved to a plastic storage bottle. Each plastic storage bottle was marked with
an individual identification number which was recorded along with the corresponding port (1,
2....8) and sample time. After sufficiently filling the plastic storage bottle, the male quick-
disconnect adapter was removed from the output tube. Storage bottle size was nominally 250
ml, which appeared to match the 10 to 20 mg target mass range for SS analysis for the media in
Phase I. Cold storage was only required for Phase III samples, as all other samples contained
inert material (e.g., sand and clay). Combined sewer samples taken during Phase III were stored
in coolers with ice in the field and during transport, and in refrigerators back in the laboratory
2-2
-------
according to Standard Methods (1995).
CERGRENE Columns
The CERGRENE columns were designed following a series of tests. CERGRENE,
Centre d'Expertise en Gestion des Eaux d'Orages (CEGEO, a subsidiary of John Meunier, Inc.),
and the University of Aston undertook a study to optimize settling-velocity- distribution
measurements. The objectives of the European study were to:
• Compare the results of existing methods and protocols on identical samples;
• Compare and contrast the advantages of each method; and
• Understand the influence of each parameter (settling height, column diameter,
concentration of SS, temperature, etc.) on the settling-velocity distribution.
In light of the European study results, a new column test was proposed to meet the
following criteria:
• The sample should not to be pretreated;
• The sample should remain in its original matrix (water and SS) for tests;
• A sufficient sample size should be collected for analyses;
• The column should be easy to use; and
• The column should be compact for in-situ measurements.
The resulting settling test apparatus was the CERGRENE columns. These were tested
with a prototype and then in full scale. John Meunier, Inc. constructed four replicas of the
CERGRENE column based on equipment available in North America. The column had a 65-mm
(2.5-in.) inner diameter constructed of clear PVC and stood approximately 1 m (3 ft) tall. The
volume of this column was approximately 2.5 L (0.66 gal). The column (Figure 2-2) had three
valves located at the top, bottom, and middle. The middle valve, a 65-mm (2.5-in.) inner
diameter ball valve, was approximately 2/5 of the length from the bottom and divided the
column into two sections.
to vacuum
valve-
clear PVC
65 mm
inner diameter
ball valve
t
1 m
valve
from mixing basin
Not to Scale
Figure 2-2. CERGRENE® Column Configuration
2-
-------
The column was filled by vacuum aspiration to minimize the variation of SS
concentration between the bottom and top sections. The vacuum was applied to the top valve of
the column and drew wastewater up into the column through the bottom valve. A vacuum pump
doesn't break SS apart as readily as a typical positive displacement water pump which tends to
grind and shear particles, thereby changing the settling characteristics. After filling, the top and
bottom valves were closed. At the specified sampling time, the two sections were separated by
turning the middle ball valve. The bottom portion was the sample. This volume of
approximately 1 L (0.26 gal) was drained and SS analysis was performed. As previously
discussed, the major premise for performing this comparison was to determine whether the
CERGRENE column provided truer representation of high-settling velocity SS at the test starting
time, to, than the Long column.
Each CERGRENE column was sampled independently at discrete times and represented
one sample for settling-velocity-distribution analysis; this differed from the Long column which
required multiple samples to be taken per time interval. Filling of the columns was conducted
sequentially. Results from Chebbo et al. (1995) indicated that sequential filling of the columns
did not significantly impact results.
Mixing Basins
Two mixing basins were used in the course of the experiments to provide a well mixed
matrix, i.e., CSO or water and testing media, to supply both column testing systems. The John
Meunier mixing basin design was used for Phases I and II. The mixing basin was 0.66 m (2 ft)
in diameter, 1.3 m (4 ft) high, and had a 300-L (80-gal) holding volume. Four vertical baffles
were inserted at 90-degree intervals to prevent the formation of a vortex (Dickey and Hemrajani,
1992; Etchells et al., 1992). The mixer was mounted on a sawhorse above the basin. The mixer
shaft was in the middle of the basin and two impellers were used; a marine impeller at the
bottom and a Rushton impeller above.
A marine impeller with three blades was placed about 1 cm (0.4 in.) from the bottom of
the basin to create an axial flow in the basin, provide complete mixing of fluids, and suspend
particles that may settle naturally (Dickey and Hemrajani, 1992; Etchells et al., 1992). The
mixer manufacturer (Greey Lightnin) recommended a 25.4-cm (10-in.) diameter impeller based
on the existing basin configuration. The Greey Lightnin mixer, model XJ-43 with 1/3 hp of
power, has a constant mixing velocity of 350 rpm.
The use of a Rushton impeller was based on previous mixing studies done by John
Meunier, Inc. (Gagne and Bordeleau, 1996) and was also verified by the CERGRENE group
(Chadirat et al., 1997). The dimensions of the Rushton impeller (four equally spaced 60-mm
wide by 90-mm tall paddles) were also linked to the physical dimensions of the basin. The
Rushton impeller creates an axial flow that keeps the particles suspended by the marine impeller
and well mixed throughout the basin. During Phase I, the Rushton impeller was tested at 20 cm
(8 in.) above the marine impeller.
CERGRENE showed that mixing velocities between 200 and 600 rpm were adequate to
generate complete mixing in the basin and achieve consistent results (Chadirat et al., 1997). The
mixer was turned on 15 min prior to sampling and stayed on throughout.
During the Phase II experiments, the mixing basin was filled to an initial SS
concentration of 300 mg/1. On the sixth side-by-side experiment, the volume in the mixing basin
was changed from 200 to 250 L and the initial mass of 60 g of media was increased to 75 g to
maintain the known concentration at 300 mg/1. This extra volume prevented the water level
from falling too low which caused excessive vibration in the mixer after filling the Long column.
2-4
-------
The Long column and CERGRENE system were filled from the mixing basin in an alternating
sequence.
The point where water was withdrawn from the mixing basin for Phase II was 7 to 10 cm
(3 to 4 in.) from the wall, set halfway between two baffles and at an alternating height of 18 to
35 cm (7 to 14 in.) from the bottom, which maintained a volume of water in the basin throughout
the experiment. The alternating sequence for filling and height of withdrawal was intended to
reduce bias in the experiment as described in Experimental Design. The four wooden baffles
deteriorated slightly during the tests and may have had a minor impact on the loadings.
The EPA also developed a mixing system that was used for Phase I, the additional
laboratory testing of the CEGRENE columns after Phase II, and for the field experiments of both
columns in Phase III. The EPA mixing system used a 208-L (55-gal) tank with a 2.54-cm (1-
in.) NPT fitting in the center of a conical bottom. The tank, which can be drained from the
bottom, stood 94 cm (37 in.) tall and is 66 cm (26 in.) wide with a 6-mm (0.25-in.) wall. The
mixer was a Chemineer, model 2JTC 350 RPM, and had a 0.25 hp motor with a constant mixing
velocity of 350 rpm. There were two 20.3-cm (8-in.) diameter marine impellers, one at the
bottom and the other 25 cm (10 in.) above, attached to the 76-cm (30-in.) long and 1-cm (0.375-
in.) diameter shaft. Initial testing showed that baffles were necessary to eliminate a vortex.
Three vertical baffles made of polypropylene were attached to the basin wall at 120-degree
intervals. Only three baffles were chosen instead of four, as the EPA mixer shaft was inserted
into the basin on an angle instead of straight down as in the John Meunier, Inc. design. It was
felt that more room was needed to place the bottom marine impeller between baffles due to the
tilt. The EPA baffles were made of the same material as the wall of the basin and were attached
with metal blocks and U-bolts as mounts. A PVC three-way ball valve was attached to the
bottom effluent; this valve could be switched to deliver the sample volume to the Long column
or return flow to the top of the tank for recycle, which provided additional mixing. During Phase
III, withdrawal was made from the top 5 to 10 cm of the mixing basin for the CERGRENE
column, while the Long column was filled from the bottom drain. The CERGRENE columns,
except for the 2-h sample, were filled first as there was not adequate volume to keep the mixer
on after filling the Long column.
The John Meunier mixing configuration used in Phase II was more thoroughly tested, and
indications were that sampling closer to the surface of the water was better due to the power of
the mixer and mixer configuration. While configuration of the EPA column was different, the
results of samples taken directly from the mixing basin during Phase III were consistent with a
well mixed system.
Sampling
Long Column
The Long column was sampled at only four of the potential eight port depths in the actual
side-by-side comparison to increase the number of samples that could be obtained at different
times and depths in the first few minutes of sampling. Samples taken from the top of the column
required more time, i.e., 7 s at Port 1 in Figure 2-1 when the column was full, while samples
taken from the bottom six ports had sufficient head to ensure a shorter sampling time period, i.e.,
< 3 s at Port 3 and below. Also, if all depths were sampled there would not have been enough
sample for the top ports for the later times. An analysis of discharge velocities from the ports
showed good correspondence to theoretical values.
The fill time of the Long column ranged from 69 to 76 s for a height of 0.24 m (95 in.).
On average, the upflow velocity of water in the 10 out of 15 experiments where both height and
2-5
-------
time were measured was 3.3 cm/s.
For each set of Long column samples (one set = samples from ports 3, 5, 7, and 8),
average decrease in water height in the column was 4.5 cm (1.8 in.), averaging 1.1 cm (0.45 in.)
per sample. These average values were used to adjust the depth of the Long column in the
calculation of overflow rates. The decrease in water level height for one set includes both purge
and sample volume.
CERGRENE Columns
In Phase II, the vacuum pump filled the CERGRENE columns in an average of 8.2 s
where time was measured from the appearance of water in the column until shut off. Using a
standard height value of 0.91 m (3 ft) for the columns, the estimated velocity of water pumped
upwards into the columns was approximately 11 cm/s. The slowest filling time was 11.5 s which
was most likely a function of a cold start of the vacuum pump. Even with a worse case sample
height of 0.76 m (2.5 ft), the upflow velocity in the CERGRENE would still exceed 6.6 cm/s.
The filling velocity in the CERGRENE column was at least twice as fast as the Long column.
During Phase II, the CERGRENE columns were never actually filled to capacity,
always being a little short from the top. This height of the sample level in the CERGRENE
columns was measured for calculation of the settling-velocity distribution. Problems with the
sample height measurement of the upper chamber of the CERGRENE columns are discussed in
Chapter 3, and specific recommendations and modifications are presented in Chapter 4.
For the additional sampling and Phase III experiments, a more powerful vaccuum pump
was used which significantly reduced the filling time. The average time to fill the column was
4.9 s with a standard deviation of 0.3, and ranged from 4.5 to 5.5 s for five measurements. As
the CERGENE columns were overflowed by approximately 0.5 L, time was measured from the
first appearance of water in the column to its arrival at the top. The average velocity was 20
cm/s.
The intent for overflowing the CERGRENE column was two-fold. Primarily,
overflowing eliminated the height measurement of the upper half of the CERGRENE columns
needed for analysis. Completely filling the columns simplified the analysis procedure; problems
with the filling procedure during Phase II are detailed in Chapter 3. Also, overflowing should
have minimized start-up velocity flow impacts and hopefully allowed the system to achieve a
steady state velocity, which might lead to a better equilibrium in concentration between the top
and bottom components of the column.
Quality Assurance Samples
During each experiment, a set of triplicate samples was taken to show that the
"background" SS concentration was uniform. During Phase II, these samples were taken using
the pump for the Long column and were dubbed "Recycle" samples. In phase III, triplicate
hand-grab background samples were taken from the mixing basin. The background samples
served as the t0 concentration for the Long column and were a basis of comparison to the
concentration of the t0 CERGRENE column. Three samples were also collected to perform
gravimetric settleable solids analysis, discussed in Solids Analyses below. Other QA samples
included laboratory and field blanks.
Solids Analyses
Solids analyses were the critical measurements of these experiments. Table 2.1 presents
the summary of Standard Methods used. The selection of analytical methods was based on the
following priorities:
2-6
-------
1. Standard Methods, 19th Edition
2. EPA Method
2-7
-------
Table 2.1 Summary of Standard Methods and Procedures*
Parameter
Suspended Solids (SS)
Settleable Solids
Volatile Solids (VSS)
Total Solids (TS)
Sample
Type
Water
Water
Water
Water
Method
No.
2540 D
2540 F
2540 E
2540 B
Method Title
Total SS Dried at 103 to 105 'C
Settleable Solids
Fixed and Volatile Solids Ignited at 550 'C
Total Solids Dried at 103 to 105 'C
* All methods used from Standard Methods 19m Edition (1995).
The upper limit for SS sample sizes is 200 mg of residue. The lower limit for SS is 4 mg/1
as specified by EPA method 160.2. Suspended solids were calculated by the following
procedure:
mg suspended solids / L =
(A- B)x 1000
sample volume (mL)
(2-1)
where:
A = weight (mg) of filter and dried residue, and
B = weight (mg) of filter, mg.
The grain size of the particles for Phases I and II and the additional laboratory
experiments was larger than the filter paper pore size (1.5 • m, Whatman 934-AH) of the filters
being used for the SS analysis. Filters with a 47-mm diameter were used.
To perform settleable solids in Phase II, a graduated cylinder was filled from the pump
used for the Long column. After 1 h, a 250-mL sample was siphoned from the approximate
center of the graduated cylinder and was analyzed for SS. These are the non-settleable solids.
This concentration was subtracted from the initial SS concentration as derived from separate
samples. As settleable solids are an extension of SS analysis, the same limits apply.
Settleable solids were calculated by the following method:
mg settleable solids / L = mg suspended solids / L - mg nonsettleable solids / L
(2-2)
Volatile solids were measured for Experiments 1 and 2 of Phase III. The equations for
this method are:
mg volatile solids / L =
mg fixed solids / L =
(A- B) x 1000
sample volume (mL)
(B- C) x 1000
sample volume (mL)
where: A = weight (mg) of residue, filter, and dish before ignition,
B = weight (mg) of residue and dish after ignition, and
C = weight (mg) of dish and filter.
(2-3)
(2-4)
2-8
-------
Total solids were measured for Experiments 2 and 3 of Phase III. The equations for this
method are:
(A- B) x 1000 n ,.
mg total solids / L = (z~-))
sample volume (mL)
where: A = weight (mg) of dried residue and dish, mg, and
B = weight (mg) of dish, mg.
The CERGRENE developers originally envisioned multiple types of analysis from one
CERGRENE sample. Ideally, for the purposes of these experiments, the whole sample from
either column method was to be evaluated as one sample using SS analysis to measure the
settling rates. However, during Phase II the CERGRENE column samples held more SS than
could be efficiently collected on a single filter, and all samples of CSO from Perth Amboy
clogged the filters before the whole sample could be filtered for SS as one sample. Therefore,
only a portion of the sample was actually analyzed. While 70-mm diameter filters are
recommended by Standard Methods for the raw wastewater samples, at the time of the
experiments filter housings were only available for 47-mm filters. In lieu of splitting each sample
and performing several SS analyses in Phase III, only a portion of the collected samples was
used for SS analyses in Experiments 1 and 2. Total solids analysis was performed on the
remaining portion of the sample of Experiment 2, and the whole collected sample was analyzed
for TS for Experiment 3. This should not affect the outcome of the experiment, as obtaining an
accurate measurement of the solids concentration in the sample, especially the settleable portion,
was what was most important.
Identification of Experimental Materials for Phase II
Initial experiments were performed using microsand, Foullon's Earth, glass beads, and
natural soils. The media were mixed in tap water prior to introduction into the mixing basin.
Microsand was chosen to test the mixing basin and for the laboratory experiments. The initial
characteristics of the microsand used for the design of the experiments were:
d10 = 85 • m
d60 = 125 • m
• •= 2.62 g/cm3
The calculated settling velocities for these particles using Stoke's Law (ideal sphere assumed)
were <1 cm/s, which is less than the upflow velocities measured in the columns. An analysis of
the microsand by a Coulter® LS Particle Analyzer determined the following particle diameters:
d10= 156.2-m
d50 = 232.8-m
which translates to less than 10% of the microsand having calculated settling velocities of less
than 1.36 cm/s and the microsand having an average settling velocity of 2.15 cm/s. This analysis
was performed in January, 1999 after Phase II was complete and the Coulter Counter first
became available for use at UWMB. Appendix J shows the Coulter Counter report.
A surficial soil excavated near Princeton, NJ, was used as an additional reference. The
soil (Neshaminy) is a silty clay loam, containing 17% sand, 46% silt, and 37% clay. Although
2-9
-------
this soil has a greater proportion of very small particles, it also has a wider particle size variation
with d10 = 75 • m and d60 = 700 • m (Fischer, 1995).
Experimental Design
Table 2.2 shows the original experimental design of the Phase II side-by-side
experiments which were designed to account for bias due to the order of filling the columns and
variations in the height of water in the mixing basin. One duplicate was performed for each
medium (microsand, Neshaminy, and mixture) for a total of 15 experiments. The experiment
order was randomized to reduce bias due to one soil type being repeated or increased precision
from experience.
Table 2.3 provides a prototype sampling strategy for microsand. A set of samples from
the Long column were withdrawn at specified intervals (e.g., 1, 3, 5, or 10 min) with time
measurements to the nearest tenth second for the individual sample end-times. Four samples
comprised one set of samples for the Long column. For each CERGRENE column, the valve
was turned to capture SS at a time corresponding to a time interval for the Long column. The
initial time for the individual CERGRENE column was designated to be t0; where i represents
the number of the column. The time for the CERGRENE column was the time from the moment
each column was filled to the time the sample was isolated by turning the valve of each column.
The Long column t0 was the time that filling was completed. Initial concentration in the entire
column was assumed to be a background or "recycled" concentration taken from the mixing
basin by the pump, as explained in Chapter 3.
Table 2.2 Experimental Design for Phase II Laboratory Experiments
Media
Microsand
Microsand
Microsand
Microsand
Microsand
Clay soil
Clay soil
Clay soil
Clay soil
Clay soil
Mixture
Mixture
Mixture
Mixture
Mixture
Experiment #
1
2
3
4
Duplicate - 13
5
6
7
8
Duplicate - 14
9
10
11
12
Duplicate - 15
Randomized
Experiment #
Order
8
15
4
13
12
5
10
14
1
6
9
11
3
2
7
Withdrawal
Height (cm)
18
36
18
36
18
18
36
18
36
36
18
36
18
36
18
Column
Filling Order
Long /CERGRENE
Long /CERGRENE
CERGRENE / Long
CERGRENE / Long
Long / CERGRENE
Long / CERGRENE
Long / CERGRENE
CERGRENE / Long
CERGRENE /Long
Long / CERGRENE
Long / CERGRENE
Long / CERGRENE
CERGRENE /Long
CERGRENE /Long
CERGRENE / Long
End Time
(min)
10
10
10
10
10
60
60
60
60
60
60
60
60
60
60
2- 10
-------
Table 2.3 Critical Time Measurements for a Microsand Experiment, Phase II
CERGRENE Column
Initial Time
V
to, 2
^ 3
to, 4
to, 5
Duplicate - 10 x
Time
t0,i
1 min - 10 2
3 min - 10 3
5 min - 10 4
10 min-t0 5
x min - 10 x
Long Column
t0 - initial time
< 1 min
<2min
• 3 min
• *> min
• 10 min
Port3
ti-to
ts-to
t9-to
tl3 - to
tl7 - to
PortS
t2-t0
t6-to
tio " to
tl4 - to
tl8 " to
Port?
ts-to
tyto
tn - to
tis - to
t19 - t0
PortS
t4-t0
tg-to
tl2 - to
tie " to
t2o ~ to
Cannot perform duplicate on Long column
A similar time measurement scheme could be constructed for the Neshaminy soil and
other particle mixtures with the time of the samples extended out to 1 h. The initial estimated
settling rate for the Microsand with a d10 of 80 • m was < 15 min for the Long and < 10 min for
the CERGRENE columns. The settling times were actually much less than that, given the larger
average diameter of the particles.
Figure 2-3 shows the laboratory setup. Since only four CERGRENE columns were
available for the experimental sequence of six or seven separate time measurements for each
experiment, the CERGRENE columns were sampled, rinsed out, and refilled from the mixing
basin during Phase II. This procedure was used for the first nine experiments until it was
discovered that the individual CERGRENE columns were each producing unexpectedly random
settling results. Only one CERGRENE column was used for the remaining six experiments of
Phase II. The column was rinsed out after each time interval, before the column was refilled and
another sample was collected. Phase III also used only one CERGRENE column and a slightly
modified setup, which is presented in Chapter 5.
2- 11
-------
Zl
Zl
" 3
Zl
•
=1 5
7
8
Monarch Type A Pump
Rushton impeller
Marine impeller
John Meunier Mixer
Not to Scale
Vacuum Pump
36 cm
18 cm
tt J
1234
Figure 2-3. Configuration for Side-by-Side Phase II Experiments
2- 12
-------
3. RESULTS OF PHASE II
All graphical figures for this and the following sections are presented in the appendices
(e.g., Figures A-l - A-5 are in Appendix A, Figures B-l - B-32 are in Appendix B).
Quality Control Analysis
Several levels of experimental procedure were incorporated to ensure quality control in
all phases of the project. Those specific to the Phase II analysis are described below.
Table 3.1 Estimated Number of SS Analysis for One Experiment of Phase II
Sample Location
CERGRENE columns
Long column - various ports
Recycle - QC samples on mixing basin
3 settleable solids (for Clay soils and mixture only)
Duplicate CERGRENE column
Laboratory QC - Standard Reference Material
Blank - Tap Water in Mixing Basin
Blank - distilled water
Number of SS samples
5
15-25
3
6
1
3
1
1
Blanks and Standards
Method blanks were run either on de-ionized, tap, or basin water before the delivery of
the media. The purpose of blanks was to ensure SS were not introduced during sample
collection or analyses. All blanks had concentrations below 4 mg/1 (Figure A-l). Experiment 3
did not have a corresponding blank while experiment 9 had two blanks.
The accuracy of the SS procedure was determined from the analysis of laboratory control
samples with known concentration of SS. Standard reference materials (SRM) were the same
materials used in the experiments, namely, microsand, Neshaminy silty clay loam, and a 50-50
mix of the sand and Neshaminy.
Table 3.2 QA Objectives for Measurements
Measurement
Suspended Solids
Settleable Solids
Method
2540D
2540F
Reporting
Unit
mg/1
mg/1
Initial
Concentration*
15
242
1707
NA
Standard
Deviation*
5.2
24
13
NA
Relative Standard
Deviation*
33%
10%
0.76%
NA
Complete-
ness
90%
90%
90%
NA
From Standard Methods (1995)
3 - 1
-------
Accuracy is expressed as percent recovery. The formula used to calculate this laboratory
QC values for a SRM is:
%R=100%xCm/Csim (3-1)
where: %R = percent recovery
Cm = measured concentration of SRM
Csrm = actual concentration of SRM
Excluding one extreme, which was most likely a mislabeled sample or an error in transcribing,
the average percent recovery was 80% with media results of 66%, 83% and 88% for sand,
mixture and clay, respectively.
Figures A-2 through A-5 show graphs of the measured standards versus the expected
percent recovery (the log linear line in all the graphs) for diatomaceous silica as developed from
the values for relative standard deviation in Table 3.1. To plot a comparison versus the relative
standard deviation values in Table 3.1, %R was adjusted by subtracting from 100% to get the
relative percent difference (RPD) values.
