United States      Office of Research and    EPA/600/X-99/031
Environmental Protection  Development         September 1999
Agency         Washington, DC 20460

Development of Bench-Scale
Settling Apparatus: Settling
Velocity Data for Design and
Operation of Wet-Weather Flow
Solids-Liquid  Separation
Processes

Interim Report

-------
                                                EPA/600/X-99/031
                                                September 1999
   Development of Bench-Scale Settling Apparatus:
    Settling Velocity Data 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.
             Montreal, Quebec H4K 1K5
                      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 has been funded by the United States Environmental
Protection Agency under CRADA No. 136-96. It has been subjected to the Agency's peer
and administrative review and has been approved for publication as an EPA document.
Mention of trade names or commercial products does not imply endorsement by the United
States government. This document was originally an internal report, intended for internal
Agency use only.
                                      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 reducing risk from threats
to human health and the environment. The focus of the Laboratory's research program is
on methods for the prevention and control of pollution  to air, land,  water and subsurface
resources; protection of water quality in public water systems; remediation of contaminated
sites and ground water; and prevention and control of indoor air pollution. The goal of this
research effort is to catalyze development and implementation of innovative, cost-effective,
environmental technologies; develop scientific and engineering information needed by EPA
to support regulatory and  policy decisions; and provide technical support and information
transfer to ensure effective implementation of environmental regulations and strategies.

      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 is a side-by-side comparison of a  traditional settling-column particle-
settling-velocity distribution evaluation method and a new settling evaluation method.
This portion of the study investigates whether these column tests are capable of
capturing or representing the rapidly settling particles present in wet-weather flows
(WWF). Equipment for the two testing methods was fabricated and laboratory tested
and preliminary evaluations were made.  This  interim report reviews the sampling
procedures and analytical methods used and presents data and results. Laboratory tests
were conducted with well characterized settling media, in order to measure suspended
solid (SS) concentrations and develop settling distributions on known substances in the
columns prior to testing actual WWF which exhibits variable SS concentrations  and
settling distributions.  The main purpose of this ongoing study is to obtain design data
for WWF  SS separator treatment devices, e.g.,  vortex separators,  grit chambers  and
settling tanks.

A summary of the performance as measured by predicted percent removal of both
columns for 15 laboratory tests is presented, as well as a comparison of the advantages
and shortcomings of the two methods.
                                     IV

-------
                                  Contents

Notice	  ii

Foreward  	iii

Abstract  	iv

Contents	v

Figures	vi

Tables	vi

Acknowledgment	  vii

       Chapter 1     Introduction	1-1
                   Background 	1-1
                          Objectives  	1-2
                   Combined Sewer Overflow	1-2
                   Settling Columns	1-3
                          Traditional Column	1-4
                          CERGRENE Columns	1-4
                          Other Columns	1-5
                   Theory of Settling Design 	1-5
                   In Field Sampling	1-7
                          Field Test Site 	1-7
                          Field Sampling Review	1-8

       Chapter 2     Materials and Methods	2-1
                   Columns Description and Delivery	2-1
                          Long Column	2-1
                          CERGRENE Columns	2-2
                          Mixing Basin	2-4
                   Sampling  	2-5
                          Long Column	2-5
                          CERGRENE Columns	2-5
                   Suspended Solids Analysis  	2-6
                   Identification of Experimental Materials	2-6
                   Experimental Design	2-7

       Chapter 3     Results	3-1
                   Quality Control Analysis	3-1
                          Blanks and Standards	3-1
                          Completeness	3.2
                   Mixing Basin - Recycle	3-4
                   Settleable Solids	3-5
                   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 Removed Long Versus CERGRENE	3-8
                          Matrix Iteration Process for
                                CERGRENE Columns	3-9
                          Eckenfelder Analysis for Long Column  	3-10
                          Design Removal  Comparison  	3-10

-------
                            Contents (continued)
      Chapter 4     Conclusions and Recommendations 	4-1

References  	5-1
Appendices

      A
      B
      C
      D
      E
      F
             Graphs of Standards, Blanks, Recycles, and Standard Methods	  A-l
             Graphs of Concentration Versus Time for Long and CERGRENE . . . . B-l
             Graphs of Percent Removal for Long and CERGRENE (exp. 10-15) .. C-l
             Results of VICTOR Analysis	  D-l
             CERGRENE Matrix Analysis	E-l
             Data Set 	F-l
2-1.
2-2.
2-3.
                                  Figures

      Long Column General Arrangement ................................. 2.2
      CERGRENE® Column Configuration ................................ 2.3
      Configuration for Side-by-Side Field Analysis ......................... 2.9
                                   Tables

2-1.   Summary of Standard Methods and Procedures 	2-6
2-2.   Experimental Design for Laboratory Test	2-8
2-3.   Typical Critical Time Measurements for One Run of Microsand	2-8
3-1.   QA Objectives for Measurements	3-1
3-2.   Completeness of Suspended Solids Analysis 	3-3
3-3.   Duplicate Analysis for Recycle Concentration and CERGRENE Columns ... 3-7
3-4   Comparison Predicted Removal between Long and CERGRENE Columns .  3-10
                                        VI

-------
                                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) administered by the Oak Ridge Institute for Science and Education
through an interagency agreement between the U.S. Department of Energy and the EPA.