Figure A-2 shows RPD for all the media and Figures A-3 through A-5 are media specific.
The Neshaminy (Figure A-5) exhibits the best RPD, equivalent to the recoveries of
diatomaceous silica. The mixture (Figure A-4) had several RPD within acceptable limits but for
most cases exceeded the limits of diatomaceous silica by up to 45%. The microsand (Figure A-
3) exceeded the expected RPD for diatomaceous silica in all but one case.
The microsand was the most difficult media to work with when performing SS, as
particles tended to stick to the surfaces of analytical equipment due to water tension. The
analysis of microsand produced the most pronounced losses. Neshaminy soil was much easier to
analyze (except when the filters began to clog) as the results indicate. Except for the one
extreme outlier of-189% for a mixture sample (A-4) which represented a gain in mass, all SRM
analysis indicated a loss in SS.
Another factor that may have contributed to this error in the SRM was the smaller sample
sizes, 125 mL bottles. All Long column samples were 250 to 300 mL and CERGRENE samples
were 250 mL aliquots for Neshaminy and 960 mL for the microsand and the mixture.
Completeness
The completeness is defined for this study as the ratio of the number of valid
measurements to the total number of measurements planned for each parameter. A
completeness objective of 90% is expected to ensure that enough valid data points are collected
to evaluate the settling velocity distributions. Samples were assumed valid unless voided due to
the following QA procedural problems: samples with no identification, duplicate identification,
missing data (which could not otherwise be discerned), torn filters, incomplete solids transfer to
filters or spills. Table 3.2 shows the completeness for critical SS measurements made for all 15
experiments. The formula used to determine completeness is:
%C = 100%xV/T (3-2)
where % C = percent completeness
V = number of measurements judged valid
T = total number of measurements
3 -2
-------
The Recycle and the Settleable Solids samples were the only sample types that did not
achieve the completeness criteria of 90% in Table 3.2. However, they approach 90 %, and are
within one sample of passing this criterion.
As noted in Table 3.3, only 17 of the 340 Long column samples were voided for faulty
procedures (e.g., sample volume not noted). Additionally, three microsand samples fell below
the 4 mg/1 limit of detection for the EPA SS method, but these samples do appear in the graphs
as approaching zero concentration, or 100% removal.
Table 3.3 Completeness of Suspended Solids Analysis
Type or location
Blanks
CERGRENE
Long
Recycle
Settleable Solids
Total Samples
17
101
340
45
16
Voided Samples
0
0
17
5
2
Expected
Completeness
90%
90%
90%
90%
90%
Measured
Completeness
100%
100%
95%
88.9%
87.5%
Of the 101 separate CERGRENE samples, none of the samples were voided. However,
in lieu of incomplete information, when omissions in measurement could not be deduced,
standard values were assumed. These values had to do with measured volumes of the
CERGRENE column. The bottom of the column was assumed to be 960 mL, which did not vary
noticeably when measured directly, and in fact should not have varied at all, as the bottom
portion was filled to capacity. After the first few runs, it was decided more error was introduced
by measuring than assuming the 960 mL value. The default bottom volume of 960 mL was used
in calculating the upper chamber volume and percent removals, but when available the measured
volume was used for actual SS concentration of the bottom effluent.
Due to faulty recording procedures, height measurements, which represents the volume
of the top chamber in the CERGRENE columns, were not recorded for several individual
CERGRENE runs. This measurement was not critical at the time of the laboratory analysis but
some indication of height is essential during settling velocity analysis using CERGRENE's
iterative matrix program. Where no height measurement was available, a value of 41.9 cm (16.5
in.) was used. This value is the mode for the 86 out of 101 samples for which a height
measurement is available, with a mean of 41.4 cm (16.3 in.), a standard deviation of 1.7 cm (0.65
in.), and a coefficient of variance (CV) of 0.040. As the experiments proceeded, better control of
the level in the CERGRENE columns was exhibited. For experiments 10 through 15, the mode
was 41.9 cm (16.5 in.), the average was 42.1 cm (16.6 in.), with a tighter standard deviation of
1.3 cm (0.53 in.) and CV of 0.032 for fewer samples (31 of 40 samples). Measurements could
only be made to the nearest eighth-inch, which implies a virtually identical mean and mode.
The error introduced by the lack of height measurements is minor in comparison to errors
introduced from other sources. The major cause of variability in the SS analysis was the use of
the microsand itself. During sample preparation, sand could be observed clinging to the filter
housing, and was subsequently scraped onto the filter. Testing media sticking to the filter
housing was not as apparent with the mixture, as the Neshaminy clay soil clung to the sand and
reduced water tension of the sand to the filter housing.
3 -
-------
Neshaminy soil tended to clog the filters for the larger volume CERGRENE samples.
For this reason, 250 mL aliquots were obtained of Neshaminy CERGRENE samples from
beakers with magnetic stirrers, and these aliquots were analyzed for SS. Thus filters were not
overloaded for Neshaminy runs (where overloaded is defined by SM 2540 as exceeding 200 mg
of residue). The aliquot method could not be used for microsand or mixture runs, as sand
particles tended to be more discrete and a representative sample could not be obtained with
magnetic stirrers as with the Neshaminy which formed a more uniform mixture. The larger sand
and mixture CERGRENE samples exceeded 200 mg thus overloading the filters.
Mixing Basin - Recycle
Three background or "recycle" samples were obtained immediately prior to filling the
Long column. Recycle samples were an additional level of quality assurance. The known
concentration in the mixing basin was 300 mg/1, so recycle concentrations should have centered
around that number. The recycle concentration of each experiment was used as the t0 of the
Long column in lieu of averaging the initial measurements at each port in the Long column. The
recycle samples were taken from the same pump that filled the Long column, at the same side
wall height that the CERGRENE and Long columns were being filled. The recycle
concentration was thought to better represent the concentration delivered to the Long column.
The average recycle concentration for all 15 experiments was 272 which represented less than a
10% loss overall from the known concentration.
Figure A-6 shows all recycle concentrations for all experiments. The data appear to be
spread over a wide range of concentrations, but when viewed by soil type (Figure A-7), it
becomes apparent that Neshaminy soil, with its higher percentage of clay and silt, yields tighter
distributions around 300 mg/1, while the microsand tends to be much more widely distributed
and unpredictable. This is due in part to the nature of the sand particles, which are discrete and
dense, and may elude the sampling container or settle beneath the mixer. In addition, the SS
analysis was much more robust for Neshaminy than for sand. Thus the mixed soil, with its
combination of Neshaminy and sand, shows a distribution not quite as tight as Neshaminy, but
not as widely distributed as sand.
Figures A-8 and A-9 show recycle concentrations by order of filling and depth of
sampling, respectively. The recycle concentration distribution is closer to 300 mg/1 for the
experiments where the long column was filled first and for the experiments where the height
above the bottom of the basin was to 35 cm (14 in.) rather than 18 cm (7 in.). A second finding
is that filling concentrations are closer to the known concentration when the intake is closer to
the surface of the water in the basin. The first finding shows tighter distributions when the Long
column was filled first, which is probably a result of larger volumes of water providing better
mixing. When the CERGRENE columns were filled first, enough water was removed from the
basin to interfere with the mixing process. An analysis of variance showed that the order of
filling, media type and intake depth did not significantly impact recycle concentrations.
Settleable Solids
The gravimetric settleable solids analysis (SM 2540F) entails first performing a SS
analysis on a representative sample. During these experiments, settleable solids samples were
taken the same way the Recycle samples were taken.
Figure A-10 shows the non-settleable solids concentration for this method for Neshaminy
and mixture media. The settleable solids method was not performed on the microsand as
insignificant concentrations of non-settleable solids were expected. The concentrations of the
mixture medium is about half that of the Neshaminy, which is expected as the mixture contains
3 -4
-------
half the mass of Neshaminy soils. An interesting result, however, is the tighter distribution of
the mixture results, possibly due to a flocculant effect, where charged clay particles may cling to
microsand particles and settle more predictably.
Laboratory Experiments 1-15
Concentration versus Time
Figures B-l through B-30 show plots of the raw data for the Long and CERGRENE
columns for the 15 experiments. The first nine CERGRENE graphs are shown for completeness
only. As mentioned previously, the inconsistency between CERGRENE columns makes it
unadvisable to use the data from the first nine CERGRENE runs. Experiments 10 through 15,
where one CERGRENE column was used repeatedly, show a pattern of increasing concentration
in the graphs, as would be expected.
The graphs of the Long column results show the pattern of settling for particles of each
type. Note the rapid settling for the sand experiments, where concentrations at all ports quickly
tail off to near zero. Neshaminy experiments exhibit a more gradual settling pattern, with the SS
concentration of the higher ports decreasing gradually, and the lower ports less so, as they
receive the settled particles from the higher elevations. The mixed soil type shows an initial
rapid settling of the sand particles, followed by the more gradual Neshaminy pattern. This
becomes even more apparent in Figure B-31, which shows the results of the 15 experiments on
the Long column, averaged by soil type.
Long Column Shortcomings - Initial Concentration Gradient
An inherent problem in the design of the Long column is the lack of reliable uniformity
in initial concentration (C0). The height and volume of the column makes it difficult to deliver
the sample quickly enough to ensure minimal settling of solids during the delivery time. Thus,
depending on the density and particle size in the sample being delivered, a concentration
gradient appears in the time zero measurements. This is compounded by the fact that
simultaneous t0 measurements were impossible to achieve in the Long column by hand (three
peolple were perfoming the sampling). Because of the very nature of the sampling
methodology, a lag will develop between completion of sample delivery and initial
measurements, and between the port measurements themselves. A full minute may elapse
between end-of-delivery and first sampling at port 8.
Figure B-32 shows port-by-port (represented by height above the column bottom)
average concentrations for each soil type and each time interval. For a well mixed column, to
measurements should yield a straight line with zero slope and a y-intercept equal to the recycle
concentration. The slope of the t0 line indicates the severity of the gradient. While the sand
shows a severe lack of mixing, due to the size and density of the particles, the Neshaminy soil
shows more uniform concentrations and exhibits better mixing. This is because Neshaminy
contained clay particles which generally have lower specific gravities than pure sand. The
Neshaminy also took longer to settle with significant concentrations after one hour while the
sand had settled out within five minutes. The lack of reliable uniformity in initial concentration
of heavy material has long been recognized as a shortcoming of the Long column, and causes the
scatter that can be seen in the initial measurements of these experiments.
Concentrations, which were expected to decrease with time, were relatively flat
throughout the experiments, except for an initial dip. Due to slower settling rates, concentrations
for the lower Ports 5, 7 and 8 , especially 7 and 8, during the experiments using the Neshaminy
soil may have been demonstrating hindered zone settling (defined, in part, as a suspension of
intermediate concentration in which interparticle forces are sufficient to hinder the settling of
3 -5
-------
neighboring particles, from Tchobanoglous and Burton, 1991). This could be observed visually
over the extended sampling times (30 - 60 minutes) when a distinct supernatant developed in the
top of the Long column. The concentrations for the lower ports 7 and 8 even began to exceed
the known delivered concentration of 300 mg/1 after one hour as demonstrated in Figure B-31.
CERGRENE Shortcomings - Lack ofRepeatable Results
The precision of the measurements for SS concentration in the mixing basin can be
calculated from the analysis of triplicate samples. The precision of the CERGRENE columns
can be calculated from the duplication of a sample at a specified time.
Precision for duplicate analysis was estimated by calculation of the relative percent
difference using the following equation:
RPD = ((d - C2) x 100) / ((d + C2)/2) (3-3)
where RPD = relative percent difference
d = the larger of the two observed values
d = the smaller of the two observed values
When three or more replicates were available, the relative standard deviation (RSD),
instead of the RPD, was used as follows:
RSD = ('/y)xlOO% (3-4)
where: • • = standard deviation, and
y = mean replicate analysis.
The standard deviation is defined by:
(3-5)
where: • • = standard deviation
Yi = measured value of the rth replicate
y = mean of replicate measurements
n = number of replicates
For each run, a duplicate CERGRENE column was tested and three Recycles were taken
The differences in these values is presented in Table 3.4.
These data indicate that the variation of the recycle concentration for each experiment
was random and not media driven (microsand, Neshaminy and mixture). This variability is
driven partly by the limited amount of samples but may also be due in part to the size of the
samples taken (about 250 mL, a larger sample may have had lower variability between samples)
and the force with which the sample bottles were filled by the pump (the same pump used to fill
the Long column and the pump may have been too powerful for the sample bottle size).
The duplicate results for the CERGRENE column tell a different story. A duplicate
analysis was performed by filling a CERGRENE column at the same or duplicate time interval.
The duplicate served to measure the settling performance and QA for recovery from the
3 -6
-------
experimental apparatus. During the first nine experiments, four different CERGRENE columns
were used, and any CERGRENE column could have been used to perform the duplicate. The
same CERGRENE column was used for the remaining six experiments.
Table 3.4 Duplicate Analysis for Recycle Concentration and CERGRENE Columns
Experiment
#
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
QA
#
8
12
11
3
5
14
15
1
9
6
10
13
4
7
2
Withdrawal
Height
(cm)
36
36
18
18
18
36
18
18
18
36
36
18
36
18
36
Filling
Sequence
C/L
C/L
C/L
C/L
L/C
L/C
C/L
L/C
L/C
L/C
L/C
L/C
C/L
C/L
L/C
Media
Neshaminy
Mixture
Mixture
Microsand
Neshaminy
Neshaminy
Mixture
Microsand
Mixture
Neshaminy
Mixture
Microsand
Microsand
Neshaminy
Microsand
Recycle
RPD
(%)
--
26.1
--
--
--
2.6
--
--
--
--
--
--
17.8
4.9
--
RSD
(%)
0.53
14.7
24.2
6.2
11.8
13.3
9.1
5.6
8.8
5.3
9.6
CERGRENE Duplicate
Duplicate
Time (min)
NA
3
3
1
0
60
10
5
3
1
0
10
5
60
1
RPD
(%)
No Duplicate
28.6
18.6
7.2
23.8
9.8
35
5.9
8.2
0.56
0.25
22.2
12.6
1.1
2.1
The duplicate analyses of the CERGRENE columns for the first nine do not appear to
follow any discernable pattern, not dependent on what time sampling occured, which media is
measured or criteria of experimental design. The CERGRENE columns appear to behave
randomly or independently for the first nine experiments. The duplicate analyses improve
dramatically for the last six experiments when only one column is used as a distinct pattern can
be correlated from the media being tested. The Neshaminy and mixture appear to have better
duplicate analyses than the microsand. While the amount of data may not be large enough to
state this finding with statistical certainty, this observable result is expected. The percent
recovery analyses of SRM as discussed under Blanks and Standards indicated that microsand
had the largest variation while the Neshaminy had the least with the mixture somewhere in
between. That the difference in the CERGRENE columns duplicate analyses in the last six
experiments was better than the expected recovery from the SRM analyses may be due in part to
the larger sample size of the CERGRENE columns than was used in the SRM analyses.
The measured Recycle SS concentrations continued to behave randomly during the final
six experimnets. This indicates that the improved CERGRENE performance in the latter
experiments was not merely due to overall enhanced technique of the experimenters as the
experiments progressed. One column, instead of several columns, each with their own
idiosyncracies, produced repeatable results. This duplicate analysis only compared the SS
concentration and not settling rates.
3 -7
-------
Percent Removal Long Versus CERGRENE
Traditional methods of computing settling velocity distribution based on settling column
data rely on a simple depth per time relationship. In the Long column, the depth measurements
from each port are divided by time of the sample to calculate the settling velocity distribution or
the design overflow rate which can then be used in settling tank design. Though this computed
number is in units of length per time (cm/s), it is not equivalent to a discrete particle settling
velocity. The C0 in the Long column are presumed to be uniformly distributed at all depths;
however no direct measurement can be made to the length of the particles' flow path.
This design overflow rate was plotted versus the percent removal. For the Long column,
percent removal is defined as the SS concentration at the port compared to the average recycle
concentration for that run, which is the theoretical C0 at every port. Though the actual C0 at
each port was obviously not equal to the recycle concentration (see discussion on lack of well-
mixed conditions), this is a necessary assumption to construct a plausible settling curve and
resulted in practical results with little scatter for the slower settling solids.
For the CERGRENE column, the overflow rate is compared to percent removal for each
column of time greater than zero. Overflow rate is computed by dividing the distance an average
particle traveled by the time measurement of the column. Thus, for each CERGRENE column,
the number is computed by dividing one-half of the length of the upper portion of the water
column (from the middle ball valve to the top water level) by the column's time, be it 1, 3, 5, 10
or 60 minutes. The C0 in each column is assumed to be equal to the time zero column's bottom
portion concentration. Percent removal is defined to be each column's top portion concentration
(computed by comparing to the bottom portion concentration) divided by the column's assumed
C0 multiplied by the ratio of bottom volume over the top volume.
Figures C-l through C-9 show Long column results for experiments 1 through 9. The
CERGRENE results are not reliable and are therefore not shown. Figures C-10 through C-21
show Long and CERGRENE results for experiments 10 through 15. The shapes of settling
curves are comparable for the two methods; the immediately obvious differences are apparent in
the sand experiments (experiments 12, 13, and 15; Figures C-l4 — C-l7, C-20 and C-21,
respectively). The Long column may overstate overflow rates for the fast settlers, as the right
hand side of the Long column graphs should represents a larger spread of settling rates, and the
CERGRENE columns under estimates removal rates, as they should attain 100% on the left side
of the graphs. For the Long column, this is most easily explained due to the lack of adequate
initial mixing. The assumption of C0 being equivalent to average recycle concentration may
yield a false concentration for the rapid settlers. For the CERGRENE column, the very large
volumes of analytes made it difficult to do the SS analysis for sand. Theoretically, all the sand
should have settled in the first three minute which corresponds to a settling rate of 0.5 cm/sec for
a diameter of 85 • m, and the percent recovery at five and ten minute should have approached
100% . Losses of mass result in prediction of lower concentrations. Besides the problems with
the SS analysis already discussed in this section under Completeness, some microsand particles
may have been trapped in the ball valve mechanism.
Matrix Iteration Process for CERGRENE Columns
The CERGRENE group of France (Lucas-Aiguier et al., 1997) developed a software
application called VICTOR to use the data from the small columns to produce settling velocity
distributions. VICTOR utilizes an iterative method to solve simultaneous equations, resulting in
a matrix M(i,j) which contains mass removed for each particular time interval / and pollutant/
Based on this matrix, a distribution of settling velocities may be constructed. The resultant
3 -
-------
analysis is presented in spreadsheets and figures. A more complete description of the derivation
is in Appendix L.
The nature of the CERGRENE settling column method sometimes results in experiments
for which the matrix analysis cannot find a solution or a limited solution at best. Thus a graph of
M(ij) for a particular pollutant concentration, in our case SS, versus time, which should increase
monotonically, may have discrete points which exceed the following temporal point. Certain
data points may not follow this pattern, and this may cause the VICTOR spreadsheet application
to be unable to find a solution. The software allows the user to choose either analysis of all data
points, which includes all data in the computation of the velocity distribution, or analysis with
"suppression," which allows data points to be excluded. In the analyses performed for this
project, the choice of suppression or non-suppression was based upon whether or not data points
showed the expected increase in concentration versus time. Unless the points in question are
significantly skewed to one direction, it should not have a severe impact on the resultant velocity
distribution, and use of suppression can be minimized.
As discussed earlier, further analysis of CERGRENE experiments 1 through 9 is not
warranted due to the inconsistencies between columns. Figure D-l illustrates experiment 9,
where the column order was randomized. Suppression was not possible with this data, as no
pattern can be inferred from the mass removal graph. This experiment shows the limitations to
the VICTOR software in relationship to the data presented to it. The points at 200 and 300 s
(figure to left) exceed the points at later times 600 and 3600 s, which confounds the ability of the
software to breakdown the experiment into a meaningful velocity distribution profile (figure to
right). This may be contrasted with Figure D-2, which shows the results from experiment 10,
where one column was cleaned and reused for each time interval. Monotonically increasing
mass removal values are seen (left), as well as a more varied velocity distribution (right).
Figure D-3 shows the results of experiment 11, utilizing all points. Figure D-4 shows the
results after suppressing points 2 and 5, which fall outside of the expected pattern. Note the
differences in the velocity distributions, especially in the slowest reaches, where the 3600 s point
was suppressed. The basic shape of the distribution, however, remains the same.
Figure D-5 shows experiment 12, where suppression was not possible, because
suppression would leave only 2 points for analysis (a minimum of three is necessary). The shape
of the velocity distribution, however, is similar to that of experiment 13 (Figure D-6), both of
them being sand experiments. The third point in Figure D-6 could have been suppressed, but it
does not fall far outside the curve, and thus is probably more useful being left in the analysis.
Suppression of this point results in a sharper drop-off at the slower settling velocities.
Figures D-7 through D-10 illustrate experiments 14 and 15, with and without
suppression. Note that the shapes of the velocity distributions remain similar whether or not
points are suppressed.
VICTOR can prove a useful tool for computing a settling velocity distribution for the
CERGRENE columns, as long as the data increases monotonically. Otherwise VICTOR may
require data suppression, a euphemism for selectively choosing data points. In general
eliminating or throwing out data points may be viewed as a questionable practice and is not
advocated by the authors; it is presented here to complete demonstration of the method and the
matrix calculations. Care should be taken in trying to compare these velocity results to the
results from other types of analyses, which use different assumptions, computational techniques
and methods for deriving settling velocity distribution. Additionally, VICTOR potentially may
work better with higher numbers of samples and time intervals. The capability of the software to
3 -9
-------
track several pollutants could be very useful for partitioning tests, though this feature was not
examined here.
Eckenfelder Analysis for Long Column
Use of Eckenfelder plots proved to be an inappropriate analysis for the media in Phase II.
The Eckenfelder analysis is generally used to provide flocculant analysis. For the analysis to be
successful, iso-concentration lines need to be developed from plotting concentration values for
each sample depth (y coordinate) and time (x coordinate). The clay particles were not settling,
and in fact concentration increased at the lowest port for several experiments at the one hour
mark. This, as stated earlier, seemed typical of hindered zone settling. During the sand and
mixture experiments, the sand settled within a five minute time frame and exhibited properties of
discrete settling, not flocculant settling.
Design Removal Comparison
Table 3.4 shows calculated overflow rates versus percent removals for the Long and
CERGRENE columns for experiments 10 through 15. Calculations were identical to those used
to develop the graphs in Appendix C, except that outlying CERGRENE points were deleted if
the Victor algorithm suppressed the points automatically. Results from the Victor analysis were
wildly divergent from the calculated results in Table 3.4, possibly due to lack of sufficient data
points to effectively utilize the Victor tool. In fact, settling velocity results of Victor runs do not
even show a noticeable difference between media. Comparisons of Long to CERGRENE results
show some similarities, though CERGRENE analysis is complicated by the lack of sufficient
data points.
The calculated Stoke's Law settling velocities for ideally spherical sand at 15 'C ranged
from 0.5 cm/s for 85 • m diameter sand to 12.6 cm/s for 400 • m. At 230 • m, the d50 value as
calculated by the Coulter® LS Particle Analyzer for the microsand, the settling velocity is 9.3
cm/s.
A graphical comparison of percent removal of the microsand for the two column methods
is shown in Figure 3-1 and 3-2 (O'Connor et al, 1999). Concentration decreases over time in
the Long column, while in the CERGRENE columns the concentration accumulates because
samples are withdrawn from the bottom section. Figure 3-1 shows that the Long columns
predicted nearly 100% removal for the microsand experiments (experiments 4, 8, 12, 13 and 15
as listed in Table 3.3) which is expected, as the microsand should settle out in the first 3-5 min,
whereas the CERGRENE columns only predicted 50% removal with a wide variation. The
microsand as a medium to measure SS recovery contributed a significant amount of loss during
the SS analysis because of the physical properties of the microsand (e.g., stuck to wall of sample
bottles). This bias error was estimated by calculation to be a 20% loss over a range of
concentrations for standard reference samples. This expected loss of 20% was incorporated into
the percent removal values for the microsand. Figure 3-2 indicates that there was still a
significant amount of variation and loss for the CERGRENE column system which under
predicts the expected 100% removal for microsand after ten minutes.
3- 10
-------
Table 3.5 Measured Overflow Rate and Predicted Percent Removal for Long and CERGRENE
Columns
Experiment
(Media)
10
(Neshaminy)
11
(Mixture)
12
(Microsand)
13
(Microsand)
14
(Neshaminy)
15
(Microsand)
Percent
Removed (%)
30
50
70
30
50
70
30
50
70
30
50
70
30
50
70
30
50
70
Overflow Rate r2
LongT (cm/s)
0.069
0.0077 0.81
0.00086
4.4
0.14 0.61
0.0046
100
18 0.64
3.1
120
20. 0.65
3.4
0.014
0.00059 0.43
0.000024
220
30. 0.66
4.0
Overflow Rate
CERGRENE1 (cm/s)
0.072
0.010
0.0015
0.30
0.0014
0.0000069
100
1.27
0.016
1.8
0.076
0.0032
0.61
0.094
0.015
31000
30.
0.00028
,
(#pts)
0.96
(5)
0.99
(3)
0.22
(4)
0.92
(4)
0.82
(4)
0.90
(3)
T Based on average recycle concentration. Points with calculated %Removal<0 were deleted.
* Points removed by the Victor "suppression" algorithm were deleted, often yielding few points.
Number of points used in the analysis is shown with the correlation coefficient.
3- 11
-------
Percent Removal of Suspended Solids
110
100
90
80
70
60
50
40
30
?0
|=^=^=|
I
—
~r
— ' —
Long Cergrene
Percent Removed from Time Zero: Time =10min, SAND
T" Non-Outlier Max
Non-Outlier Min
l~l Median; 75%
25%
o Outliers
105
-g 95
o
co
«
CO
1 85
65
55
45
Figure 3-1 Comparison of Removals of Microsand
Long
Cergrene
T~ Non-Outlier Max
Non-Outlier Min
l"~l Median; 75%
25%
o Outliers
Percent Removed from Time Zero: Time =10min, SAND
Incorporating a 20% Assumed Loss in Analysis
Figure 3-2 Comparison of Removals of Microsand with Error Included
3- 12
-------
Initial Concentration Comparison
Figure 3-3 (O'Connor et al. 1999a ) shows that the CERGRENE column for determining
the initial SS concentration for all experiments had less variation and was closer to the initial
known concentration of 300 mg/1 than the Long column. The CERGRENE column, a nearly
instantaneous initial concentration measurement, approached 300 mg/1 with a small deviation.
The Long column required at least 40 s, and up to a minute as previously stated, before the four
ports could be sampled for initial concentrations, thus "missing" many of the faster settling
particles. Figure 3-3 also shows that initial concentrations, which should be equivalent
throughout the column, decreased towards the top of the Long column (Port 8>7>5>3).
Background concentrations for the Long column, measured by taking samples directly from the
pumped influent to the column, averaged 270 mg/1 with a standard deviation of 30 mg/1.
400
350
2- 300
0)
~ 250
o
1 200
8
g 150
100
50
0
8
T
_l_
e
T
_l_
o
o
1
_L_
o
*
o
T
|
o
o
V1
f
1
if,
Ports Ports Port7 Ports Cergrene
Concentrations at Time Zero (Influent = 300 mg/1)
T Mean; Mean+SD
Mean-SD
o Outliers
*E> Extremes
Figure 3-3 Comparison of Initial Concentrations
Discussion and Recommendations
The objective of the Phase II laboratory experiments was to compare, in side-by-side
analysis, two methods of measuring settling velocity distributions of solids in water. Laboratory
study results indicated that the CERGRENE columns have some advantages, such as ease of use,
smaller testing volumes and a consistent initial concentration, but also significant problems such
as loss of SS mass and lack of reproducibility for other time measurements. The Long column
had its own advantages, such as repeatability and consistent (predictable) SS removals, while the
disadvantages included poor initial concentration measurement, large testing volumes and large
number of SS analyses required.
Phase II experiments were also designed to determine the optimal withdrawal point from
the mixing basin as well as the withdrawal order on experimental results. Statistical analysis
indicated that neither of these aspects of the experimental design were significant. Phase I
3- 13
-------
preliminary testing of microsand (not shown) indicated that the John Meunier mixing basin set
up provided more than adequate lift to suspend particles and that the sampling points with the
least random concentrations were toward the top of the water column in the mixing basin.
The original experimental design did not account for the three factors which most
affected results during Phase II of the experiment:
1. Individual behavior of the CERGRENE columns
2. Overloading of the filter by the samples (when trying to pass whole sample through filter)
3. Wide variation in measured SS concentrations when analyzing micro-sand suspended in water
Initial conclusions from Phase II were:
1. Microsand was more difficult to work with than anticipated. It was difficult to recover
all of the microsand during SS analysis. This is a result of the physical characteristics of
the sand. While behaving discretely during settling in the water column, when not fully
submerged, the particles tended to cling to analytical equipment due to surface water
tension and losses of mass easily occurred. Each sand grain contains significant mass, so
even a loss of several grains can increase errors. This loss was especially noticeable in
the larger CERGRENE samples where the large volume of analyte caused the filters to
become overloaded. The microsand was a major source of error to the experiment;
however this error had a distinct bias toward a loss of microsand.
2. The performance of the CERGRENE columns was erratic for experiments 1-9 in which
four different CERGRENE columns were used. While the expected results of a plot of
concentration versus time would show an increase in the concentration with time, the
columns in experiments 1-9 behaved almost randomly. In contrast, experiments 10-15,
where only one column was used, indicated this trend of increasing concentration versus
time and seemed to point to a lack of consistency between the four columns used during
experiments 1-9. However, even this trend of increasing concentration for experiments
10-15 had scattered results.
3. The Long column was never fully mixed at t0, especially for the microsand and mixture
experiments due mostly to an inadequate way of delivering the mixed analyte to the
settling column.
4. The CERGRENE columns were examined for defects and sizing specifications for
volumes and height. No defects or deviations from the specifications were found, though
a small indentation in the ball valve, part of the functional design, could be trapping some
solids.
Recommendations for the CERGRENE column were:
1. CERGRENE Columns should be modified to allow filling to a constant head (which has
already been adapted by the CERGRENE research group in France). The water height in
the column is an essential measurement of the analysis procedure. Starting at the same
height would also allow for better duplicate analysis.
2. The filling procedure should allow at least a /^ volume to overflow the column.
3. More data points should be evaluated. The matrix analysis spreadsheet "Victor"
currently allows for nine data points not including the initial time and final time. Only
four and five data points were collected for the microsand and other experiments,
respectively.
4. Sample volumes should be reduced to ease analysis. This could be accomplished by
reducing the volume of the bottom portion of the column and/or splitting samples to
avoid overloading.
5. Evaluations of other types of valves should be conducted to attempt to minimize
interference of the valve mechanism on solids settling.
6. The concentration of the top and bottom parts of the column should be measured at t0, to
3- 14
-------
ensure that the initial concentrations, before settling, are the same.
Long column recommendations were:
1. The long column should be retrofitted with a device that allows an overflow to achieve
better mixing and to allow for a repeatable starting depth.
2. A more powerful pump and mixer should be employed to reduce the concentration
gradient of SS in the column at time zero.
3. Delivery of a well mixed plug flow at the top of the column would allow for easier
computation of settling velocity. This method should be further investigated.
General recommendations:
In future experiments using the microsand, the last sample time should be reduced to five
minutes with sampling frequency increased from t0, i.e., 15s, 30s, and 60s. As previously
discussed, the microsand diameters were found to be larger than originally reported which
increased the settling rate.
It was a benefit from the standpoint of this phase to analyze "standards" using the same
media that was used in the experiments. This data confirmed an expected bias (or losses) for the
media, i.e., microsand, in the course of the SS analysis. However, for the sake of confirming the
SS analysis technique, diatomaceous silica should have been used for standard reference
material. In addition, the same sample bottle sizes should have been used for the standard
volumes as the experimental samples.
3- 15
-------
4. RESULTS OF ADDITIONAL LABORATORY SAMPLING
Following the recommendations of the interim report, further analyses were performed
on the CERGRENE column at the UWMB laboratory. These additional experiments were
conducted using microsand and the Neshaminy clay soil as the characteristics of these solids
were already known. To standardize these additional experiments further, all the samples were
all taken at t0, as this was the most consistent sampling time for the CERGRENE columns during
Phase II performed in John Meunier's laboratory in Montreal. The recommendations for the
CERGRENE columns that were addressed during these experiments are:
• A procedure was developed where the CERGRENE columns could be filled to a constant
head, which allows for the columns to start settling at the same height and eases
measurement and computation,
• The same procedure allowed the CERGRENE columns to be overflowed,
• The sampling hose was purged between samples, and
• The SS concentrations in the top and bottom portion of the column were measured at t0,
to ensure that the initial concentrations, before settling, are the same.
An integral aspect of the CERGRENE design, as described in Section 2, is to fill the
columns by vacuum aspiration instead of using positive displacement pumps which, when used
with sewage, tend to breakup the particles and thus change the settling-velocity distribution of
the sample. During the Phase II side by side experiments there was no precise control to the fill
level in the CERGRENE column which was a critical measurement for the procedure.
Overflowing the CERGRENE column eliminated the need for the height measurement and
standardized the volume of the top chamber which was a required input for the Victor model. As
previously mentioned in Section 2, the vacuum pump used for the remainder of the experiment
was also much more powerful and filled the CERGRENE column faster than in Phase II.
The problem associated with the consistent height measurement of the upper chamber of
the CERGRENE column was also addressed by the CERGRENE developers in France, who
developed a similar overflow solution, although the exact configuration was not detailed. Figure
4-1 shows the configuration that was used for the remainder of the experiment, which basically
entails the addition of a filtering flask to capture the overflow from the column. The numbers in
Figure 4-1 detail how the experiment was broken down into several filling capacities to see if
overfilling the samples would improve results.
Only the one column deemed reliable during Phase II was used in these and the Phase III
field experiments. The samples were transferred to 1 L sample bottles and analyzed later. The
intake lines from the mixing basin between each time measurement were purged by vacuuming
two sequential overflow volumes into the CERGRENE columns of potable water between
samples in this and the Phase III experiments. In Phase II, the influent line was only allowed to
drain under its own head and the CERGRENE column was washed out under a tap.
Standard Method analyses were limited to SS as listed in Table 2.4. Two experiments
were performed using the same experimental materials as were used in Phase II. Experiment 1
used Neshaminy clay soil, and Experiment 2 used the microsand. The sampling was conducted
by EPA personnel and laboratory analyses were conducted by U. S. Infrastructure, Inc. (USI).
4- 1
-------
Experimental Volume CERGRENE column filled to:
1 - Bottom - filled column to ball valve
2 - Full - filled column to top (normal procedure)
3 - Overflow - overflowed column into flask
Not to Scale
Flow Direction
Vacuum pump
fills column in 5 s
Filtering flask
filled with dessicant
Filtering flask
for overflow
EPA Mixing Basin
Figure 4-1 New configuration with additional flask to overflow CERGRENE column.
Quality Control Analysis
Blanks and Standards
Blanks were taken with either de-ionized, tap, or basin water before the delivery of the
media. The purpose of blanks was to ensure the cleanliness of SS analysis and sample transfer.
For experiment 1, the measured SS concentration in the lab blank sample was -1 mg/1. A
method blank, tap water in a sample bottle, had a SS value 2 mg/1. Two method backgrounds,
water taken from the full mixing basin before the Neshaminy clay soil was put in, had SS
concentrations of 2 and 17 mg/1. The concentration of 17 mg/1, may indicate some residual
concentration in the mixing basin or the sample bottle. For experiment number 2, a laboratory
blank of 0 and a method blank of 0.93 mg/1.
The SRM samples were from a manufacturer of laboratory control samples
(Environmental Resource Associates, WasteWatR™ Quality Control Samples). The accuracy
expressed as percent recovery (equation 3-1) for the two laboratory standards that were run (one
for each experiment) was 101% and 104% respectively. The SRM samples were within the
control limits as provided by the manufacturer.
Two 250 mg/1 QA samples of microsand were prepared. The percent recovery of these
samples were only 48% and 67%. These results were typical of the Phase II results of the SRM
samples (refer to Figure A-3 in Appendix A) and as different personnel performed the SS
4-2
-------
analyses this serves as an independent confirmation of the typical results found during Phase 1
for microsand. Since the Neshaminy clay soils did not have similar recovery problems sas the
microsand samples during phase 1, no QA Neshaminy samples were prepared
Completeness
Table 4.1 shows the completeness for critical SS measurements made for both
experiments. Completeness is calculated using equation 3-2. Of the 24 separate CERGRENE
samples, two were voided due to faulty data recording.
Table 4.1 Completeness of Suspended Solids Analysis
Samples
ALL
CERGRENE
Samples
31
24
Voided Samples
2
2
Expected
Completeness
90%
90%
Measured
Completeness
94%
92%
Laboratory Duplicate
One duplicate was performed for these two experiments. The laboratory RPD was
calculated to be 0.51. The duplicate was performed on the Neshaminy clay soil as only a portion
of the total sample was used, 100 ml out of approximately 1 L. The microsand experiments used
the whole volume of the sample which was rounded off to 1000 L for all CERGRENE samples,
which may have introduced some error to be discussed under the heading Mass Balance.
Results of Experiments 1 and 2
Tables 4.2 and 4.3 show measured concentrations and statistical analysis of the two
experiments. Here RPD (equation 3-3) or RSD (equation 3-4) represents duplicates or triplicates
of the CERGRENE columns dependent on the volume flowing in and through the column. The
initial concentration of the Neshaminy clay soil was 270 mg/1 and the microsand was 290 mg/1
assuming the mixing basin was filled to capacity. The precision of the CERGRENE columns
can be calculated from the duplicates or replicates of a sample at a specified time.
Results would appear to indicate two things, one positive and the other somewhat
negative. The samples with the flow-through method had the lower RPD or RSD values and
therefore overflowing the columns is a better method of filling the CERGRENE columns.
However, averages at t0 appear to be lower than the expected results of 270 mg/1 and 290 mg/1
for Neshaminy and microsand, respectively. Phase II t0 results approached 300mg/l, which was
the known starting concentration. This could indicate that the EPA mixing system was not as
proficient at lifting the heavier particles as the John Meunier set up used in Phase I; however,
this results still represent > 90% recovery. As noted during the experiments some solids were
seen in the fitting grooves of the effluent valve of the CERGRENE column.
By sampling the top portion of the CERGRENE column of the experiments that either
filled or overflowed the CERGRENE column, an attempt was made to verify that the
concentrations were equivalent around the center ball valve of the CERGRENE column. This
comparison was made by looking at the average concentrations of the experiments listed as
Through and Top in Tables 4.2 and 4.3. This may not be as easy since the very act of turning the
ball valve to drain the top portion will capture and separate a small volume of water from either
the top or bottom portions.
4-
-------
Table 4.2 Experiment 1 Suspended Solids Concentrations of Neshaminy Clay Soil
Sample
Bottom 1
Bottom 2
Bottom 3
Full 1
Full 2
Full3
Through 1
Through 2
Through 3
Topi
Top 2
Top3
Description
15 cm above center ball valve
12 cm above center ball valve
8 cm above center ball valve
full, no water in flask
full, some water in flask
full, flask < 1/2 full with water
full, flask > 1/2 full with water
full, flask > 1/2 full with water
full, flask > 1/2 full with water
Top of Full 1
Top of Full 3
Top of Through 3
Concentration
mg/1
129
197
233
198
55
220
176
192
206
209
197
209
Mean
186
158
191
205
• •
53
90
15
6.9
RSD
(%)
28
57
7.8
3.4
Table 4.3 Experiment 2 Suspended Solids Concentrations of Microsand
Sample
Bottom 1
Bottom 2
Bottom 3
Full 1
Full 2
Full 3
Through 1
Through 3
Top 2
Top 3
Description
1 1 cm above center ball valve
6.4 cm above center ball valve
1.3 cm above center ball valve
full, no water in flask
full, some water in flask
full, flask < 1/2 full with water
full, flask > 1/2 full with water
full, flask > 1/2 full with water
Top of Through 2
Top of Full 3
Concentration
mg/1
164
134
218
162
266
89
145
169
159
185
Mean
172
172
157
172
• •
42
89
—
—
RPDor
RSD (%)
25
52
15
16
In addition, volume measurements of the CERGENE column were done in the laboratory
after Phase III field sampling was completed. The total volume of the column was
approximately 2470 ml. The volume of the bottom portion of the column averaged 967 ml but
was set at 960 ml for all Phase II and III calculation. The volume of the top portions of the
column was more difficult to measure, as by the time of these volume measurements were made
the column began to leak. Precise volumes for the top portion could not be obtained without
including the ball valve component for a total volume of 1510 ml. The top chamber ranged from
4-4
-------
1160 to 1235 ml, although it was approximately 1280 ml. The ball valve component measured
approximately 230 ml.
This difficulty with obtaining volumes of the top portion of the column underscores the
problems with trying to confirm equal concentrations on either side of a ball valve. The bottom
portion of the CERGRENE column had to be drained before the top portion could be drained.
The samples bottles used to collect the top portion held approximately 1 L, which meant the
whole 1280 ml top volume could not be drained nor could the 230 ml contents of the ball valve.
In the time it took to set up the CERGRENE column to drain the top portion, some settling may
have occurred in the top portion and sample SS concentrations may not have been uniform.
This may have been more critical in the case of the Microsand where the whole sample
volume was analyzed, the mean SS concentrations of the upper chamber reported at 172 mg/1
(reported as Top in Table 4.3) may have been smaller by a 1000/1280 ratio if settling of the
microsand occurred. The SS concentration of the upper chamber may have been as low as 134
mg/1. The lower chamber (Through in Table 4.3) measured at 157 mg/1 for the samples.
For the Neshaminy, the volume differences may not have been as significant as only a
100 ml volume of the sample was analyzed for SS concentration of the lower and upper
chambers (labeled Through and Top in Table 4.2) with concentrations of 191 and 205 mg/1
respectively. The slight increase in concetration of the Neshaminy in the upper chamber may
have been caused by lighter material being carried up by the large Reynolds numbers (6000 -
50,000 as calculated in Appendix M) produced by the flow to fill the column. It could also be
caused by the settling mentioned above. However, this increase of the upper chamber
concentration is not statistically significant and therefore shows good agreement, and is evidence
of a mass balance between chambers for the lighter material. The same cannot be conclusively
said for the Microsand. What is significant is that in the cases measured, overflowing the
column led to a measurable SS concentration in the upper chamber.
Other Considerations
These experiments indicated that filling-through gave the most consistent and easiest to
perform testing method for the CERGRENE columns. The selected procedure to fill the
CERGRENE column was to turn off the vacuum pump once the first filtering flask was half
filled (corresponding to the number 3 position in Figure 4-1).
By overflowing, the pump was on longer. This decreases the fraction of start up time for
the pump to the total time period the pump is on. This may minimize start-up velocity flow
fluctuations and possibly allows the pumped sample to achieve steady state flow through the
system, i.e., column and influent hoses. For the idealized case, the time to one half the value of
the steady flow velocity could be several seconds, i.e, 3 - 6 for the CERGRENE column and up
to approximately /^ s for the influent tube. If all head losses, due to bends and turns, surface
roughness, and kinematic viscosity are taken into account, these approximations may change.
Appendix L shows the calculations for clean water using the measured flowrates of the EPA
vacuum pump. In actuality, due to elastic waves and dampening, steady state is eventually
achieved, while theoretically, the idealized case never reaches equilibrium. In practice, steady
state flow should deliver a more uniform concentration then a flow that is oscillating.
Another consideration calculated in Appendix L is the wall effects on the settling rates in
the column and transport of the sample to the column by influent tubes. The wall effects appear
to be insignificant for the settling rates at the expected concentrations. The column or tube
diameter would have to approach 0.5 mm before a 5% effect were measured.
4-5
-------
Conclusions and Recommendations
Results indicated overflowing column achieved a more consistent result and had the
lowest RSD of the methods tested in these additional experiments. The overflowing of the
CERGRENE Column system was adopted for the field sampling and allowed filling to a constant
head. The water height in the column is an essential measurement for calculation of the settling
rates and also allowed for better duplicate analysis.
Use of diatomaceous earth as the SRM validated the performance of the USI laboratory.
Again the microsand was a major source of error to the experiment, with a distinct bias as the
analysis of a loss of sand.
A measurement of the upper and lower chamber SS concentrations was performed, but
there were limitations to the effectiveness of this comparison. A 2 L sample bottle was needed
for the top portion. The current configuration of the CERGRENE columns uses a ball valve to
separate the top from the bottom; this ball valve holds a small volume of water and complicates
comparisons of the top and bottom SS concentrations.
The EPA mixing basin had a slight mixing as the John Meunier mixing basin design. The
vacuum pump used in these and the field experiment provided a much better influent flowrate.
A more rigorous approach may be needed to determine the fluid dynamics of the
CERGRENE columns and the component pieces.
4-6
-------
5. RESULTS OF PHASE III FIELD TESTING
Several recommendations were made for the CERGRENE and Long columns in the
interim report that were intended to improve the final Phase III field experiments. The
recommendations of further analysis performed on the CERGRENE column were detailed in the
previous section. The Long column recommendations were:
• The long column should be retrofitted with a device that allows an overflow during
filling to achieve better mixing and to allow for a repeatable starting depth.
• Delivery of a well mixed plug flow at the top of the column would allow for easier
computation of settling velocity.
• A more powerful pump and mixer should be employed to reduce the concentration
gradient of SS in the column at time zero.
Other recommendations for the CERGRENE column following Phase II and not addressed in the
additional laboratory sampling were:
• More data points should be evaluated. The matrix analysis spreadsheet "Victor"
currently allows 10 data points not including the initial time and final time.
• Aqueous volumes should be reduced for SS analysis. This could be accomplished by
reducing the volume of the bottom portion of the column and/or splitting samples to
avoid overloading filters used in SS analysis.
Only one of the three Long column recommendations was directly addressed. No
adjustment was made for the first recommendation, although the additional laboratory
experiments covered in Section 4 demonstrated the apparant usefulness of overflowing for the
CERGRENE column. At a minimum, provisions should have been made for at least a minor
controlled overflow so that the column could be filled to the top without guess work or spillage.