The efforts of Mr. Larry Fradkin of the EPA's Office of Science Policy who assisted with the
establishment of the Cooperative Research and Development Agreement (CRADA 136-96) are
greatly appreciated.

Mr. Chi-Yuan (Evan) Fan of the EPA's Urban Watershed Management Branch (UWMB) lent
invaluable support to this project by establishing the CRADA and providing technical input to
the project.  Thanks also to Mr. Asim Ray of the UWMB for reviewing this document.
                                        vn

-------
                                  1.  INTRODUCTION
Background
       This Interim Report pertains to the laboratory evaluation of a field-testing apparatus
developed by the Centre d'Enseignement et de Recherche pour la Gestion des Ressources
Naturelles et de 1'Environnement (CERGRENE) of France which uses several small columns to
measure particle-settling velocities. This method was adapted for North American application by
John Meunier, Inc. The settling-velocity distribution and pollutant content may be used for wet-
weather flows (WWF) treatment process selection and design and for evaluation of preliminary
or existing process operation.

       The U.S. Environmental Protection Agency (EPA) National Risk Management Research
Laboratory's Water Supply and Water Resources Division, Urban Watershed Management
Branch in Edison (UWMB), 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.  John
Meunier, Inc. and the UWMB are jointly responsible for the prototype apparatus design,
construction, testing, and field evaluation as well as procedures to significantly improve
selection, design, and operation of full-scale WWF treatment processes that depend on solids-
liquid separation through settling and vortex separation.  This CRADA will compare the new
apparatus with the larger traditional column, the advantages and disadvantages of each method,
the projected standard operating procedures, QA procedures, expected results and the limitations
for both settling-velocity distribution tests for WWF.

       The newer settling testing method is 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
tests is presented and the two methods are compared.

        This study is a side-by-side comparison of an existing settling-testing method and a
newer settling-testing method.  The ongoing study is attempting to determine whether these tests
are capable of capturing or representing the rapidly settling particles present in WWF. Two
separate column testing methods were fabricated and laboratory tested and preliminary
evaluations were made. This interim report reviews the sampling procedures and analytical
methods used and presents data and results. Well characterized particles were used in the lab to
measure SS concentrations and develop  settling-velocity distributions in the columns prior to
testing WWF, particularly combined sewer overflow (CSO), which has extremely variable SS
concentrations and settling-velocity distributions.
                                          1-1

-------
       This study was planned in three phases:
•      Phase I:   Preliminary screening - Conducted in a laboratory setting to identify
       experimental parameters and determine process variables using well characterized
       particles. Aspects of this phase were performed by both parties (UWMB and John
       Meunier, Inc.). Procedures were then adjusted to allow for any difficulties encountered
       during this phase.
•      Phase II:   Laboratory bench-scale testing - The official QA approved experimental test
       runs of the side-by-side analysis of this phase were conducted in the laboratory of the
       John Meunier, Inc.
•      Phase III:  Field study - Side-by-side comparative study of the two settling
       characterization methods (conventional and CERGRENE) will be conducted at an offsite
       location with actual CSO samples. This side-by-side comparison will 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 will
       better represent settling velocities because sample storage and transport may change the
       naturally occurring settling velocities. Samples will be delivered immediately to the
       UWMB laboratories and analyzed for SS and other parameters. The Perth Amboy, New
       Jersey treatment works has been confirmed by both parties for the sampling of CSO.

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 and SS and associated pollutant content are essential for proper
assessment of high-rate settling and vortex separation technologies.

       The objective of this study is to compare sampling and analytical procedures of two
settling column techniques used to characterize the settling velocity of SS in WWF.  These
results will aid engineers in obtaining pertinent WWF pollution-abatement facility selection and
design data by analyzing particle-settling-velocity distribution, and settleable, suspended,
floatable fractions and design parameters. These design parameters include facility dimensions,
overflow rate, design flowrate, detention time and predicted removal efficiencies.

       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 precedence was needed. Thus a traditional column settling method was used..

Combined Sewer Overflows
       The recent 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:
                                          1-2

-------
•      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 wet weather flows

       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 existing physical treatment, 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 data evaluations are needed for designing CSO treatment facilities, as CSO differs from
dry-weather flow (DWF).  CSO settleable solids build up and characteristics in the  sewer system
are a function of the length of the antecedent dry-weather period, sewer slope, drainage area
(catchment) size, fiowrate, and drainage area soil characteristics, etc., whereas DWF solids
characteristics (barring industrial sources) are similar from place to place.  Furthermore,
suspended and settleable solids concentrations can vary with time during the storm  events and
from storm to storm.