An overflow did occur during the third field experiment and the Long column was not sampled
due to the spill which interrupted the integrity of the experiment. This also impacted the 1 hr
time measurement for the CERGRENE column during the third experiment.
No attempt was made to evaluate the second point of delivering a well mixed plug flow
to the top of the Long column as an alternative method. For this method to be successful, it
would have required significant redesign of the experiment and delivery system. This would
have minimized the comparability of pumping from the bottom as the delivery system for both
columns. During Phase I, rudimentary attempts to introduce hand mixed plug flows at the top of
the Long column proved unsuccessful.
The third point was addressed as a more powerful pump was employed to reduce the
concentration gradient of SS in the Long column at time zero. A Moyno pump (300 series) filled
the Long column and pumped the raw combined-sewage from the Perth Amboy, NJ grit
chamber. The Moyno progressing cavity pump, capable of 7.6 m (25 ft) maximum lift at 1750
rpm, was used to minimize the maceration of the combined-sewage sample. Due to the power of
the pump, which had more than sufficient intake velocities, only one sampling port was exposed
to the influent combined sewer flow at the Perth Amboy grit chamber. A 2.54 cm (1 in.) inner
diameter, reinforced tube was manually dropped into the grit chamber and it was bobbed up and
down, and back and forth to provide a variety of material in the mixing basin ( 210 L (55 gal)
were collected). The original John Meunier, Inc. design, described at the end of Section 1, called
for two intake points, but this was envisioned for a mounted, in-sewer sampling location.
The CERGRENE concerns were addressed in the following manner:
• Additional times of 7, 20 and 30 min were added to the experiment.
• The CERGRENE column was not redesigned so the volume of sample from the bottom
5- 1
-------
portion remained the same as in previous experiments. The second field experiment used
split samples for SS and total solids analysis. The first experiment only used a portion of
the sample for SS analysis. The third experiment used the whole sample for total solids
analysis.
Experimental Design
The experiments produced three data sets from the first two field experiments for the
Long column and four data sets from three field experiments for the CERGRENE column.
During these experiments the CERGRENE columns were filled first, except for the last one hr
time measurement. This was due in part to the smaller size of the EPA mixing basin than the
John Meunier Mixing Basin. As the Long column was filled, the mixer would vibrate due to the
decreasing water volume weighing down the mixing basin. For the CERGRENE columns, the
sample collection tube was placed approximately 5 to 10 cm (2 to 4 inches) from the top of the
water level in the mixing basin. Previously, the John Meunier mixing basin was shown to more
than adequately lift the microsand from the bottom of the mixing basin. The EPA mixing basin
did not have the same configuration, and as shown in Section 4, may not have provided the
equivalent mixing. The EPA basin also had a port in the bottom through which the Long column
was filled. This was easier to work with in the field, where there was an attempt to minimize
direct contact between the sampling personnel and the combined sewage, than drawing the
sample out from the top. Figure 5-1 shows a schematic of the field experimental setup.
The columns were transported to Perth Amboy, NJ which is approximately five miles
away from the EPA laboratory in Edison, NJ. The Perth Amboy Wastewater Treatment Plant is
an enclosed facility which allowed for a complex setup and rapid startup as the equipment was
stored securely between events. The sampling and settling distribution analysis was conducted
only during wet-weather flow events when the plant flow charts exceeded 90%, which indicated
a significant wet-weather flow. At measurements of 100% on the flow chart, the system
typically overflowed. The combined sewage was collected from the Perth Amboy grit chamber.
Samples collected from the Long and CERGRENE column were brought back to the laboratory
for solids analysis. At the time of the study, accumulated grit was removed from the wet-well
once every three months.
Several field blank samples had unexpectedly high solids values. To help determine the
cause, additional field blanks were taken at the sampling site after the original Phase III
experiments were completed. An additional round of grab samples were taken from the grit
chamber to perform Settleable Solids (SM 2540F listed in Table 3.1). This sampling also
occurred during a rain event with > 90% flow.
5-2
-------
Long Column
EPA Mixing Basin
Vacuum Pump
Moyno Pump
CERGRENE
Column
Not to Scale
Figure 5-1 Configuration for Side-by-Side Field Analysis
Ideally, the whole sample, i.e., the approximate 1 L bottom portion of the CERGRENE
column or the 250 ml Long column sample bottle, was to be analyzed for solids. During
Experiments 1 and 2, samples were shaken and only a portion were measured for SS and VSS
analysis. The SS concentration of Experiment 1 samples from Perth Amboy were for the most
part under 100 mg/1 and were still clogging the filters before the whole sample could be filtered
for SS. During Experiment 2 only 100 ml was used for SS analysis of all samples. The
remaining portion of the sample from Experiment 2 was analyzed using Total Solids (TS), which
became the preferred method and was also used for Experiment 3 and the additional sampling.
The TS analysis, which uses evaporation, was used as a more pratical approach to completely
analyze the samples without splitting than the SS analysis, which uses filtration. A small portion
of the sample at a time was poured into one dish for TS analysis instead dividing into several SS
samples. Either procedure can introduce error, and the TS analysis was chosen as a better
method to capture all the settleable solids of the sample which was the most important aspect of
sample analysis.
Changing from SS to TS analysis should not have otherwise affected the outcome of
Phase III results. A spreadsheet analysis of the average ratio of TS to SS values was similar to
text book values, approximately 3.4, even though SS was completed within the holding time, and
TS was not. Either solids method may be acceptable, and a portion of the sample may be used
(as was also detailed in Section 3 with regard to the Neshaminy analysis), as long as the samples
are well mixed when split. This is explained further under Quality Control Analysis.
5-
-------
Quality Control Analysis
Blanks and Standards
The laboratory blanks were drawn from laboratory de-ionized water. SS analysis of three
of the four laboratory blanks yielded between -2 and 0 mg/1, which is acceptable. The first blank
for TS analysis registered 29 mg/1, which was high. One explanation offered by the laboratory
personal was that they were working with porcelain dishes which had a tendency to accumulate
water vapor with time. The purpose of the laboratory blanks was to determine whether
laboratory SS procedures introduce error in the SS analysis.
Field blanks were taken from a tap of the Perth Amboy potable water supply located in
the grit chamber room. The purpose of the field blanks was to determine whether other aspects
of the procedure, e.g. storage bottles and handling procedures, introduce error into solids
analysis. Because of an apparent problem with the analysis of the field blanks taken on 3/28/00
and 5/19/00, some additional field blank samples were taken on 6/6/00 after the Phase III field
measurements. All laboratory and field blanks results are listed in Table 5.1.
Table 5.1 Analysis of Blanks
Phase III
Experiment
1
2
3
Additional
Field
Blanks
Date
3/17/00
3/28/00
5/19/00
6/6/00
Blank
Type
Laboratory
Field
Laboratory
Laboratory
Field
Laboratory
Field
Field
Field
Field
Field
Laboratory
Laboratory
Concentration (mg/1)
SS
0
0
95
VSS
-2
0
TS
29
543
0
244
199
188
172
118
-25
-7
Comment
Voided by laboratory
Perth Amboy tap water
Perth Amboy tap water
Perth Amboy tap water
Perth Amboy tap water
Perth Amboy tap water
Laboratory tap water (Edison)
De-ionized in dish
De-ionized in sample bottle
Because the field blank for Experiment 1 was voided (due to difficulties with static
charge build up on porcelain dishes during weighing, as were three other samples), only the
Experiment 2 field blank SS and TS concentration requires further explanation. The TS results
for Experiment 3 are not significantly different from the results obtained from 6/6/00 additional
field samples. Both results indicate that there is a large loading of TS in the Perth Amboy tap
5-4
-------
water, approximately 200 mg/1, compared to the Edison potable water supply, approximately 100
mg/1. The use of tap water for blanks was not a problem during the Phase I and Phase II results
because the samples were filtered and dissolved solids were not analyzed.
The Experiment 2 field blank had unexpectedly high results for both the SS and TS
analysis. This could have been caused by any number of reasons, i.e., faulty sampling technique,
using a dirty sampling bottle, mislabeling of the sample in the field or the laboratory, or cross
contamination of samples in the laboratory. Because additional field blanks results were not
taken during Experiment 2, it is difficult to specifically identify where the fault lies.
QA objectives for solids analyses were limited to the parameters listed in Table 5.2 and
those previously listed in Table 3.2. Two types of SRM were provided. One from a
manufacturer of laboratory control samples which USI purchased. The second were made by
EPA and provided to the USI laboratory personnel.
Table 5.2 QA Objectives for Measurements *
Measurement
Total Solids
Volatile Solids
Method
2540 B
2540 E
Reporting
Unit
mg/1
mg/1
Initial
Concentration
NA
170
Standard
Deviation
6.0
11
Relative Standard
Deviation
NA
6.5%
Complete-
ness
90%
90%
From Standard Methods (1995)
The manufacturer SRM samples were from Environmental Resource Associates,
WasteWatR™ Quality Control Samples. For SS and TS analysis, the manufacturer SRM
samples were all within the manufacturer provided control limits. The accuracy expressed as
percent recovery was calculated using equation 3-1. SS percent recovery ranged from 89 —
104% and was well within the expected values in Table 3.1. The analysis of the manufacturers
TS samples had a standard deviation of 2.9 for 5 samples which was less than the Table 5.1
value of 6.0.
The VSS samples were specially made and did not have manufacturer control limits.
Even though the true value of the VSS SRM, i.e., 100 mg/1, did not coincide with the value in
Table 5.2, 170 mg/1, the values of the RSD for two sets of five VSS SRM samples were within
QA objective limits (< 6.5%) at 6.49% and 3.3%, respectively.
Two SRM samples were made from the Neshaminy clay soil by EPA. Results from
Phase II indicated that the Neshaminy clay soil performed well as a SRM and provided SS
results that were comparable to the Diatomaceous-silica reported in Table 3.1. The %R
(equation 3-1) of the samples were 0.68% and 2.4% for known values of 979 and 232 mg/1,
respectively. These values of %R fit within the expected RPD of Diatomaceous-silica in Table
3.1.
Because the standards and the laboatory blanks tested well, the anomalous results
obtained from the analysis of field blank for Experiment 2 cannot be adequately explained and
appears to only be a glitch in the sampling and analysis procedure. Results from experiment 3
and the additional field blank analysis confirm this. In future experiments it is recommended
that more field blanks are collected during each experiment.
5-5
-------
Completeness
Table 5.3 shows the completeness for critical solids measurements made for all
experiments. Completeness was calculated using equation 3-2. Of the 31 separate CERGRENE
samples, two were voided due to faulty data recording. Samples from Experiment 2 had more
than one type of solids analysis performed.
Table 5.3 Completeness of Solids Analysis
Type* or location
Lab Blanks
Field Blanks
Manufacturer Lab Standards
Prepared Standards
CERGRENE
Long***
Background (start and end)
Total
Samples**
4
3
27
2
31
65
12
Voided
SamplesT
0
1
0
2
0
5.5
1
Expected
Completeness
90%
90%
100%
100%
90%
90%
90%
Measured
Completeness
100%
66%
100%
100%
100%
92.3%
92%
*
**
***
Does not include additional field blank or Settleable Solids sampling.
There was an extra sample reported by the lab, which was not recorded in the log book or
on the chain of custody sheets for Experiment 2.
No samples were taken during the third experiment. Recorded as 5.5 because the second
solids analysis (total solids) on sample was voided due to result being below zero.
T One Long and two CERGRENE column samples were not included in the graphical
settling-velocity analyses because the values were extremely large and threw off the
analyses. These data omissions are noted here and later, and do not reflect laboratory
performance as a Void because the laboratory was instructed to analyze everything
contained in the sample.
Additional QA Concerns: Ratio of Total Solids to Suspended Solids
TS analysis was performed on the samples of Experiment 2, even though samples did not
meet holding requirements according to Standard Methods at the time of analysis (after
completion of SS analysis samples were stored at ambient temperature awaiting disposal). The
analysis was performed because it was more important to attempt to analyze the settleable
material, which is less likely to have been impacted by biochemical degradation, than to adhere
strictly to Standard Methods. Also, because all samples were from the same place and date and
experienced the same processes, storage effects should have been similar.
To assess whether the TS values for experiment 2 had any validity, an analysis of a
typical ratio factor of TS to SS results was performed. The results presented in Table 5.4 are the
averages, • *and RSD of the RPD of TS measured to TS calculated of individual samples. The
RPD of individual samples was used as values were changing with time.
5-6
-------
Table 5.4 Comparison of Total Solids to Suspended Solids Ratio
Column
Type
CERGRENE
Long
Relative Percent Difference *
Samples
11
o o
JJ
Average
(mg/1)
30
18
• •
(mg/1)
28
32
RSD
(%)
92
170
Relative Percent Difference less extreme *
Samples
10
32
Average
(mg/1)
23
13
• •
(mg/1)
12
10
RSD
(%)
52
79
Ratio from
data**
2.9
3.6
* Calculated TS (where TS/SS = 3.4) to Measured TS
** Ratio of measured TS to SS
Often, TS or SS values can be predicted by using an assumed ratio between the two. A
derived ratio of 3.4 from literature values (averaged values from Metcalf and Eddy ,1991 , p. 109,
Table 3.16 ) was used on SS results to predict TS, and overall, the results bear out the validity of
the TS analysis. There appears to be no substantial change in the results of the between the SS,
completed within the holding time and that of the TS, which was not. While there was large
individual variability, as borne out by s exceeding the average, this can be reduced by
eliminating the one extreme data point from each column type. The extremes were caused by
exceptional large pieces of grit, i.e., pebble sized, greater the 2 mm, and once the extreme values
are removed, the averages dropped with • -dropping below the average. The multipliers as
calcualted from the data were 2.9 and 3.6 for the CERGRENE and Long columns respectively,
with • «=0.6 and RPD = 0.2, approach the literature value used.
The analyses conducted here was on portion of samples. As results would indicate that
SS values can approximate the TS, when using a portion of the sample, multiple analysis of
discrete volumes from one sample may be valid for any settling test, as it appears valid for these
two methods. To decrease values of s between the split samples, whether performed as
duplicates or separate analysis, samples should be thoroughly and continually mixed during
withdrawal.
Field Experiments 1-3
Mixing Basin - Background
Several background samples were obtained during the experiments. The background
samples consist of combined sewage samples taken by hand grabs from the top of the mixing
basin and were collected and analyzed to establish starting and ending concentration in the
mixing basin. The average background concentration of each test compared well to the t0 of the
CERGRENE and Long columns. The results of the background sample analyses are presented
in Table 5.5.
T-test for paired two sample for means was performed to determine whether
concentrations of the SS, VSS, and TS changed significantly during the course of the Experiment
2, with the following results:
Suspended Solids Total Solids
tStat 1.052632 -2.2
P(T<=t) one-tail 0.24184 0.1358
t Critical one-tail 6.313749
The P(T<=t) one-tail value is in excess of 0.05, which indicates that the SS concentrations in the
mixing basin before and after filling of the columns are indistinguishable.
5-7
-------
Table 5.5 Solids Concentrations in Mixing Basin during Field Experiments
Phase III
Experiment
1
2
3
Sample
Type
Start
Start
Start
End
End
Average
Start
Start
End
End
Average
Start
Start
Start
Average
Suspended Solids
(mg/1)
46
50
60
Void
56
53
200
157
161
156
169
Volatile Suspended Solids
(mg/1)
26
31
32
Void
22
28
Total Solilds
(mg/1)
607
518
497
482
526
276
304
301
294
Initial Concentration
An inherent problem in the design of the Long column is the lack of reliable uniformity
in initial concentration (C0) throughout the length of the column. C0 is used in calculations for
the Percent Removal. However, while this was very noticeable in the Phase II experiments, it
did not appear as a significant problem with the well mixed CSO in the field. Table 5.6 shows
that the initial C0 measured in both columns compare well to the average background
concentration and to each other.
A T-test for two samples assuming unequal variances was run on Experiment 1 VSS
results because these had the widest variance, due in part to the low concentrations. The initial
three samples of the Long column and the starting Background concentration were used to
determine the concentrations were distinct, with the following results:
Mean
Variance
Observations
Hypothesized Mean Difference
Degrees of freedom
tStat
P(T<=t) two-tail
t Critical two-tail
Long Column
35
43
O
0
O
1.276964215
0.291485258
3.182449291
Background Start
29.62416107
10.16873114
5-
-------
The P(T<=t) two-tail value is in excess of 0.05, which indicates that VSS concentration values
and therefore SS concentration values in the mixing basin before the experiment began and in
the columns at t0, are indistinguishable.
Table 5.6 Comparison of Initial Column Concentrations to Background
Phase III
Experiment
1
2
3
Sample Type
Long, tj- 10 Port 3
Long, t2- 10 Port 5
Long, t3- 10 Port 7
Long, t4- to, Port 8
Average Long
CERGRENE, toa
Average Background
Long, tj- to, Port 3
Long, t2- to, Port 5
Long, t3- to, Port 7
Long, t4- t0, Port 8
Average Long
CERGRENE, toa
Average Background
CERGRENE, toa
Average Background
Suspended
Solids (mg/1)
Void
52
46
89
62
68
53
165
171
180
151
167
167
169
Volatile Suspended
Solids (mg/1)
Void
28
36
41
35
33
28
109
112
Total Solids
(mg/1)
518
522
526
547
528
554
526
241
294
CERGRENE Duplicate Analysis
Table 5.7 shows the duplicate analyses for the CERGRENE Columns. All duplicates
were performed on the samples collected t03, 3 minute time step.
Table 5.7 CERGRENE Duplicate Analysis
Phase III
Experiment
1
2
O
Relative Percent Difference
Suspended Solids
21
0.47
Volatile Suspended Solids
15
Total Solids
3.9
0.11
Phase III RPD results were similar in range to Phase II results (0.25 - 22.2); however, the
Phase II experiment used a known SS concentration of 300 mg/1 and higher RPD values was
primarily influenced by the use of Microsand. The higher RPD value for Experiment 1 in Phase
5-9
-------
Ill is due in part to the lower SS concentration of experiment 1 (< 100 mg/1) of the combined
sewage.
Concentration versus Time
Figures E-l through E-10 show plots of the raw data for the Long and CERGRENE
columns for the 3 experiments of Phase III. As in Phase II experiments, the CERGRENE
columns should show an increasing pattern in the graphs, and the Long column should show a
decreasing concentration (which can be confound in the lower ports by Zone settling and
compaction, as previously mentioned). In Experiment 2 for the Long column a sample with a
concentration exceeding 9000 mg/1, has been omitted from Figure E-7 because it would impact
the scale of the figure.
The concentrations measured in the Long column show a decreasing trend in Figures E-l,
E-3, E-5, and E-7, and especially in E-10 which is an isolation on Port 3 of Experiment 1.
Figure E-10 shows an approximate 40% decrease in concentration occurring in the upper portion
of the column, though data are not corrected for water depth.
The concentrations measured in the CERGRENE columns show only slight, if any
increase with time in Figure E-2, E-4, E-6, E-8 and E-9. These figures also show the sharp
changes and spikes in concentrations which cause problems with the Victor Matrix Analysis
tool. This tool requires monotonically increasing data points which exceed the preceding
sampling time step and in all cases the initial concentration. Inflection points (the 1, 3, 5, 10 and
20 minute time steps in Figure E-9) would have to be eliminated before a graphical result could
be produced, leaving only three remaining data points for use in analysis
Percent Removal Long Versus CERGRENE
The design overflow rate was plotted versus the percent removal as was described earlier
in Section 3. Two changes to the procedure were made. For the Long column, percent removal
is defined as the SS concentration at the port compared to the average of the initial readings of
all four ports which is the theoretical C0 at every port, which is how the analysis is traditionally
conducted. This is different than in Phase II, where the recycle concentration for each run was
used as C0. This change was made because the concentrations in the basins compared well to the
initial concentrations in the columns as shown in Table 5.6. In Phase II discrete particles, e.g.,
microsand, were used which had rapid settling rates, while Phase III used combined sewage,
which had a wider range of particle sizes and settling rates. Also, the recycle samples of Phase
II were drawn from the same pump used to fill the Long column and the background samples of
Phase III were hand grab samples.
For the CERGRENE column, the volume of the lower chamber was assumed to be 960
mL, as was done previously in Phase II, and the upper chamber was set to 1510 mL. The method
of overflowing the CERGRENE column ensured that standard volumes could be used. This
eased calculations and fulfilled one of the recommendations from Phase II. These volumes were
also used in the Victor analysis.
Figures F-l through F-10 show plots of the Percent Removals of solids for the Long and
CERGRENE columns for the experiments of Phase III. As in Phase II Experiments, the graphs
are expected to have a higher Percent Greater value for the slower settling rates and a lower
Percent Greater value for the higher settling rates. Logarithmic best fit lines are presented for
direct comparison between columns. Several of the CERGRENE graphs had correlation
coefficients, R2, higher for linear than logarithmic plots, i.e., Figure F-6 with R2 = 0.33 and
Figure F-8 with R2 = 0.69. Scales were not normalized because there was widespread scatter in
the results. The subsection "Design Removal Comparison" below will normalize results.
5- 10
-------
In Experiment 2 for the Long column a sample with a concentration exceeding 9000
mg/1, has been omitted from Figure F-7, and as noted on the Figures F-6 and F-8, two other data
points were omitted which would have caused inverted trend lines, for the CERGRENE column
analysis. Inclusion of the data points would have created trend lines that predict suspension of
particles with time instead of settling. These latter data points are not included in the Victor
analysis, with or without suppression.
The Long column represented in Figures F-l, F-3, F-5, and F-7, shows greater percent
removal potential than the CERGRENE column. The CERGRENE columns show only slight, if
any, increase in solids removal with increasing overflow rate in Figure F-4, F-6, F-8 and F-9.
The exception is F-2, which has a large variance, R2 = 0.11. When the two data points in excess
of 100% removal are eliminated, the variance is decreased (R2 = 0.33) but the removals are
nearly zero and increase with decreased settling rates only slightly, as shown by F-10.
Matrix Iteration Process for CERGRENE Columns
The settling velocity distributions for Phase II using the spreadsheet application,
"VICTOR," are presented in Appendix G. Each figure in Appendix G is comprised of six
graphs from the experiment and the solids analyses performed. Graphs in the top row included
all data, except those points previously described as being omitted from analysis and graphs in
the bottom row were analyzed with suppression.
Reading from left to right, the first graph (represents the matrix-calculated mass removed
over time (described as M(i,j) in Section 3 and Appendix L), the second is the idealized percent
removal of mass from the upper chamber over time, and the last represents Settling Velocity
Distributions, (at Vs90, 90% of the particles measured have a settling velocity less than the y-
coordinate value). The scale on the graphs of Mass Removed (mg) versus Time (s) and Settling
Velocity (mm/s) versus Velocity Percentile are the same for both sets of graphs (without and
with suppression). The y-coordinate for Percent Removed (%) versus Settling Velocity (mm/s)
was not scaled for both analysis. This graph typically had a default setting from 0% to 100%, a
sensible limitation on percent removals. However, in order to expose in greater detail the
problems with the analysis, this default was not used.
In the first row of each of the figure where all data was used (no suppression), there is
much scatter. This is demonstrated in the Mass Removed graphs and the Percent Mass Removed
graphs which show severe scatter and wildly exceed the theoretical limits of 0% or 100% for all
five data sets. When the data is too scattered, the Settling Velocity Distribution graph defaults to
a null set as it did in Figure G-2 and G-3. When data points are suppressed the graphs have less
scatter. Results make more sense for Mass Removed and approach the limit of 100% removal,
i.e. the error exceeds 100% by 40% in Figure G-4 which is a 120% improvement. However,
eliminating data, while allowing the matrix to work, may overall yield less predictive results.
While the use of data suppression resulted in the creation of Settling Velocity Distributions for
all data sets instead of null sets, there are less Velocity Percentiles produced for some of the
experiments as a result of using suppression, e.g., Figure G-l, G-2 and G-5.
Experiment 3 was shown to have limited removals by the previous analysis, as
represented in Figure F-9. In Figure G-5, the suppressed data set predicts that 90% of the
particle measures have settling rates of less than 0.5 mm/s, in comparison to the graph above
which indicates a Vs90 of 4.5 mm/s for all the data and has a larger spread of Velocity Percentiles.
There was scatter in all the data sets, and suppressing data points appears to limit not improve
settling velocity distribution interpretation.
5- 11
-------
Design Removal Comparison
The objective of this project was to compare the accuracy and precision of solids
concentrations taken over a specified time range from two separate settling column methods.
These measurements were used to develop settling velocity distributions. Currently no direct
graphical method for comparison is available for the two columns as each method is derived
through separate mathematical procedures. Direct graphical comparisons were made for
measured parameters, i.e., solids concentration removal and time. This type of analysis
demonstrated the differences in the processes and was used to indicate problems with the
CERGRENE design (Appendix C).
Comparison of the predicted settling velocity distributions of the Long and CERGRENE
columns can be inferred from Table 5.8., which shows calculated overflow rates versus percent
removals for the Long and CERGRENE columns for experiments 1, 2 and 3. Calculations were
identical to those used to develop the graphs in Appendix F. Comparisons of Long to
CERGRENE results shows some similarities, though CERGRENE analysis is complicated by an
insufficient number of data points. Interestingly, the results of the Long Column are similar for
each experiment, predicting initial removals followed by a flattening out at 50% removal, which
is typical of primary treatment removals. Some of the CERGRENE results, i.e., the total solids
analysis, show no removal at all.
Results from the Victor analyses were wildly divergent as shown in Appendix G and are
not represented in Table 5.8. The Victor matrix analysis results were limited to a range of 0.004
to 0.5 cm/sec, however it is difficult to link this to a percent removal because of the scatter in the
data. It is interesting to note that the CERGRENE column's limited maximum settling velocity
predictions, 0.5 cm/s by the Victor analysis and 0.54 cm/s from Table 5.8, is an order of
magnitude larger in the Long column, 1.5 cm/s. This is partially driven by the configuration of
the CERGRENE column which is shorter than the Long column, and the minimum time interval
used during the experiments of 1 minute.
5- 12
-------
Table 5.8 Comparison Predicted Removal between Long and CERGRENE Columns
Experiment
(Analysis)
1
(SS)
1
(VSS)
2
(SS)
2
(TS)
3
(TS)
Percent
Removed (%)
10
30
50
10
30
50
10
30
50
10
30
50
10
30
50
Overflow Rate (cm/s)
Long
1.5
0.057
0.0022
4.4
0.14
0.0046
1.5
0.29
0.00055
0.54
0.0034
0.000022
R2
0.23
0.38
0.55
0.30
--
CERGRENE
0.13
0.057
0.026
0.029
0.00038
0.0000049
0.54
0.0034
0.000022
0.0046
0.00000074
1.2 xlO'12
0.000049
7.9x10-"
1.3 xlO'16
R2
0.11
0.33
.31
0.47
0.31
5- 13
-------
6. CONCLUSIONS AND RECOMMENDATIONS
General Conclusions
The CERGRENE settling velocity distribution test did not achieve all theoretical
expectations. The anticipated benefits of a smaller, easier to use settling system has to be
weighed against the ability to reliably produce usable results. There appears to be a tradeoff
between settling testing volume and the number of experiments; larger settling volume may
require less experiments, while smaller testing volumes may require more experiments to
develop reproducible results.
Experimental Conclusions
The Phase I Preliminary Results established the predicted homogeneity of the mixing
basins and the initial performance of the Long and CERGRENE columns. This testing showed
that adequate mixing was provided in the mixing basin, that SS were transferred to the settling
columns for further testing, and that the testing material of Phase II, i.e., microsand and
Neshaminy clay particles, was recoverable in the columns. The Long column had insufficient
head to sample quickly from the top two ports, so only ports 3, 5, 7 and 8 were used during the
experiments.
The objective of the Phase II experiments was to compare, in side-by-side analysis and
under the same conditions, the Long and CERGRENE column methods for measuring settling-
velocity-distribution. Phase II was also intended to determine the optimal withdrawal point in
the mixing basin and to examine the effect that filling order of the columns has on experimental
results. Phase II study results indicated that the CERGRENE columns had some comparative
advantages, such as ease of use, smaller testing volumes, and a consistent initial concentration,
but also significant problems such as loss of SS mass during testing, lack of reproducibility, and
large SS analysis volumes. The Long column had its own advantages, such as repeatability,
consistent (predictable) SS removals, and smaller SS analysis volumes while the disadvantages
included lack of uniform concentrations through column, large testing volumes, and large
number of SS analyses required.
The original experimental design did not account for several important factors affecting
results during Phase II of the experiment:
1. Individual behavior of the CERGRENE columns - one CERGRENE column does not
behave like another
2. Overloading of the filter during SS analysis by the CERGRENE samples
3. Wide variation in SS concentrations in the attempt to capture the heaviest settling
particles Microsand
4. Size of the Microsand, which had a larger diameter than anticipated and thus faster
settling velocity.
Theoretically all the sand should have settled in five minutes or less and percent recovery
should have approached 100%. Incomplete recovery of the Microsand resulted in predictions of
concentrations lower than 100%. Besides the problems with the SS analysis already discussed
under Section 3 Completeness, a small indentation in the ball valve, part of the functional design,
could remove the particles from the water column and prevent any chance of recovery.
Conclusions from Phase II were:
1. Microsand was more difficult to work with than had been anticipated, especially during
SS analysis. This was a result of the physical characteristics of the sand. The analysis of
6- 1
-------
microsand produced the most pronounced losses because particles tended to stick to
analytical equipment through water tension. Each sand grain contains significant mass,
so even a loss of several grains can increase errors. This loss was especially noticeable in
the CERGRENE samples where the large volume of analyte caused the filters to become
overloaded. Overall, the use of microsand was a source of error to the experiment;
however, this error was distinctly biased in the direction of a loss of sand or incomplete
recovery as demonstrated by the analysis of the "standard reference material."
2. There was a lack of consistency between the CERGRENE columns. The performance of
the four CERGRENE columns was erratic for experiments 1-9. While the expected
results of a plot of concentration versus time should have shown an increase in the
concentration with time, the four columns in experiment 1-9 behaved randomly. In
contrast, experiments 10-15, where only one column was used, indicated a trend of
increasing concentration versus time. This indicated that the four CERGRENE columns
being tested each had unique properties that influenced the settling experiments.
3. The initial SS concentrations through the Long column was never uniformly mixed at t0,
especially for the microsand and microsand/clay mixture experiments.
4. The CERGRENE columns were examined for defects and sizing specifications for
volumes and height. The four CERGRENE columns appeared similar, no anomalies
were found, and no explanation for varying results could be found. A small indentation
in the ball valve, part of the functional design, could trap some solids.
After Phase II, the following recommendations were adopted for the Phase III analysis:
1. The CERGRENE column procedure was modified to allow filling to a constant head as
the water height in the column is an essential measurement for this test. The filling
procedure allowed at least a 0.5 L overflow. Starting the CERGRENE columns at the
same height allowed easier mathematical analysis, especially with the use of Victor.
2. Additional data points, i.e., times of 7 and 20 minutes, per experiment were evaluated for
the CERGRENE column.
3. The SS concentration of the top and bottom parts of the column were measured at t0, to
ensure that the initial concentrations, before settling, were the same above and below the
ball valve.
4. More powerful pumps were employed for each column procedure to reduce the
concentration gradient of SS in the columns at time zero.
During the Phase III field test, results of the CERGRENE column settling velocity
distributions were consistently flat as in Phase II. The setup may have provided more solids to
the Long column, as it was filled from the bottom of the basin. However, concentrations taken
from the mixing basin by hand were statistically similar before and after column filling, and to
initial concentrations of both columns. It is also noted that concentrations measured in the
CERGRENE columns did exceed the background concentrations in the mixing basin and initial
concentrations in the CERGRENE column but did not do so in a monotonically increasing
manner that is required to complete analysis using the Victor matrix program.
Problems with filling the Long column to achieve a uniform initial concentration t0 were
clearly detailed during the Phase II analysis but were not significant during Phase III. The
primary advantage in the Long column system was the larger number of samples, which means
that the Long column analysis does not rely as heavily on the outcome of any one sample as the
CERGRENE procedure.
There should have been more interim CERGRENE sampling times, i.e, 15 sec, 30 sec, 1
min, 2 min, for the development of settling velocity distributions. This is critical for heavier,
discretely-falling particles as the configuration of the CERGRENE column with an effective
6-2
-------
settling length < 1 m is much shorter than the Long column.
While the Long column may have only limited benefit in the direct measurement of
settling velocities of sewer sediments or discrete solids, it appears to be more than capable in
terms of predicting the percent removal of sedimentary treatment devices with equivalent depths.
In retrospect and for purposes of design, samples taken from the upper ports of the Long column
should have been used; the advantage to Long column is that you are measuring the solids going
away, not the solids that are accumulating as in the CERGRENE columns. Samples were not
collected from the upper ports due to the longer sampling time necessary because of decreased
head, the top ports (1 and 2 in Figure 2-1) eventually fell below the water line as samples were
taken and the perceived need to get as many sample as close together (in time) throughout the
column to improve analyses. However, more samples from the upper ports and fewer from the
bottom ports may actually have improved the settling velocity distribution curves. The Long
column had a visible supernatent developing at the top of the Long column in the Phase III
experiments.
Where possible during these experiments, the whole liter sample size was intended to be
used to reduce additional error in splitting the samples. The sample volumes of the CERGRENE
columns should be reduced prior to solids analysis. This could be accomplished by reducing the
volume of the bottom portion of the CERGRENE column and/or splitting samples to avoid
overloading filters. Splitting CERGRENE samples into replicates or triplicates may help
somewhat with the analysis; however this would not change measured settling characteristics.
Discussion and Recommendations
With the current configuration and analysis protocols, use of the CERGRENE columns
will most likely not result in fewer samples or a decreased sampling effort. The Victor analysis
program did not work properly when results were scattered and not monotonically increasing as
expected. In nearly every experiment performed results were thrown out to run the Victor
program, which limited the number of samples used to create the settling-velocity-distribution
curves. If continued field trips are required until the data resembles a pattern that the Victor
Analysis Matrix can interpret, there is no discernable benefit to adopting this method. Selecting
which data points for the Victor matrix analysis may also bring into question the validity of this
approach. The theoretical basis of the CERGRENE approach did not fulfill expectations.
The adoption of additional procedures from other methods or further analysis of
alternative settling test may be warranted to improve the current CERGRENE method. Gentle
agitation of the column may keep some of the particles from adhering to the sides of the
CERGRENE column especially during the longer experiments. Standard Methods settleable
solids procedure (2540.F.a) recommends to "gently agitate sample near the sides of the (Irnhoff)
cone with a rod or by spinning" to keep material off the sides for a better representation of
settleable solids.
The larger Long column was more consistent and can produce usable results despite the
variability of CSO because it uses more samples than the CERGRENE approach. The Brombach
method, an alternative settling test mentioned in Chapter 1 and not appropriate for direct
comparison, is significant not in the design of the column or the analysis, which separates the
solids from the analyte of sewage before re-introducing solids into clean water, but in the overall
approach of sampling. Brombach (Michelbach and Wohrle, 1993) developed settling distribution
graphs from thousands of data points. While this approach is also subject to data scatter, the
larger number of data points minimizes the effect of variability of individual measurements. The
settling curves (from 35-98 curves for solids from different types of flows) became apparent
because of the sheer magnitude of the sampling program (350 samples, each subject to settling
-------
analysis). Over many storms and many sites, a pattern emerged. The CERGRENE column
analysis currently relies heavily on individual data points, which, when not adhering to an
anticipated concentration curve over time, result in limited settling velocity distributions and
inconclusive results.
More data points could be collected for the CERGRENE column, and the matrix analysis
program Victor should be modified to handle both variable data and larger data sets. These
changes would incorporate data management techniques so that many storms could be sampled
and experiments performed. Incorporating statistical parameters such as averages, standard
deviation, and variance into the analysis may help solve the problem of the variability inherent in
settling data of any one storm or any one sample, and would also provide statistical proof of
outliers. Research may yet show that reliable data may only be achieved by multiple, large scale
sampling programs, which use consistent sampling techniques. A cursory attempt to use Victor
to analyze results using averages results from several of the Phase II experiments performed did
not arrive at monotonically increasing concentrations suggesting that at a minimum a statistically
appropriate number of well run experiments will exceed five.
The nature of the Matrix iteration analysis of the Victor program is not as flexible as are
other forms of analysis due to the constraint of monotonically increasing results. Barring an
outside forces causing resuspension, the amount of solids settling should increase with time.
This common knowledge in itself is not sufficient to eliminate data points or minimize the
influence of unwanted results. Only large amounts of data can minimize the influence of
individual data points that upset expected results.
As both column methods tested in this project used pumps as the means to fill the
columns, the size of the particles that can be tested was limited by the top sampling speed of the
pumps. In practice, a pump that has achieved steady state flow should deliver a more uniform
concentration then a flow that is oscillating. Ideally the purpose of the pump is to deliver a well
mixed sample to the column so that there is a good representative mixture. Each of the two
columns tested here, the Long and the CERGRENE, could be improved with flow through
volumes to achieve steady state flow. This would require switching to a positive displacement
pump for the CERGRENE column from the vacuum pumps which have a limited water volume
that can be drawn up before the water reaches the pump.
The sample from the CERGRENE column represents 1/3 of the total volume sent to the
column. The Long column samples only represent 1/280 of the fluid pumped. The Long column
represented 1/3 of the CSO pumped to the mixing basin while the series of CERGRENE column
experiments represented less than 1/20. The Long column reduces error due to the increased
volume being settled and the larger number of SS samples. The reliance of the CERGRENE
method on smaller settling volumes, fewer samples and increased size of the CERGRENE
samples in comparison to the rest of the column may increase the possibility of error.
As previously mentioned, the ball valve in the design of the CERGRENE column
probably inhibits settling (and filling) in the bottom chamber by intercepting near-wall settleable
material. Evaluations of other types of valves should be conducted to attempt to minimize
interference of the valve mechanism on solids settling. A valve designed on the workings of a
camera shutter could function as an alternative to the current off-the-shelf ball valve. Old
fashioned camera curtain or peephole shutters operate by a series of very thin metal leaves
opening, then closing to expose the film, with the curtain shutter moving from one side to
another, like a curtain and the peephole shutter, opening and closing radially, leaving a peephole,
as the names imply. The ball valve is designed for pressurized flow which is unecessary in the
current application. A shutter valve would have its own mechanical problems like leakage or
6-4
-------
diffusion, rates that could be measured and accounted for in calculations to offset concentration
changes. A shutter mechanism would be easier to automate than a ball valve. Figure 6-1A
addresses some of these and the other recommendations mentioned above. Figure 6-IB presents
an alternative configuration for the bottom ball valve to minimize headloss during filling and
draining for sample analysis.
A.
Top ball valve can remain,
should not affect setting
Another valve configuration could be
used, i.e., camera shutter design or
gate valve with plastic or teflon
components. Ball valves are
designed for pressurized flow. Since
there is no flow when the valve is
used, this configuration only
warrants valve supporting < 1 m of
head.
to vacuum
top valve
clear PVC
65 mm inner diameter
center ball valve
Lower center valve to reduce sample
volume; this may ease analysis.
from mixing basin
1 m
bottom
valve
The current bottom 3-way ball valve causes head losses due to right angle
to flow path slowing influent and effluent and possibly settling out heavier
particles. An Iternative configuration is pictured in the next figure.
Figure 6-1A Suggested Modifications to the CERGRENE Column Design
-5
-------
B.
column
Y-fitting
ball valve
effluent
sample bottle
I t
influent
A Y-fitting configuration with separate valves attached
may yield flow paths with lower head losses.
Figure 6-IB Suggested Modifications to the CERGRENE Column Design
The CERGRENE columns should be attached to a more permanent mounting to
minimize movement of the columns and keep them erect during the settling portion of the
experiment. However, a permanent mount might negate some positive effects of "the ease of
use" in the field, transport and cleaning requirements. Without a proper mounting, the
CERGRENE column, which is currently made of off-the shelf components, may not be robust
enough for field work.
The CERGRENE columns appear to be behaving as independent events. Test developed
around multiple settling devices need a higher level of precision in construction as results
between columns are compared and need to be clearly identified to trace problems or
malfunctions. An evaluation of the CERGRENE column's performance may also not be
complete without specifications of peripheral equipment, i.e., vacuum pump, tubing, sample
bottle size. In general, a much more rigorous treatment of fluid dynamics should be performed
on the current design and any future modifications of the CERGRENE column. Multiple
evaluations of the CERGRENE columns without the use of the same influent flowrate, intake
sampling depth, and tubing size could influence results so much as to obscure any direct
comparisons that could be made for an already statistically variable measurement such as CSO.
Because each storm event is an independent event with varying concentrations of suspended and
settleable solids and variable flowrates in a combined sewer, it may be difficult to produce
repeatable results with CSO.
Development of a totally automated sampling system might reduce labor costs (initial
deployment, removal, and transfer of samples for analysis), but it would significantly increase
capital cost. Standard Methods currently does not have standardized settleability test
requirements. Therefore laboratories and consultants can use low cost or in-house equipment to
devise sampling plans. An automated settling sampler would require more sophistication than is
currently available from a typical autosampler, and would therefore retail at a much higher price.
6-6
-------
7. REFERENCES
Aiguier, E., G. Chebbo, J. L. Bertrand-Krajewski, P. Hedges, and J. N. Tyack. "Methods for
Determining the Settling Velocity Profiles of Solids in Storm Sewage." Proceedings of
lAWQ/IAHR's International Conference on Sewer Solids - Characteristics, Movement, Effects
and Control, University of Abertay Dundee, Scotland, 1995.
Camp, T. R. "Sedimentation and the Design of Settling Tanks." In Proceedings American
Society of Civil Engineers, Paper No. 2285, p. 895 - 958 April, 1945.
Chadirat, S., E.Quouillot, J. Tradwell, E. Lucas-Aiguier, and G. Chebbo. Mise aupoint d'un
protocole de me sure de la repartition de la pollution par classes de vitesses de chute des solides
en suspension dans les effluents urbains de temps sec et de temps de pluie. Centre
d'Enseignement et de Recherche pour la Gestion des Ressources Naturelles et de
1'Environnement (CERGRENE), France, 1997.
Champigny, M., B. Gagne, M.. Couture and J. Cigana. Assessment of Known Practices:
Sampling Methods. Internal report for John Meunier, Inc., Montreal, Quebec, Canada, 1997.
Chebbo, G., P. Musquere, V. Milisie and A. Bahoc. Characterization of Solids Transferred into
Sewer Trunks During Wet Weather. Water Science Technology, 22, (10/11): 231-238 (1990).
Chebbo, G., A. Bachoc, D. Laplace, and B. Leguennec. The Transfer of Solids in Combined
Sewer Networks. Water Science and Technology, 31 (7): 95-105 (1995).
Daugherty, R. L., J. B. Frazini, and E. J. Finnemore (1985) Fluid Mechanics With
Engineering Applications, McGraw-Hill, Inc., ISBN 0-07-015441-4.
Dalrymple, R,J., S. L. Hodd, and D. C. Morin. Physical and Settling Characteristics of
Particulates in Storm and Sanitary Wastewaters, EPA-670/2-75-011 (NTIS PB 242 001),
Cincinnati, OH - Edison, NJ: U.S. Environmental Protection Agency, 1975.
Dickey, D. S. and R.R. Hemrajani. Recipes for Fluid Mixing. Journal of Chemical Engineering,
March 1992, pp. 82-89 (1992).
EPA. Combined Sewer Overflows - Guidance for Nine Minimum Controls, EPA 832-B-95-003,
Washington, D.C. : U.S. Environmental Protection Agency, 1995.
EPA. Combined Sewer Overflows - Guidance for Long-Term Control, EPA 832-B-95-002,
Washington, D.C. : U.S. Environmental Protection Agency, 1995.
Eckenfelder, W. W., Jr. "Sedimentation" in Industrial Water Pollution Control. McGraw-Hill
Book Company, 1966, 28-51.
Etchells, A.W., R.R. Hemrajani, D.J. Koestler, and E.L. Paul. The many faces of mixing.
Journal of Chemical Engineering, pp. 92-94, March 1992.
Fischer, D. "Vapor Transport of TCE from Ground Water into Residence Basements: Model
Experiments," PhD Dissertation, Rutgers University, New Brunswick, NJ, October, 1995.
7-7
-------
Gagne, B. and J. Bordeleau (1996). "Vitesse de chute des particules Essais Comparatifs des
methodes allemande et americaine", Fevrier 1996, CEGEO, pp 46.
Lager, J. A., W. G. Smith, W. G. Lynard, R. M. Finn, and E. J. Finnemore. Urban Stormwater
Management and Technology: Update and Users' Guide. EPA-600/8-77-014 (NTIS PB 275-
654), U.S. Environmental Protection Agency, Edison, NJ, 1977.
Lucas-Aiguier, E., J. Bertrand-Krajewski, and G. Chebbo (1997). "Protocole Victor de Mesure
de la Distribution des Polluants en Fonction des Vitesses de Chute dans Les Effluents Urbans"
CERGRENE, Agence de 1'Eau Seine Normandie (In French).
Lygren, E. and T. Damhaug (1986). "The Swirl Concentrator as an Urban Runoff Treatment
Device", In Proceedings of the NATO Advanced Research Workshop on "Urban Runoff
Pollution. " August 26-30 1985 Montpeliier, France, pp.713 - 724. Ed. by H. C. Torno et. al.,
Springer-Verlag Berlin Heidelberg.
Michelbach, S. and C. Wohrle. Settleable Solids in a Combined Sewer System, Settling
Characteristics, Heavy Metals, Efficiency of Storm Water Tanks, Water Science Technology. 27,
(5-6):153 - 164(1993).
O'Connor, T., J. Cigana and D. Fischer (1999a), "A Protocol for Determining WWF Settling
Velocities for Treatment Process Design Enhancement." Proceedings 26th Annual Water
Resources Planning and Management: Water Resources and the Urban Environment, ASCE,
Tempe, AZ, June 6-9, 1999.
O'Connor, T., D. Fischer, C. Fan, R. Field, J. Cigana, B. Gagne and M. Couture. Project Work
Plan and Quality Assurance Plan for Conducting Laboratory and Field Evaluation of Wet-
weather Flow Treatment Design Protocols (1998); QA addendums 1(1999) and 2 (2000).
O'Connor, T., D. Fischer, C. Fan, R. Field, J. Cigana, B. Gagne and M. Couture. (1999b)
Development and Evaluation of Bench-Scale Apparatus for Obtaining Wet Weather Flow
Sedimentation Treatment: Design and Operation Settling Velocity Data. Interim/Internal Report
(October, 1999).
Pisano, W.C., and Brombach, H. "Solids Settling Curves: Wastewater solids data can aid design
of urban runoff controls." Water Environment Technology 8 (4):27-33, (1996).
Pisano, W.C., Connick, D. J., and Aronson, G. L. Swirl andHelical Bend Regulator/
Concentrator for Storm and Combined Sewer Overflow Control. EPA-600/2-84/151 (NTIS PB
85-102 523), Cincinnati OH - Edison, NJ: U.S. Environmental Protection Agency, 1984.
Standard Methods for the Examination of Water and Wastewater (1995). Edited by A. D. Eaton,
L. S. Clesceri, and A. E. Greenburg. Published Jointly by American Public Health Association,
American Water Works Association and Water Environment Federation, 19th Edition.
Tchobanoglous, G. and F. Burton. "Physical Unit Operations" In Wastewater Engineering:
Treatment, Disposal and Reuse. Metcalf & Eddy, Inc. - 3rd Edition McGraw-Hill, Inc. 1991, 193
-300.
Tyack, J. N., P. D. Hedges, and R. P. M. Smisson. "A Device for determining the Settling
Velocity Grading of Storm Sewage." In Proceedings Sixth International Conference on Urban
Storm Drainage, Volume II pp 1805-1811, Niagra Falls, Ontario, Canada, September 12-17,
7-8
-------
1993.
Walker, D., J. Golden, D. Bingham, and E. Driscoll. Manual: Combined Sewer Overflow
Control. EPA/625/R-93/007 (NTIS PB 93-144 649), Cincinnati, OH: U.S. Environmental
Protection Agency, 1993.
7-9
-------
-------
100
QO
""x 80
Q70
Q.
c
- 50
.5
Q
re
c 90
ra 20
* 10
.1
•
• RPD+
• Sand as %
o Additional
Log. (RPD+)
o
*
« o
^^^^^^ «,
A ^^^
£ 10 100 1000 10000
SS Concentration (mg/L)
Figure A-3: Relative Percent Difference, SAND
50
Relative Standard Deviation (RPD
-50
-100
-150
-200
RPD+
RPD-
N\\x as %
• Log. (RPD-)
• Log. (RPD+)
10
100 1000
SS Concentration (mg/L)
10000
Figure A-4: Relative Percent Difference , MIX
-------
Q
0.
(V
c
o
_re
0)
S 55
re
c
re
•4-i
CO
1
OS
0)
0^
en
45
40
35
30
25
20
15
10
• RPD+
• Clay as %
0
. Log. (RPD+)
.
^V^
c^^^
^^^^
^^^^^
^^^^
^^^^ i
^^^^
_ ^^^^ I
* ^^"^^
^^*»_
^•S^A
^^^^-rfft
10 100 1000 10000
SS Concentration (mg/L)
Figure A-5: Relative Percent Difference, NESH
-------
Recycle Concentration, mg/L
d^n
400
350
300
250 -
200
•icn
* *
_*_ *.._*. * A _*_ . ^ . .
, * :•*•!* t * >
* + * * * * •
* * *
* ^
•
1 OU
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Experiment Number
Figure A-6: Recycle Concentration, All Experiments
4DD
H-UU
W ^
•D O)
•- c Qnn
o S JUU
CO c
•o o
0 4-> 9nn
•S 2 ^uu
c i:
0 C
& 8 mn
OT w I UU
5 §
n
Microsand Mixture
t
* Neshaminy t
•
Figure A-7: Recycle Concentration by Soil Type
-------
400
w "^»
!5 ?
"o ^
co c
•o o
CD +•>
^ (u
CD C
&S 100
co o
0
C/L
L/C
!
Figure A-8: Recycle Concentration (mg/1), by Order of Filling (C/L=CERGRENE-Long, L/C=Long-CERGRENE)
Suspended Solids
400
HUU
c
^^ *}C\C\
2
o 100
w | vjU
n
.
I 1
• 36 cm (14 in.)
18 cm (7 in.)
Figure A-9: Recycle Concentration by Intake Depth (height above bottom of basin)
-------
100 -,
I UU
•n "B) an
II
00 £ 60
•n o DU
® ''S
c £ 40
-------
350
Pbrt8
Pbrt7
Pbrt5
Pbrt3
20
40
time (m)
60
80
Figure B-l: Experiment 1, Long Column [Withdrawal height 36 cm, filling sequence C/L, media Neshaminy]*
concentration (mg/l)
'SOD
400 ^
L * * *
^00 l
\J\J\J ^l
900
£-\J\J
100
I UU
n
V-/ III
0.00 20.00 40.00 60.00 80.00
time (m)
Figure B-2: Experiment 1, CERGRENE (multiple columns) [36 cm, C/L, Neshaminy]
: Experimental parameters in brackets, refer to Table 3.3
-------
300
0
20
40
time (m)
60
80
Figure B-3: Experiment 2, Long Column [36 cm, C/L, Mixture]
concentration (mg/l)
4^0
4DD
"T\Jvy
^n
\j\j\j
^00 t
\j\j\j '
9^0
£.\J\J
9nn
£-\J\J
1^0
1 \J\J
mn
1 UU
^n
ou
n
•
^ ^
%
^
^
V 1 1 1 1
0.00 20.00 40.00 60.00 80.00
time (m)
Figure B-4: Experiment 2, CERGRENE (multiple columns) [36 cm, C/L, Mixture]
-------
O)
o
~03
0)
O
O
O
Port 8
Port?
A— Port 5
Ports
0
20
40
time (m)
60
80
Figure B-5: Experiment 3, Long Column [18 cm, C/L Mixture]
concentration (mg/l)
ROD
^00
\j\j\j
4DD
^
+
*TVJVJ
?nn I
OVJVJ 1
L* *
?nn
^VJVJ
100
i \j\j
n
vy 1 i i i
0.00 20.00 40.00 60.00 80.00
time (m)
Figure B-6: Experiment 3, CERGRENE (multiple columns) [18 cm, C/L Mixture]
-------
Port 8
Port?
A— Port 5
Ports
0
4 6
time (m)
8
10 12
Figure B-7: Experiment 4, Long Column [18 cm, C/L, Microsand]
^nn
\j\j\j
^ 400
c ^\j\j
c ^nn ^
Q OUU
"(S
*- 9nn
^— • ^.VJVJ
0
^ mn
^_- I VJVJ
o
0 o
•
>
•
• +
\J I I I I I I
0.00 2.00 4.00 6.00 8.00 10.00 12
time (m)
00
Figure B-8: Experiment 4, CERGRENE (multiple columns) [18 cm, C/L, Microsand]
-------
350
0 100
o
o
o
50
0
0
20
40
time (m)
6d
Ports
Port?
Ports
Ports
Figure B-9: Experiment 5, Long Column [18 cm, L/C, Neshaminy]
^00
\J\J\J
1?400
^_ ^ \j \j
*
5 ^oo
Q \J\J\J
•*3 '
2 200
_l_i ^LVJVJ
C
0
o 100
ii I ww
O
0 0
+
+ *
>
. ^
>
>
V III
0.00 20.00 40.00 60.00 80.00
time (m)
Figure B-10: Experiment 5, CERGRENE (multiple columns) [18 cm, L/C, Neshaminy]
-------
400
0
20
40
time (m)
Figure B-ll: Experiment 6, Long Column [36 cm, L/C, Neshaminy]
concentration (mg/l)
^oo
400
*-rVJVJ
^00 -
OVJVJ
900
£-\j\j
100
I UU
n
+ 4
^
^
fr
\J \ \ \
0.00 20.00 40.00 60.00 80.00
time (m)
Figure B-12: Experiment 6, CERGRENE (multiple columns) [36 cm, L/C, Neshaminy]
-------
O)
o
U— •
03
0)
O
O
O
0
20
40
time (m)
60
80
Figure B-13: Experiment 7, Long Column [18 cm, C/L, Mixture]
concentration (mg/l)
^nn
400
*T\J\J
<
^00
\J\J\J
900
£-\j\j
100
I \J\J
n
•
•* *
• •
^
vy i i i
0.00 20.00 40.00 60.00 80.00
time (m)
Figure B-14: Experiment 7, CERGRENE (multiple columns) [18 cm, C/L, Mixture]
-------
O)
o
'-I— •
03
0)
O
O
O
Port 8
Port?
A— Port 5
Ports
0
6
time (m)
8
10
12
Figure B-15: Experiment 8, Long Column [18 cm, L/C, Microsand]
concentration (mg/l)
^oo
400
*rUU
^00
\J\J\J <
900
£-\J\J
100
I \J\J
n
•
>
* t
\J 1 1 1 1 1
0.00 2.00 4.00 6.00 8.00 10.00 12.00
time (m)
Figure B-16: Experiment 8, CERGRENE (multiple columns) [18 cm, L/C, Microsand]
-------
O)
400
350
300
250
200
-£ 150
8 100
o
~03
O
O
50
0
0
20 40
time (m)
60
80
Figure B-17: Experiment 9, Long Column [18 cm, L/C, Mixture]
finn
uuu
o) ^nn
>•" OUU
Ss-x ^nn
__ tuu
O
'^3 onn
CO JUU
~c 9nn *
0 ^uu
o
a mn
^ i \j\j
o
0 o
+
4
4
» * *
>
0.00 20.00 40.00 60.00 80.00
time (m)
Figure B-18: Experiment 9, CERGRENE (multiple columns, randomized) [18 cm, L/C, Mixture]
-------
400
0
20
40
time (m)
Figure B-19: Experiment 10, Long Column [36 cm, L/C, Neshaminy]
concentration (mg/l)
600
500
400
300
200
100
0
0
*
.*
f
1 1 1
.00 20.00 40.00 60.00 80.00
time (m)
Figure B-20: Experiment 10, CERGRENE (one column) [36 cm, L/C, Neshaminy]
-------
O)
CD
o
o
o
0
20
40
time (m)
60
80
Figure B-21: Experiment 11, Long Column [36 cm, L/C, Mixture]
concentration (mg/l)
ROD
^00
bUU ^ ^
400 k ^
4uu r^
?nn J
OVJVJ |
9nn
£-\j\j
mn
I UU
n
'
vy 1 i i i
0.00 20.00 40.00 60.00 80.00
time (m)
Figure B-22: Experiment 11, CERGRENE (one column) [36 cm, L/C, Mixture]
-------
O)
0
o
o
o
Ports
Pbrt7
A—Pbrt5
Pbrt3
0
time (m)
10
15
Figure B-23: Experiment 12, Long Column [18 cm, L/C, Microsand]
RDD
VJUU
*^» ^oo
\jt \j\j\j
E
^^ 400
*T\J\J
0
•^3 ^00
+ +
4
•
^
TZ OUU i
£ I
C 900
^~ £-\j\j
0
c 100
^ I UU
O
o
n
\j i i i i i
0.00 2.00 4.00 6.00 8.00 10.00 12
time (m)
.00
Figure B-24: Experiment 12, CERGRENE (one column) [18 cm, L/C, Microsand]
-------
300
Ports
Port?
Ports
Ports
0
468
time (m)
10
12
Figure B-25: Experiment 13, Long Column [36 cm, C/L, Microsand]
concentration (mg/l)
ROD
^00
\j\j\j
4nn
*rUU
^nn ^
ouu
9nn
^.VJVJ
100
I \J\J
n
* «
* *
>
\J 1 1 1 1 1
0.00 2.00 4.00 6.00 8.00 10.00 12.00
time (m)
Figure B-26: Experiment 13, CERGRENE (one eoliimn) [36 em, C/L, Mierosandj
-------
400
^ 350
0)
E
o
i—»
03
CD
O
O
O
50
0
0
20
40
time (m)
Figure B-27: Experiment 14, Long Column [18 cm, C/L, Neshaminy]
concentration (mg/l)
RDD
^nn
\j\j\j
400
^nn
OVJVJ
?nn
^VJVJ
mn
i \j\j
n
* *
* +
\J 1 1 1
0.00 20.00 40.00 60.00 80.00
time (m)
Figure B-28: Experiment 14, CERGRENE (one column) [18 cm, C/L, Neshaminy]
-------
O)
o
'-i— •
CD
0)
O
O
O
0
8
time (m)
10
12
Figure B-29: Experiment 15, Long Column [36 cm, L/C, Microsand]
concentration (mg/l)
ROO
500
400
tuu
^00
A A ^
* ^ ^
\j\j\j
900
^.uu
100
i \j\j
n
V-/ 1 1
0.00 5.00 10.00 15.00
time (m)
Figure B-30: Experiment 15, CERGRENE (one column) [36 cm, L/C, Microsand]
-------
o
o
•-§ 200
CD
100
o
0
0
PortS
Port 7
PortS
PortS
6
time (m)
8
10
12
I400
o
MIX
Ports
Port?
Ports
Ports
20
40
time (m)
60
80
PortS
Port?
PortS
PortS
30 40
time (m)
50
60
70
Figure B-31: Average of Long Column Concentrations by Soil Type
-------
^ 400
0
height (ft)
Figure B-32: Average Concentrations for Sand, Mix, and Neshaminy at Each Port and Time Interval
-------
-------
40
*rU
^\ ^^
^J^J
!_
£ 20
(0
2 10
o
-£ o
^ u
0
^ -10
0 IU
Q.
30
C
4-
*
^ ^ ^
A
*
) 0.5 1 1.5
Overflow Rate (cm /sec)
Figure C-l: Percent Removal versus Overflow Rate, Experiment 1, Long Column [36 cm, C/L, Neshaminy]
Percent Greater
Rn
OVJ
?n
* \j
60 4
50
40
tu
^n
O\J
9n
^.u
10
1 U
n
A
v
t
t'~*4~ %
•
^
W 1 I ii
0 0.5 1 1.5 2 2.5
Overflow Rate (cm /sec)
Figure C-2: Percent Removal versus Overflow Rate, Experiment 2, Long Column [36 cm, C/L, Mixture]
-------
0
0
0.5
1
1.5
Overflow Rate (cm /sec)
Figure C-3: Percent Removal versus Overflow Rate, Experiment 3, Long Column [18 cm, C/L, Mixture]
Percent Greater
190
I ^VJ
mn
I VJVJ
ftn
OVJ
RO
VJU
40
*rVJ
90
£-\J
n
"*\ ~* , *
+
•
U n i i i i
0 0.5 1 1.5 2 2.5
Overflow Rate (cm /sec)
Figure C-4: Percent Removal versus Overflow Rate, Experiment 4, Long Column [18 cm, C/L, Microsand]
-------
0 0.5 1 1.5 2 2.5
Overflow Rate (cm /sec)
Figure C-5: Percent Removal versus Overflow Rate, Experiment 5, Long Column [18 cm, L/C, Neshaminy]
Percent Greater
70
70
RO I
uu ^
so
OVJ
40
H-U
^o
OVJ
90
£-\J
10
n
•
• •
* V < * «• *
% * ^
U n i i
0123
Overflow Rate (cm /sec)
Figure C-6: Percent Removal versus Overflow Rate, Experiment 6, Long Column [36 cm, L/C, Neshaminy]
-------
Percent Greater
80
\j\j
70^
fin
\j\j
^0
\j\j
40
^\j
^0
\j\j
?o
^. \j
10
1 W
n
t
^
4
* ^* A+* *
* * * * «« *
*4
\j ^ i i i i
0 0.5 1 1.5 2 2.5
Overflow Rate (cm/sec)
Figure C-7: Percent Removal versus Overflow Rate, Experiment 7, Long Column [18cm, C/L, Mixture]
Percent Greater
100
80
\j\j
RO
\J\J
40
*rU
90
£-\J
n
*
*
\j \ i i i i i
0 0.5 1 1.5 2 2.5 3
Overflow Rate (cm /sec)
Figure C-8: Percent Removal versus Overflow Rate, Experiment 8, Long Column [18 cm, L/C, Microsand]
-------
Percent Greater
80
OVJ i
70 T ^
60 ***
VJU
50
40
*rVJ
^0
ou
90
^.u
10
i \j
n
* v+ *
* •• *.
V
•
W n I I I I I
0 0.5 1 1.5 2 2.5 3
Overflow Rate (cm/sec)
Figure C-9: Percent Removal versus Overflow Rate, Experiment 9, Long Column [18 cm, L/C, Mixture]
-------
Percent Greater
60
0
0.5 1 1.5
Overflow Rate (cm /sec)
2.5
Figure C-10: Percent Removal versus Overflow Rate, Experiment 10, Long Column [36 cm, L/C, Neshaminy]
Percent Greater
fin
VJ W
^n
\j\j
40
T^ W
^n
O \J
20
^- W
10
1 \J
o
•
•
•
* *
w n i i i
0 0.1 0.2 0.3 0.4
Overflow Rate (cm/sec)
Figure C-ll: Percent Removal versus Overflow Rate, Experiment 10, CERGRENE Column [36 cm, L/C, Neshaminy]
-------
Percent Greater
RO
(JW
70 L
^
RO
VJU
*SO
ow
40
tu
^0
ou
90
^.w
10
I U
n
*
t ** ««* ^
* + +
+
U n i i i i
0 0.5 1 1.5 2 2.5
Overflow Rate (cm /sec)
Figure C-12: Percent Removal versus Overflow Rate, Experiment 11, Long Column [36 cm, L/C, Mixture]
60
\j\j
»- ^0
-------
a 10°
8 80
O 60
§ 40
| 20
0
0
12
Overflow Rate (cm/sec)
Figure C-14: Percent Removal versus Overflow Rate, Experiment 12, Long Column [18 cm, L/C, Microsand]
80
\j\j
^ 70
S 60
ro DU
P ^0
L. w W
0 40
| . T-W
m ^0
QJ O W
" 20
(1) ^U
°- 10
1 \J
o
A
4 ^
^
A
w
•
W ^ 1 1 1
0 0.1 0.2 0.3 0.4
Overflow Rate (cm/sec)
Figure C-15: Percent Removal versus Overflow Rate, Experiment 12, CERGRENE Column [18 cm, L/C, Microsand]
-------
Percent Greater
1?0
1 £-\J
100
80
\j\j
60
\j\j
40
^\j
?0
£-\J
n
A -- - _ A
* **** +0+ ^
4 *
V
+ ^
\J \ I I I I
0 0.5 1 1.5 2 2.5
Overflow Rate (cm/sec)
Figure C-16: Percent Removal versus Overflow Rate, Experiment 13, Long Column [36 cm, C/L, Microsand]
Percent Greater
60
\j\j
40
^0
\j\j
10
1 \J
o
*
* *
«
\J \ 1 1 1
0 0.1 0.2 0.3 0.4
Overflow Rate (cm/sec)
Figure C-17: Percent Removal versus Overflow Rate, Experiment 13, CERGRENE Column [36 cm, C/L, Microsand]
-------
a_
0
+j
OJ
0)
a_
0
•+J
c
0)
B
0)
a.
60 -,
50 -
t
40 -
^\ ^^
^1 ^J
?0
10
0
-10
^
^
^^
^
A ^
A ^
T ^ ^^ A ^
^ + i ^ ^ i ^
0 0.5 1 1.5 2 2.5
Overflow Rate (cm/sec)
Figure C-18: Percent Removal versus Overflow Rate, Experiment 14, Long Column [18 cm, C/L, Neshaminy]
100
£ 80
20
o_
0
0
0.1 0.2 0.3
Overflow Rate (cm/sec)
0.4
Figure C-19: Percent Removal versus Overflow Rate, Experiment 14, CERGRENE Column [18 cm, C/L, Neshaminy]
-------
Percent Greater
120
100
I VJVJ
80
ou
60
VJU
40
*rU
90
^.VJ
n
+**« *«• ^ ^
•
\J n i i i i
0 0.5 1 1.5 2 2.5
Overflow Rate (cm /sec)
Figure C-20: Percent Removal versus Overflow Rate, Experiment 15, Long Column [36cm, L/C, Microsand]
Percent Greater
70
i \j
60
\j\j
^0
\j\j
40
^ \j
^0
O \J
20
£. w
10
i \j
o
* * ,
I
w n i i i
0 0.1 0.2 0.3 0.4
Overflow Rate (cm/sec)
Figure C-21: Percent Removal versus Overflow Rate, Experiment 15, CERGRENE Column [36cm, L/C, Microsand]
-------
-------
Experiment 9
300
250
^ 20°
35
"§ 150
o
E
o>
K 100
50
4.00q
500 1000
1500 2000 2500
time (s)
Experiment 10
3000 3500 4000
Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs10
Velocity Percentiles
Figure D-l
3.50
3.00 --H
0.50 --H
0.00
500 1000
1500 2000 2500
time (s)
3000 3500 4000
Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs10
Velocity Percentiles
Figure D-2
-------
180 :
D)
E 140 i
T3
0
>
E 100 ;
0
o:
w
w
CD
80 :
60 -
0
200 q
180 -
160 -.
O)
E 140 -.
>
o
0
o:
ro
100 T
80 i
40 -
Experiment 11 (all points)
10.00 q
500 1000
1500 2000 2500
time (s)
3000 3500 4000
\/s90 \/s80 \/s70 \/s60 \/s50 \/s40
Velocity Percentiles
\/s20
Figure D-3
Experiment 11 (with suppression)
10.00 q
100 200 300 400
time (s)
500
600
700
Figure D-4
Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs10
Velocity Percentiles
-------
Experiment 12
100 200 300 400
time (s)
Experiment 13
500
600
9.00 q
700
Figure D-5
Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs10
Velocity Percentiles
9.00 q
zou
onn -
zuu
en
E, !
"m -i^n -
CD loll
o
E
CD
fV *i nn -
u- 1UU
w
w
CD
^
en -
t)U
0_
n
c
)
<
r-
/
1C
^
)0
^
^
4
k-
2C
)0
3C
)0
t
— •
irr
i —
e
.— — •
(s
i — •
4C
)
.
)0
— •
L — •
I •
—
5C
—- •-
)0
I— —
-—
— — - -
-4
6C
>
)0
7C
l
)0
Fie
Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs10
Velocity Percentiles
Figure D-6
-------
Experiment 14 (all points)
en
E,
T3
CD
O
E
CD
a:
w
w
CD
300
250
200
150
100
50
0
8.00
500 1000 1500 2000 2500 3000 3500 4000 .. nn .. on .. vn .. on .. cn .. .. .. on .. on .. Hn
Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs10
Figure D-7 Velocity Percentiles
8.00 T
time (s)
Experiment 14 (with suppression)
500 1000
1500 2000 2500
time (s)
3000 3500 4000
Figure D-8
Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs10
Velocity Percentiles
-------
300
250
35
— 200
T3
0
1 150
0
a:
co
100
50
0
D)
T3
0
0
a:
CD
300
250
200
150
100
50
0
Experiment 15 (all points)
100
200
300
400
time (s)
500
600
700
Experiment 15 (with suppression)
100
200
300
400
time (s)
500
600
700
10.00 n
800
Figure D-9
Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs10
Velocity Percentiles
10.00 i
800
Figure D-10
Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs10
Velocity Percentiles
-------
-------
Suspended Solids Concentration, mg/L
100 n
on
yu
on
oU
70,
/ u
cr\
DU
50
AC\
tu
^n
OU
on
^u
m
I U
n
I "^F
. r
A r
. c
* -*-F
I A\
3ort3
3_-4 c
Jon o
3ort 7
3ort8
Ar^!^?^^>^ •
•t™^^*^ ^ — —
^C^r^ ^ — «^__^
^^^^^^
w I I I I I I
0 10 20 30 40 50 60
Time, minutes
Figure E-l: Experiment 1, Long Column, Suspended Solids
350 -i
_j
onn
c"
~ 9^,0,
2 ^3U
+•»
1 200
o
o
V)
~o Hen
O
OT
•c
ai /inn
c
Ol
= *.n
OT &U
c
/ \
/ \
/ \
/ \
/ \
/ \
/ \
/ \
/ \
/ \
/ \
/ \
_^-~J * A
) 10 20 30 40 50 60
Time, minutes
Figure E-2: Experiment 1, CERGRENE Column, Suspended Solids
-------
50 n
_l
O A C
o 4U
'•7
2 35
0)
O on
J; OU
0
O 0(-
(/) ^3
;o
CO
"S -IK.
Q) IO
•o
0) -|Q
I/)
= C
OT O
\
C
._, , „
^^ Port 3
T , r->-*-L c
A rort o
• • pnrt 7
nj • ^-^ '/^~ "~~~-^/ » ~~~— — -___ ~~~^-~~^_
^L^f^y7^^^^^^^^^^^-^^^
V ^^^^--_^^^^^=^*
^^
i i i i i i
) 10 20 30 40 50 60
Time, minutes
Figure E-3: Experiment 1, Long Column, Volatile Suspended Solids
Volatile Suspended Solids Concentration, mg/L
50 -,
45
40 -
35 -
i
30 -
25
20
15
10 -
5 _
o
A,
\ / \
1 V ~ ~^
A *, '
/*( \ /
\^
0 10 20 30 40 50 60
Time, minutes
Figure E-4: Experiment 1, CERGRENE Column, Volatile Suspended Solids
-------
Suspended Solids Concentration, mg/L
200 n
180 1
1
1RD » -"
^1
r l ^» --
140 > *tv^_
19D
[
100
an
RO
AT\
on
n
' VT,--.,-:"' ><
\ — ~~~~~>^^ ~=^-_
V" ^^
0
• Port3
A Ports
» Port 7
• — Ports
x Port 2
o Port 3
VSS
n Port 7
VSS
0 10 20 30 40 50 60 70
Time, minutes
Figure E-5: Experiment 2, Long Column, Suspended Solids (with several Volatile Suspended Solid measurements)
600 n
_J
"Si
E 500
c"
.0
"S /inn
ds Concentn
O CO J
3 § S
O i
(0
"5 -inn >
•o
c
0)
«" 0
—^- SS mg/L
A x VSS mg/L
T
L x
WO 10 20 30 40 50 60 70
Time, minutes
Figure E-6: Experiment 2, CERGRENE Column, Suspended Solids (with two Volatile Suspended Solid measurements)
-------
1
"Si
E
c"
o
?
2
+-
0)
u
c
o
0
-------
_J
c"
o
+J
2
+*
0)
u
c
o
o
I/)
•o
"o
V)
I
2Q5 n
255
250
245
240 -
235
230
225
220
/
/
/
/
/
.A, /
/ •-, /
/ \ /
/ \ /
^A / \A/
\ S
\ /
A /
^ /
\ /
\ /
\ /
V
0 5 10 15 20 25 30 35
Time, minutes
Figure E-9: Experiment 3, CERGRENE Column, Total Solids
Suspended Solids Concentration, mg/L
60 n
en
OU
/in
'tU
on
OU
on
ZU
m
I U
n
-+- Port 3
V-_^^
^^^*_^^
^^~~^»
\j i i i i i i
0 10 20 30 40 50 60
Time, minutes
Figure E-10: Experiment 1, Long Column, Suspended Solids, Port 3 Only
-------
-------
Chapter A
0)
•4-i
re
£
O
•4-i
0)
0)
Q.
0.01
A Suspended Solids Removal
y = -6.1813Ln(x) + 12.304
R2 = 0.2386
0.1 1
Overflow Rate, cm/sec
10
Figure F-l: Experiment 1, Percent Removal vs. Overflow Rate, Long Column, Suspended Solids
250 n
200
£ 150
1
e>
•g 100
8
0)
°- 50
n
"^n
A Suspended Solids
Removal
A
A y = -25.138Ln(x)-41.813
R2 = 0.1154
^^
A A A A^*^»_ A.
A A A ^^
0.001 0.01 0.1 1
Overflow Rate, cm/sec
Figure F-2: Experiment 1, Percent Removal vs. Overflow Rate, CERGRENE Column, Suspended Solids
-------
0)
•4-i
re
O
•4-i
0)
0)
Q.
y
60 n
en
AC\
on
on
m
1 0
on
on
AC\
A Volatile Suspended Solids Removal
A A
A
^^^^ A A
A?"^^^A A A
A AA^-^^A
A ^^^^^
A
A
A
0.01 0.1 1 10
Overflow Rate, cm/sec
= -8.7358Ln(x) + 7.5897
R2 = 0.3838
Figure F-3: Experiment 1, Percent Removal vs. Overflow Rate, Long Column, Volatile Suspended Solids
0)
ra
n
r
s
0)
Q.
A Volatile Suspended Solids Removal
3Q y = -4.5965Ln(x)-6.1998
25 -
20 -
15
10 -
5
o
-5 -
-10
R2 = 0.3352
A
^^ A
A ^^--^i
"^^^^
A ^s-^>^
r*^^^
^"""•s^^ A
^^^^^
A
A
0.001 0.01 0.1 1
Overflow Rate, cm/sec
Figure F-4: Experiment 1, Percent Removal vs. Overflow Rate, CERGRENE Column, Volatile Suspended Solids
-------
50
40
o> 30
•4-1
re
£
O)
"c
d>
£
£ 10
-10
A Suspended Solids Removal
y = -5.054Ln(x) + 12.086
R2 = 0.5523
0.01
0.1 1 10
Overflow rate, crrVsec
100
Figure F-5: Experiment 2, Percent Removal vs. Overflow Rate, Long Column, Suspended Solids
Percent Greater
60 -,
50 -
40
30
20 -
10 -
0 -
A Suspended Solids Removal
T1 sample ommited as it inverted slope
A y = -6.0926Ln(x) + 4.3457
R2 = 0.3097
^^ A
^**^^^ A
^^V^^
0.001 0.01 0.1 1
Overflow Rate, cm/sec
Figure F-6: Experiment 2, Percent Removal vs. Overflow Rate, CERGRENE Column, Suspended Solids
-------
Percent greater
50 n
4D
"30
on
m
n
m
0.
-20
A Total Solids Removal
A
A y = -3.9525Ln(x) + 7.562 A
A R2 = 0.2995
A
-AAA A
^»-^. A
A \AA
^^^^ A
«A ""---^
A ^^^^
A A A ^^\ A
\ A i ^"^-^
^*AA A ^-^^^
A ^-^^
A ^^-^A
_
01 0.1 1 10 A 100
Overflow rate, cm/sec
Figure F-7: Experiment 2, Percent Removal vs. Overflow Rate, Long Column, Total Solids
k.
Pi
Si
£:
e>
^-i
S
0)
Q.
12 n
10
8
6
4
2
0
-2
y = -1.8129Ln(x) + 0.2306 A Total Solids Removal
R2 = 04681A T5 omitted
* A
^
v^
,_ A
A "<. A
^^V.
^^^^
^Sw^
A ^S^
^^^
A
I I I
0.001 0.01 0.1 1
Overflow Rate, cm/sec
Figure F-8: Experiment 2, Percent Removal vs. Overflow Rate, CERGRENE Column, Total Solids
-------
0) i
15 2
£
e>
E o
0)
e
0)
Q.
-2
y = -1.499Ln(x)-4.8659
R2 = 0.3173
0.01
0.1
Overflow Rate, cm/sec
A Total Solids Removal
Figure F-9: Experiment 3, Percent Removal vs. Overflow Rate, CERGRENE Column, Total Solids
Percent Greater
4 n
n
0
_4
s
m
14
A Suspended Solids Removal
A
^^V^^ A
A ^^^^
^\^
A
A
0.001 0.01 0.1 1
y = -2.4439Ln(x) - 10.874 Overflow Rate, cm/sec
R2 = 0.3255
Figure F-10: Experiment 1, Percent Removal vs. Overflow Rate, CERGRENE Column, Suspended Solids,
2 Data Points Removed
-------
-------
Figure G-l Victor Matrix Output for Experiment 1, Phase III, Suspended Solids (Top Row All Data, Bottom Row with Suppression)
250
200
| 150
o
E
0
^ 100
50
3000
2000
1000
0
S
c -1000
-2000
500 1000
1500 2000 2500
Time (s)
3000 3500 4000
-3000
0.001 0.010 0.100 1.000
Settling Velocity (mm/s)
10.000
7.00
100.000
6.00 - -
| 5.00 - -
Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs10
Velocity Percentiles
200 -
"en
^T isn -
Remove
I
I/)
CD
^
50 -
0 500 1 000 1 500 2000 2500 3000 3500 40
Time (s)
~a
C£
$>
0
o
0
°- 20 -
0 -
•u
••a
-
TT
*'
S
'
77 7.00 ;
$ 5.00 :
>• 4.00 -
'o
o
o !
en
I 2.00 ;
00 u-uu H
0.001 0.010 0.100 1.000 10.000 100.000
Settling Velocity (mm/s)
—
Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs10
Velocity Percentile
Note: Background suspended solids concentration used instead of to concentation for Analysis with Suppresion.
-------
Figure G-2 Victor Matrix Output for Experiment 1, Phase III, Volatile Suspended Solids (Top Row All Data, Bottom Row with Suppression)
500 1000
1500 2000 2500
Time (s)
250
. 200
150
100
3000 3500 4000
1.40
0.001
0.010
0.100 1.000
Settling Velocity (mm/s)
10.000
1.20 --
ST 1.00
E.
4-0.80
8
o
> 0.60
0.20 - -
0.00
100.000
Vs90 Vs80 Vs70
Vs60 Vs50 Vs40
Velocity Percentiles
Vs30 Vs20 Vs10
"55
E :
1 J
0
&- :
CD
-2 -
/
f
/
f-
/
f
'
/
/
t
,
| 80 -
0
^ 60 '
^
CD
S
20 -
n 5nn 1 nnn 1 snn ?nnn 95nn snnn ssnn annn
x
x
'
X
/
/
*
~" 1.40;,
S 0.80 -
^ 8
0
> 0.60 -
en
1 0.40 ;
0.20 ;
"^ 0.00 ^1
-
Time ^
Settling Velocity (mm/s)
Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs10
Velocity Percentile
-------
Figure G-3 Victor Matrix Output for Experiment 2, Phase III, Suspended Solids (Top Row All Data, Bottom Row with Suppression)
"S an '
o '
0 __ •
in
in
CD
Q
«
\
N/
/•
r
s
<
/
/
/
(
s
s
/
<
s
k
s
N
k
S
s
s
N
k
N
S
N
^
S
s
^
g •
T3
0
I 100 -
0
cc
&
§ 50 '
0
O
0
Q- o -
\
\
\
\
\
1
1
1
1
1
/
Jl
VI
i MI
i MI
1
ip
i
i ii
1 1 1
1 1 1
1 1 1
V
2.50 -
5f
| 2.00 -
1 1'50 "
I 1.00
0
ro
w 0.50
0 500 1000 1500 2000 2500 3000 3500 4000 Qm ^ Qm ^ ,„„„, mQOQ
Time (S) Settling Velocity (mm/s)
inn 3 nn -,
140 -
>
o
o __ -
CD
4
/
/
f
/-
S
f
x
X
S
s
/
'
<
80 :
^ 60-
E 40 "
a:
^
CD 20 -
^ :
0
1
1
1 1
1-J
/
1
\
\
\
V
y —
/
| 2.00 •
>.
o 1.50 •
o
CT
-u 0'50 "
n nn '.
0 500 1 000 1 500 2000 2500 3000 3500 4000 0.001 0.01 0 0. 1 00 1 .000 1 0.000 1 00.000
Time (s) Settling Velocity (mm/s)
Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs10
Velocity Percentile
r— |
i—i
—
n n n n
Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs10
Velocity Percentile
-------
Figure G-4 Victor Matrix Output for Experiment 2, Phase III, Total Solids (Top Row All Data, Bottom Row with Suppression)
100 n
80 -
"3 •
£, 60 "
~a
(D
K. '
o>
TtT g 1 00 -
0
ro
n nn -
,— I
0 500 1 000 1 500 2000 2500 3000 3500 4000 0.001 0.01 0 0. 1 00 1 .000 1 0.000 1 00.000 Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs1 0
Time (s) Settling Velocity (mm/s) Velocity Percentile
I5 60 '
-a
c 40 -
0
QL '
£ on
s
0 -
20 -
4
J
f
/
1
1
1
/
-
_
_
L.
|w
*
^ 120 :
I :
0 :
QL on '
in 80 :
(f)
CD
2 60 :
s •
D- :
20 •
0 500 1000 1500 2000 2500 3000 3500 4000 Q OQ1
Time (s)
s
1
^
ill
I
\
I
::
1 1
I ' '
I ' '
\
2.50 -
^0
1 2.00 •
.£•
0
>
cn
m ^ A ^ •
S1-00:
ro
— I
ni — I
"0""""
0.010 0.100 1.000 10.000 100.000 Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs10
Settling Velocity (mm/s) Velocity Percentile
-------
Figure G-5 Victor Matrix Output for Experiment 3, Phase III, Total Solids (Top Row All Data, Bottom Row with Suppression)
20
15
„ 10
0>
| 5
1 0
o
cc
in "5
ro
-10
-15
-20
500
1000
Time (s)
1500
2000
400
300
200
100
-100
-200
0.001
0.010
0.100 1.000
Settling Velocity (mm/s)
10.000
ST ^n
£r
(D
£=
OT
nnn -
, — .
"rrrrrrrr
100.000
Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs10
Velocity Percentile
15
"5
E
T3
0
1 0
E u
0
a:
CD
-15
«*
-^
r~
i —
i —
--
,»-
,*- •
— •
r-"
^-»
<•
x--
^
^
P-^
^
0 500 1000 1500 20
Time (s)
^^ 100 -
QL '.
CD 60 -
0 -
1
/
1
1
1
S
^\
V
t
^ 5.00 -
450
400
| 3.50
o 2.50
0
ra 2.00
w
1.00
0.50
nn u-uu -
0.001 0.01 0 0. 1 00 1 .000 1 0.000 1 00.000
Settling Velocity (mm/s)
I j j — | | — j f — ] __
Vs90 Vs80 Vs70 Vs60 Vs50 Vs40 Vs30 Vs20 Vs1 0
Velocity Percentile
-------
Experiment 1: 6/9/98 Long Column Initial Height=7'10"
Long column data
Port
Time (min) Cone (mg/l)
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
0.60
0.88
1.15
1.42
1.85
2.12
2.42
2.70
3.10
3.18
3.58
3.90
5.18
5.37
5.73
6.08
10.38
10.78
11.08
11.37
58.83
59.17
59.45
59.78
256.18
292.00
289.96
327.84
274.73
268.69
n/a
284.58
258.20
267.41
276.89
268.18
244.80
275.37
255.81
255.69
224.23
269.03
268.90
n/a
171.98
295.51
324.40
305.68
CERGRENE data
Column
1
2
3
4
1
2
3
Time (min) Height(in)
0.00
1.00
3.00
5.00
10.00
60.00
14.67
16.25
16.25
16.38
16.88
17.00
15.25
16.75
Cone, (mg/l)
276.47
338.74
420.00
357.85
447.08
379.43
365.56
Blank Cone (mg/l)
3.03
Recycle Cone (mg/l)
1 267.14
2 267.97
3 265.20
Non-Settleable Solids (mg/l)
1 85.20
2 71.54
H-1
-------
Experiment 2: 6/10/98 Long Column Initial Height=7'11"
Long column data
Port
Time (min) Cone (mg/l)
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
0.42 168.01
0.68 183.27
0.85 254.47
1.03 280.40
1.38 172.03
1 .57 n/a
1.78 185.02
1.93 145.87
2.82 128.40
3.00 143.70
3.17 142.80
3.37 151.79
5.05 128.46
5.20 141.11
5.40 143.87
5.57 142.18
10.20 88.59
10.40 133.33
10.63 141.60
10.83 141.98
60.20 104.56
60.43 117.18
60.65 136.22
60.85 148.08
CERGRENE data
Column
1
2
3
4
1
2
3
Time (min) Height(in)
0.00 16.75
1.00 18.5
3.00 15.75
3.00 14.75
10.00 16.5
60.00 16.5
5.00 16.25
Cone, (mg/l)
299.07
193.94
349.95
262.27
426.87
367.27
324.39
Blank Cone (mg/l)
0.44
Recycle Cone (mg/l)
1 335.90
2 258.37
3 n/a
Non-Settleable Solids (mg/l)
1 38.15
2 25.73
H-2
-------
Experiment 3: 6/10/98 Long Column Initial Height=7'11"
Long column data
Port
Time (min) Cone (mg/l)
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
0.25
0.50
0.70
0.90
1.32
1.57
1.80
2.00
3.02
3.17
3.37
3.57
5.25
5.47
5.67
5.92
10.45
10.75
11.00
11.25
59.95
60.15
60.33
60.57
175.27
186.29
264.17
374.35
141.53
156.28
177.05
182.95
132.33
185.16
153.97
154.58
123.36
141.31
142.28
146.01
119.85
132.05
136.51
136.47
85.94
129.37
154.73
152.67
Blank Cone (mg/l)
n/a
Recycle Cone (mg/l)
1 288.89
2 220.08
3 286.38
Non-Settleable Solids (mg/l)
1 n/a
2 33.11
CERGRENE data
Column Time (min) Height(in) Cone, (mg/l)
1
2
3
4
1
2
3
0.0015.25 351.10
1.0017 238.82
3.00 16 412.69
5.00 16.25 255.59
10.00 16.5 536.23
60.00 16.75 249.68
3.00 16.88 342.37
H-3
-------
Experiment 4: 6/11/98 Long Column Initial Height=7'11"
Long column data
Port Time (min) Cone (mg/l) Blank Cone (mg/l)
1.06
Recycle Cone (mg/l)
1 156.50
2 256.76
3 207.08
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
0.38
0.52
0.75
0.97
1.35
1.53
1.72
1.88
2.58
2.83
2.98
3.45
4.88
5.02
5.18
5.35
9.97
10.17
10.35
10.55
107.39
57.41
218.10
346.80
29.23
47.64
n/a
91.20
15.23
20.00
16.27
22.14
n/a
8.24
9.41
8.94
7.20
7.02
5.87
8.05
CERGRENE data
Column Time (min) Height(in) Cone, (mg/l)
1 0.00 15.5 315.31
2 1.0015.75 116.13
3 3.00 15.5 213.62
4 5.0015.5 116.02
1 10.00 16.38 463.14
2 1.00 124.84
H-4
-------
Experiment 5: 6/11/98 Long Column Initial Height=7'11"
Long column data
Port
Time (min) Cone (mg/l)
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
0.27
0.47
0.70
0.88
1.08
1.27
1.42
1.58
2.85
2.98
3.15
3.32
4.95
5.13
5.30
5.48
11.22
11.40
11.57
11.77
59.95
60.17
60.35
60.63
193.77
263.57
312.81
292.69
242.91
262.73
248.18
277.07
228.00
248.35
256.61
261.16
209.16
226.82
248.35
#DIV/0!
170.82
221.97
240.63
247.06
140.08
221.11
255.56
266.14
Blank Cone (mg/l)
3.94
Recycle Cone (mg/l)
1 305.28
2 286.85
3 269.55
Non-Settleable Solids (mg/l)
1 56.74
2 43.35
CERGRENE data
Column Time (min) Height(in) Cone, (mg/l)
1
2
3
4
1
2
3
0.00
1.00
3.00
5.00 16.25
10.00 14.5
60.00 16
0.00 16.25
339.48
291.85
384.09
312.08
442.47
387.99
267.18
H-5
-------
Experiment 6: 6/12/98 Long Column Initial Height=7'11"
Long column data
Port
Time (min) Cone (mg/l)
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
0.20
0.42
0.58
0.72
0.93
1.13
1.32
1.52
2.90
3.07
3.22
3.38
5.00
5.17
5.33
5.48
9.90
10.07
10.25
10.42
59.95
60.08
60.25
60.47
254.96
243.51
319.48
338.27
254.03
260.25
272.65
242.03
231.75
233.72
255.19
267.18
212.08
235.12
246.69
254.55
179.92
215.29
246.30
257.14
119.69
212.70
255.34
244.62
Blank Cone (mg/l)
0.71
Recycle Cone (mg/l)
1 283.33
2 290.71
3 n/a
Non-Settleable Solids (mg/l)
1 34.62
2 49.26
CERGRENE data
Column Time (min) Height(in) Cone, (mg/l)
1
2
3
4
1
2
3
0.00 16.25
1.00 15.75
3.00 15.25
5.00 15.75
10.00 16.5
60.00 16.5
60.00 16
322.97
290.85
369.12
339.92
464.36
411.72
454.04
H-6
-------
Experiment 7: 6/12/98 Long Column Initial Height=8'
Long column data
Port
Time (min)
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
0.25
0.45
0.63
0.83
1.35
1.50
1.63
3.25
2.73
2.87
3.02
1.82
4.83
4.98
5.13
5.28
9.95
10.10
10.30
10.37
60.48
60.75
60.93
61.12
Cone (mg/l)
188.76
241.44
n/a
357.87
145.77
155.06
196.83
157.56
135.02
147.46
157.53
184.07
123.17
136.89
149.80
137.08
114.02
129.96
139.26
150.20
86.94
129.59
155.98
159.07
Blank Cone (mg/l)
1.07
1.08
Recycle Cone (mg/l)
1 298.74
2 327.46
3 257.99
Non-Settleable Solids (mg/l)
1 27.31
2 26.78
CERGRENE data
Column Time (min) Height(in) Cone, (mg/l)
1 0.00 15.25 347.42
2 1.00 15.75 216.54
3 3.00 15.50 359.04
4 5.00 16.25 259.15
1 10.0016.25 467.11
2 60.00 16.5 256.56
3 10.00 15.75 327.98
H-7
-------
Experiment 8: 6/14/98 Long Column Initial Height=7'11.6"
Long column data
Port Time (min) Cone (mg/l) Blank Cone (mg/l)
3.38
Recycle Cone (mg/l)
1 290.98
2 238.26
3 310.80
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
Time (min)
0.22
0.35
0.50
0.62
1.07
1.22
1.35
1.52
2.87
3.00
3.15
3.32
5.00
5.15
5.30
5.47
9.80
9.95
10.10
10.23
Cone (mg/l)
60.22
153.90
388.99
n/a
n/a
78.54
130.36
120.44
9.96
25.61
23.18
37.04
6.74
7.17
10.61
16.05
4.58
5.91
7.09
3.76
CERGRENE data
Column Time (min) Height(in) Cone, (mg/l)
1 0.0015.25 291.05
2 1.0016.125 160.11
3 3.00 15.5 197.34
4 5.00 16 143.87
1 10.00 15.5 445.89
2 5.00 16.25 152.55
H-8
-------
Experiment 9: 6/15/98 Long Column Initial Height=7'11"
Long column data
Port
Time (min) Cone (mg/l)
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
0.20
0.35
0.50
0.70
1.15
1.27
1.38
1.53
2.80
2.93
3.07
3.22
4.80
4.95
5.08
5.27
9.72
9.90
10.05
10.22
59.92
60.10
60.28
60.47
157.87
189.24
329.32
345.31
126.80
155.65
145.53
201.98
93.42
128.95
137.36
144.62
94.09
114.45
128.90
116.97
82.61
105.62
118.50
118.22
63.71
111.11
142.05
136.36
Blank Cone (mg/l)
0.77
1.79
Recycle Cone (mg/l)
1 257.66
2 235.42
3 282.44
Non-Settleable Solids (mg/l)
1 24.62
2 21.84
CERGRENE data
Column Time (min) Height(in) Cone, (mg/l)
2
4
1
3
2
4
1
0.00 16.25
1.00 17
3.00
5.00 16.25
10.00 16
60.00 16.75
3.00 16
206.04
265.51
485.70
308.30
255.21
279.57
447.19
H-9
-------
Experiment 10: 6/15/98 Long Column Initial Height=7'11"
Long column data
Port
Time (min) Cone (mg/l)
3
5
7
8
3
5
7
8
3
5
7
8
3 n/a
5
7
8
3
5
7
8
3
5
7
8
0.25
0.42
0.58
0.77
1.10
1.30
1.47
1.63
2.83
2.97
3.12
3.27
4.88
5.05
5.23
9.87
10.07
10.15
10.32
59.80
60.00
60.18
60.33
260.34
302.07
327.98
347.62
257.92
285.56
276.47
291.76
229.20
260.59
276.15
272.16
n/a
246.72
256.32
264.57
187.68
236.80
258.40
257.33
131.82
212.00
286.82
n/a
Blank Cone (mg/l)
0.80
Recycle Cone (mg/l)
1 287.92
2 298.82
3 267.19
Non-Settleable Solids (mg/l)
1 48.05
2 46.56
CERGRENE data
Column Time (min) Height(in) Cone, (mg/l)
1 0.00 17.00 284.23
1 1.0014.5 329.89
1 3.00 16.50 391.54
1 5.00 422.62
1 10.00 16.75 454.89
1 1.0016.5 328.06
1 60.00 16.25 487.68
H-10
-------
Experiment 11: 6/16/98 Long Column Initial Height=7'11.5"
Long column data
Port
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
Time (min)
0.25
0.42
0.58
0.73
1.20
1.33
1.47
1.63
2.77
2.88
3.03
3.18
4.75
4.92
5.07
5.22
9.72
9.87
10.05
10.22
59.45
59.62
59.77
59.95
Cone (mg/l)
177.26
n/a
305.33
356.09
144.72
169.35
168.08
205.14
133.05
142.26
156.30
160.23
112.60
130.68
137.36
139.53
103.77
116.72
131.13
134.69
79.10
118.08
138.31
144.65
CERGRENE data
Column
1
1
1
1
1
1
1
Time (min) Height(in) Cone, (mg/l)
0.00
1.00
3.00
5.00
10.00
0.00
60.00
17
16.75
16.5
15.5
17
16.38
290.21
413.54
381.56
429.53
487.29
290.94
469.90
Blank Cone (mg/l)
-0.38
Recycle Cone (mg/l)
1 237.24
2 239.32
3 276.47
Non-Settleable Solids (mg/l)
1 30.42
2 28.99
H-11
-------
Experiment 12: 6/16/98 Long Column Initial Height=7'11.6"
Long column data
Port Time (min) Cone (mg/l) Blank Cone (mg/l)
3.57
Recycle Cone (mg/l)
1 302.26
2 274.27
3 299.62
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
0.23
0.37
0.50
0.65
0.95
1.08
1.23
1.37
2.65
2.80
2.97
3.13
5.30
5.43
5.57
5.70
9.78
9.92
10.05
10.22
62.64
152.96
291.61
354.75
26.29
68.06
113.62
138.08
15.98
10.11
35.88
48.69
10.40
9.09
11.67
13.85
8.08
6.44
5.15
11.35
CERGRENE data
Column Time (min) Height(in) Cone, (mg/l)
1 0.00 269.22
1 1.0016.5 453.78
1 3.00 16.75 533.85
1 5.00 17 529.42
1 10.00 16.75 490.94
1 10.00 17 392.67
H-12
-------
Experiment 13: 6/17/98 Long Column Initial Height=7'11.5"
Long column data
Port Time (min) Cone (mg/l) Blank Cone (mg/l)
3.96
Recycle Cone (mg/l)
1 254.29
2 304.02
3 n/a
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
0.48
0.62
0.75
0.90
1.27
1.40
1.53
1.67
2.85
3.00
3.20
3.37
4.90
5.05
5.18
5.37
10.13
10.30
10.42
10.57
50.00
123.08
166.28
253.72
24.24
54.51
98.80
110.98
8.63
3.40
n/a
43.56
5.90
7.63
9.62
10.64
3.18
5.56
4.12
8.40
CERGRENE data
Column Time (min) Height(in) Cone, (mg/l)
1 0.00
1 1.00
1 3.00
1 5.00
1 5.00
1 10.00
16.00
16.50
17.00
16.50
16.75
310.00
478.76
518.55
455.03
516.27
553.19
H-13
-------
Experiment 14: 6/17/98 Long Column Initial Height=8'
Long column data
Port
Time (min) Cone (mg/l)
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
0.23
0.38
0.52
0.72
1.23
1.40
1.53
1.68
2.82
3.03
3.18
3.33
4.87
5.02
5.15
5.28
10.22
10.38
10.57
10.75
59.47
59.67
59.83
59.97
270.33
283.14
331.52
333.86
264.20
282.87
n/a
299.23
258.61
268.60
293.49
292.28
236.96
264.37
278.60
284.56
217.99
253.01
282.73
294.80
138.24
227.84
273.91
291.97
Blank Cone (mg/l)
0.78
Recycle Cone (mg/l)
1 285.28
2 299.56
3 n/a
Non-Settleable Solids (mg/l)
1 46.69
2 56.81
CERGRENE data
Column Time (min) Height(in) Cone, (mg/l)
1 0.00 262.09
1 1.00 16.50 384.19
1 3.00 16.50 458.46
1 5.00 17.25 509.18
1 3.00 16.50 408.54
1 10.00 17.00 469.89
1 60.00 541.58
H-14
-------
Experiment 15: 6/17/98 Long Column Initial Height=8'
Long column data
Port Time (min) Cone (mg/l) Blank Cone (mg/l)
0.40
Recycle Cone (mg/l)
1 235.81
2 275.83
3 282.84
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
0.20
0.38
0.52
0.68
1.15
1.35
1.48
1.63
2.80
2.93
3.07
3.20
4.78
4.92
5.05
5.20
9.87
10.03
10.17
10.35
55.47
167.86
345.45
386.69
31.18
44.85
103.45
86.17
16.41
21.88
38.19
46.85
12.06
12.32
8.05
19.49
10.73
7.98
5.19
12.20
CERGRENE data
Column Time (min) Height(in) Cone, (mg/l)
1 0.00
1 1.00
1 3.00
1 5.00
1 1.00
1 12.00
16.50
17.00
16.00
16.75
269.33
492.50
470.94
493.68
482.38
517.29
H-15
-------
Experiment 1: 3/17/2000
Long
Port
column data SS VSS
Time (min) mg/L mg/L
3 0.67 Void Void
5 0.83 52.12 28.48
7 0.92 46.21 36.36
8 1.17 89.13 41.30
3 1.53 55.97 46.54
5 1 .82 Void Void
7 1.88 45.35 27.91
8 2.12 47.09 30.69
3 3.02 54.55 22.73
5 3.28 43.40 25.79
7 3.57 48.95 29.47
8 3.67 44.00 24.50
3 5.65 Void Void
5 5.90 43.87 27.10
7 6.02 Void Void
8 6.17 55.00 29.00
3 9.38 53.13 30.47
5 9.58 47.53 27.78
7 9.80 47.79 22.79
8 10.03 48.03 30.92
3 14.85 Void Void
5 15.00 50.28 28.81
7 15.13 53.50 34.00
8 15.28 60.00 31.00
3 31.08 43.02 20.93
5 31.18 46.67 27.33
7 31.32 47.14 27.86
8 31.45 45.20 25.99
3 59.22 32.77 18.08
5 59.38 35.83 18.33
7 59.57 50.00 23.00
8 59.73 38.85 22.29
CERGRENE data* SS VSS
Time (min) mg/L mg/L
0 68.25 33.33
1.5 64.41 35.03
3 53.33 30.37
3 66.17 35.34
5 59.56 31.62
7 68.24 42.57
10 63.12 36.88
15 289.33 46.00
20 273.00 40.00
30 70.83 38.33
60 68.97 40.00
Long Column Initial Height=7'11"
Field Blank
SS VSS
mg/L mg/L
Void Void
Background
SS
mg/L
46.00
49.66
60.00
Void
Start
Start
Start
End
End
VSS
mg/L
26.00
30.87
32.00
Void
55.86 22.07
* Values supressed in VICTOR are bolded;
backround value used in VICTOR is italicized,.
-------
Experiment 2: 3/28/2000
Long
Port
column data
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
3
5
7
8
2
3
5
7
8
Time (min)
0.02
0.06
0.08
0.11
0.66
0.82
0.92
1.10
2.80
2.97
3.18
3.37
4.78
4.95
5.13
5.28
9.53
9.70
9.92
10.08
14.98
15.15
15.37
15.57
29.38
29.60
29.85
30.02
55.15
60.48
60.65
60.83
61.15
CERGRENE data*
Time (min)
0
1
3
3
5
7
10
15
20
30
60
SS
mg/L
165.00
171.00
180.00
151.00
172.00
168.00
161.00
156.00
155.00
155.00
131.00
151.00
145.00
153.00
157.00
148.00
127.00
126.00
140.00
139.00
91.00
119.00
126.00
124.00
119.00
121.00
120.00
111.00
127.00
103.00
95.00
165.00
97.00
SS
mg/L
167
535
213
214
207
202
243
232
261
293
219
VSS TS
mg/L mg/L
518.13
521.94
109.00 525.71
546.56
502.27
457.24
495.8
489.44
537.66
535.06
517.93
522.92
598.12
493.17
500
500.6
526.32
-46.00 482.16
491.76
484.44
496.47
472.64
474.83
492.63
438.01
477.44
453.05
9652.29
474.68
83.00 430.23
434.08
110.00 481.46
484.12
VSS TS
mg/L mg/L
112 554
550
601
578
867
609
643
605
646
614
155 609
Long Column Initial Height=7'10.7"
Field Blank
SS VSS TS
mg/L mg/L mg/L
95 542.8571
Background
SS VSS TS
mg/L mg/L mg/L
Start 200.00 606.62
Start 157.00 518.03
End 161.00 496.57
End 156.00 482.81
Extreme not used in graphs
* Values supressed in VICTOR are bolded.
I-2
-------
CERGRENE
data*
Time (min)
0
1
3
3
5
10
15
20
30
TS
mg/L
241.09
240.41
232.18
231.93
221.57
237.15
248.38
239.40
260.16
X
X
X
X
X
Experiment 3: 6/9/2000
Field Blank
TS
mg/L
244.21
Background
TS
mg/L
Start 275.77
Start 303.93
* VICTOR Supressed in bold Start 301.10
Long column - no samples collected for experiment 3.
I-3
-------
COULTER& LS Particle Size Analyzer
27 Jan 1999
File name:
Sample ID:
Optical model:
LS230
tomsand.$01
sand from
Fraunhofer PIDS included
Small Volume Module
Group ID: tomsand
Run number: 1
File name:
Sample ID:
Optical model:
LS230
tomsand.$02
sand from
Fraunhofer PIDS included
Small Volume Module
Group ID: tomsand
Run number: 2
Cumulative < Volume
100
200
300
tomsand.$01
- tomsand.$02
400
500
ASTM
600
700
800
900
1000
Size
38
45
53
63
75
90
106
125
150
180
212
250
300
355
425
500
600
710
850
ASTM
999
400
325
270
230
200
170
140
120
100
80
70
60
50
45
40
35
30
25
20
tomsand. $01
Cum. <
Volume
0
0
0
0.072
0.30
0.70
2.28
7.86
20.4
38.0
59.2
80.4
93.0
98.2
98.7
98.7
98.8
99.1
tomsand.$02
Cum. <
Volume
e
0
0
0
0
0
0.051
0.15
0.35
1.68
7.01
19.5
37.4
59.0
80.6
93.6
99.3
100.0
100
100
100
-------
COULTER® LS Particle Size Analyzer
27 Jan 1999
Size
I
1,
1,
1,
1,
2,
Jim
000
180
400
700
000
ASTM
18
16
14
12
10
tomsand. $01
Cum. <
Volume
99.5
99.8
100.0
100.0
100
100
tomsand. $02
Cum. <
Volume
100
100
100
100
100
100
-------
COULTER® LS Particle Size Analyzer
27 Jan 1999
File name:
tomsand.$01
Sample ID: sand from
Optical model: Fraunhofer PIDS included
LS 230 Small Volume Module
Group ID: tomsand
Run number: 1
File name:
Sample ID:
tomsand.$02
sand from
Group ID: tomsand
Run number: 2
Optical model: Fraunhofer PIDS included
LS 230 Small Volume Module
Differential Volume
0.04
0.1
0.2 0.4
tomsand.$01
tomsand.$02
4 6 10 20 40
Particle Diameter (lam)
100 200 400
10002000
Volume Statistics (Arithmetic)
Calculations from 0.040 pm to 2,000 lam
Volume
Mean:
Median:
D(3,2):
Mode:
<
Size lam
100.0%
247.51 am S.D.:
232.81 am C.V.:
221.01am
245.2 lam
10 25 50 75 90
156.2 188.8 232.8 284.4 336.1
tomsand. $01
108.71 am
43.9%
-------
COULTER® LS Particle Size Analyzer
27 Jan 1999
Volume Statistics (Arithmetic) tomsand.$02
Calculations from 0.040 pm to 2,000 pm
Volume 100.0%
Mean: 240.6pm S.D.: 67.12pm
Median: 233.7pm C.V.: 27.9%
D(3,2): 222.0 tarn
•Mode: 245.2 pm
%< 10 25 50 75 90
Sizepm 158.8 190.4 233.7 284.3 334.0
-------
COULTER® LS Particle Size Analyzer
27 Jan 1999
Volume Statistics (Arithmetic)
Calculations from 0.040 pm to 2,000 pm
Volume
Mean:
Median:
D(3,2):
Mode:
< 10
Size pm 158.8
100.0%
240.6 pm S.D.:
233.7 pm C.V.:
222.0 pm
245.2 pm
25 50 75
190.4 233.7 284.3
e
y-
90
334.0
tomsand.$02
67.12pm
27.9%
I
100
80
60-
40-
20-
0
Cumulative < Volume
100
200
300
tomsand.SQI
*a=~ tomsand.$02
400
500
ASTM
600
700
800
900
1000
Size
}gyi
0
38
45
53
63
75
90
106
125
150
180
212
250
300
355
425
500
600
710
850
ASTM
999
400
325
270
230
200
170
140
120
100
80
70
60
50
45
40
35
30
25
20
tomsand. $01
Cum. <
Volume
0
0
0
Pi
U
0
0.072
0.30
0.70
2.28
7.86
20.4
38.0
59.2
80.4
93.0
98.2
98.7
98.7
98.8
99.1
tomsand. $02
Cum. <
Volume
0
0
0
Pi
U
0
0.051
0.15
0.35
1.68
7.01
19.5
37.4
59.0
80.6
93.6
99.3
100.0
100
100
100
-------
COULTER® LS Particle Size Analyzer
27 Jan 1999
Size
pm
1,000
1,180
1,400
1,700
2,000
ASTM
18
16
14
12
10
tomsand. $01
Cum. <
Volume
99.5
99.8
100.0
100.0
100
100
tomsand. $02
Cum. <
Volume
100
100
100
100
100
100
-------
COULTER® LS Particle Size Analyzer
27 Jan 1999
File name: tomsand.$01
Sample ID: sand from
Optical model: Fraunhofer PIDS included
LS230
Start time:
Obscuration:
PIDS Obscur:
Fluid:
Software:
Small Volume Module
15:0827 Jan 1999
9%
34%
Water
2.11
Group ID: tomsand
Run number: 1
Run length: 60 Seconds
Firmware:
2.022.02
File name:
Sample ID:
tomsand.$02
sand from
Optical model: Fraunhofer PIDS included
LS230
Start time:
Obscuration:
PIDS Obscur:
Fluid:
Software:
Small Volume Module
15:11 27 Jan 1999
9%
29%
Water
2.11
Group ID: tomsand
Run number: 2
Run length: 60 Seconds
Firmware:
2.022.02
Differential Volume
4-
2-
0-
8-
6-
4
"•—r~T
~T
2
T
—T~
10
—T-
20
—I—
40
r~^—r
10002000
4 6
Particle Diameter (pm)
200 400
Volume Statistics (Arithmetic)
tomsand.$01
Calculations from 0.040 pm to 2,000 pm
Volume 100.0%
Mean: 247.5 pm
Median: 232.8 pm
D(3,2): 221.0 pm
Mode: 245.2 pm
S. D.:
C.V.:
108.7 pm
43.9%
Size pm
10
156.2
25
188.8
50
232.8
75
284.4
90
336.1
-------
Appendix K. CERGRENE Matrix Analysis (Victor)
Each CERGRENE column has a top section and a bottom section which can be
isolated by the central ball valve. The settling height (h) is measured from the top of the
water level to the start of the bottom section. The initial mass of particles is introduced by
vacuum aspiration into each column. If the mixing basin is homogeneous, the initial mass
of solids (Moi) should be the same for columns i to (Nf-1), Nf being the total number of
sampling points.
MOJ = Mo Vi = 1 to (Nf - 1)
By noting, MH, the mass of particles in the top section and MB, the mass of
particles in the bottom section, the following relation can be written:
MOJ = MB Of + MHoj Vi = 1 to (Nf - 1)
Also, assuming the mixing is homogeneous, the top and bottom section should be
of same initial mass
MBoj = MBo and MHoj = MHo \/i = 1 to (Nf -1)
The settling height is noted h, if the columns are identical this value should be
constant:
hj =h Vi = 1 to (Nf - 1)
At time 7=0, the initial mass, MBoj, in the bottom part of the column is equal to
the initial mass, MHoj, in the top section of the column. At any given time, t= tj and for
any column /', the central ball valve can isolate the top and bottom sections of the column.
The mass of solids found in the bottom section correspond to the initial mass in the
bottom MBoi and a fraction of settled solids Mf from the top section MHoj that
effectively settled between times 7 = 0 and t = tj.
MBj = MBoj + Mj Vi = 1 to (Nf - 1)
The accumulated masses from one column to the next, in function of time is given by
MBj - MBj_i = MBoj + Mj - MBoj_i - Mj_i Vi = 1 to (Nf - 1)
Assuming an homogeneous mixing:
MBj - MBj _ i = MJ - MJ _ i = PJ Vi = 1 to (Nf - 1)
with Pi being the mass of the settled particles from the top section between time t = tj-i
and t=ti.
Thus
/'
= MBo + Mj = MBo+ £/fc Vi = 1 to (Nf - 1)
L-1
-------
MJ=
The accumulated mass becomes:
Chebbo et al. (1992) analyzed this curve in the following fashion :
dt
where
M(f) cumulated mass of settled particles between t = 0 et t
S(t) settled particles at time t with a settling velocity greater than —
/ — — mass of settled particles at time t with a settling velocity less than —
L-2
-------
Establishment of steady flow
Derived from "Fluid Mechanics with Engineering Applications", Daugherty et al. (1985), p 454, 455
tube column
Diameter, D in. 0.75 2.5
Diameter, D m 0.01905 0.0635
Length, L m 3 0.91
Velocity, Vo, assumed to be equal to average measured velocity
Calculation of Friction
Bold = given
kinematic viscosity, mA2/s
1.003E-06
Velocity
Vo JM pump- col
Vo EPA pump - col
Vo* JM pump -tube
Vo* EPA pump -tube
Area tube
Area col.
Ratio
cm/s
11
20
122.2222
222.2222
mA2
0.003167
0.000285
11.11111
m/s
0.11
0.2
1 .222222
2.222222
D
in.
2.5
0.75
D, M
0.0635
0.0635
0.01905
0.01905
D
m
0.0635
0.01905
Reynolds
Number
R
6964.108
12662.01
23213.69
42206.71
0.000005
e/D in.
0.000024
0.00008
Friction - f
0.034592
0.029789
0.025601
0.022047
0.0015
2.36E-05
1 .24E-06
Average f
0.03219
0.023824
* calculated from continuity equation
friction, f
Equation for losses, k
k = k'+fL/d
k' (not used)
L k
0.91 0.461313
3 3.751752
Time, t, to fraction of full
tube
0.023824
L/1+K
0.622728
0.631346
velocity
column
0.03219
Solve for
Vo (m/s)
0.11
0.2
1
2.222222
( f = 0.31 6/RA0.25 for 3000 < R <
V = 0.25Vo
0.25Vo
0.0275
0.05
0.25
0.555556
100000)
, O.SVo & 0.75Vo
O.SOVo
0.055
0.1
0.5
1.111111
0.75Vo
0.0825
0.15
0.75
1 .666667
0.95Vo
0.1045
0.19
0.95
2.111111
t=L/[Vo(1 +k)]*ln[(Vo+V)/(Vo-V)]
L
m
Column 0.91
Tube 3
0.25Vo
s
2.891867
1 .590527
0.322508
0.145128
O.SOVo
s
6.219422
3.420682
0.693604
0.312122
0.75Vo
s
11.01611
6.058862
1 .228543
0.552844
0.95Vo
s
20.74002
11.40701
2.281402
1 .026631
L- 1
-------
Calcualtion of Wall Effects
from Dalrymple et al. "Physical and Settling Characteristics of Particulates
in Storm and Sanitary Wastewaters" EPA-670/2-75-011.
For Column
Vd = V*[1-(d/D)A2][1-1/2(d/D)A2]A0.5
Settling veloc cm/sec Vd
Settling veloc cm/sec V
diameter part cm*10A-4 d
diameter colu cm D
1
0.86
100
6
0.76
80
6
0.48
60
6
Threshhold
For Tubing
% difference
5% Effect
Vd for 1 Vd for 2 Vd for 3
0.859997 0.759998 0.479999
0.000347 0.000222 0.000125
Settling veloc cm/sec
Settling veloc cm/sec
diameter part cm*10A-4
diameter colu cm
Vd = V*[1-(d/D)A2][1-1/2(d/D)A2]A0.5
Vd
V
d
D
1
0.86
100
0.051
5 % Effect
Vd for 1
0.818949
4.773375
L-2
------- |