       Past 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 |im) found in stormwater can achieve  settling velocities of 2.5 m/h (0.07
cm/s) or more (Chebbo et al., 1990) and 70% to 80  % will deposit within 15 min and more than
97 %  after 1 hr.
Settling Columns
       The traditional settling method for determining settling-velocity distributions uses side
ports to analyze quiescent sampling. Camp (1945) published settling curves using Stoke's Law
based on particle settling and Eckenfelder (1966) used it as a design aid for sedimentation


                                          1-3

-------
processes and for analysis of flocculation. There is substantial variability associated with this
method (hereafter Long column).

       Currently, only one method for measuring gravity separation is accepted by Standard
Methods1 (SM 2540.F.b; 19th Edition) called "settleable solids".  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 hr.
There are no control limits or substantiating data for this method.

       This  method uses a column of at least 20 cm in depth.  A sample is pipetted from the
center of the column after 1 hr of quiescent settling to determine the nonsettleable solids.
Settleable solids are equal to the initial SS concentration minus the nonsettleable solids
concentration.

Traditional  Column
       The typical 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.) with side withdrawals evenly spaced
along the column depth.  The height of the column simulates the effective settling depth which
occurs in a sedimentation tank that typically has constructed depths exceeding 2.5 m (8 ft).  This
column requires an extensive laboratory layout. Various methods have been used to pre-mix the
sample before the column test begins, e.g., plunger plates and  rotation of the settling column.
Depending on specific dimensions between 40 and 80 L (10 and 20 gal) are required to fill the
column.  The 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 intial samplig time,  t0, due to the heavy particles in
WWF. This is partially caused by the length of time required  to  fill the column and the time
required to withdraw samples from all ports sequentially. Pisano et al. (1984) went to the extent
of mounting the Long column on a device that allowed 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 at t0 using the classical settling column for WWF, which may result in predictions of
lower than actual SS fractions.

CERGRENE Columns
       CERGRENE (Chadirat et al., 1997) developed a new design that uses a sequence of 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
       1 Standard Methods describes other settling methods which are applicable to the zone
settling of sludges. Sludges have significantly higher SS concentrations and different
characteristics than CSO.

                                          1-4

-------
CERGRENE settling columns, like the long column, are designed to sample for SS
concentrations of the WWF in the original sewage matrix.

       The new CERGRENE columns have a shorter time to fill (approximately 7 s) and may be
more completely mixed at the initial sample time (better representation of to). 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 for
field as well as laboratory use. Settling-velocity-distribution samples taken in the field should
give a truer representation of the settling rates of the combined sewage. Settling velocities of
samples taken in the field should be faster than samples taken back in the lab or stored in the  lab
for longer periods of time  due to less time allowed for agglomeration.

Other Columns
       Other methods developed in Europe are:
       •     Brombach or German (Michelbach and Wohrle ,1993 and Pisano and Brombach,
             1996);
       •     Norwegian Institute for Water Research (NIVA) (Lygren and Damhaug, 1986 and
             Walker etal, 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, offer several benefits over the Long column.
They require less 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 smaller testing volumes, approximately 4tol2L(lto4 gal). This is especially true of the
German and the NIVA columns (less than 1 m deep and 5 cm wide) which  are also more
amenable to field use. 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. These methods provide truer
representation of high settling-velocity SS because  the concentrated sample is situated above the
settling column and dropped into it at i^.

       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) which 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 will be 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. The first two factors alone could be used to design
the physical dimensions of a basin, however, once the third factor is included, the characteristics
of the SS in the WWF must be taken into account. For sedimentation tanks (Tchobanoglous and

                                         1-5

-------
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 flow 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 greater than Vc or q will be removed with some
fraction of all other particles also being removed. For the purposes of this project, the overflow
rate will be used in the graphs 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 the four types of settling: discrete, fiocculant,
hindered, and compression (Tchobanoglous and Burton, 1991). The settling velocities of discrete
and fiocculant 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 POTW in secondary-settling tanks and sludge-handling devices or industrial
applications.

       Various studies used discrete settleable solids and various column devices or settling
methods to determine settling velocities. In Stake's Law (Equationl), the velocity of an ideal
sphere is proportional to the square of the particle diameter.

vs = g(ps- p)d2/(18^)                                                                  (1-3)

where:
vs = velocity of sphere, m/s
g = acceleration due to gravity, 981 m/s2
ps = density of the particle, kg/m3
p = density of the fluid
d = diameter of sphere, m
|i = dynamic viscosity, N-s/m2

As previously mentioned, settling columns have been used to observe and analyze fiocculant
settling. WWF is often a combination of discrete and fiocculant settling.

       While direct measurement of sedimentation efficiency can be made on controls after
installation by taking grab samples at the influent and effluent of the controls, the settling column
and its predicted removals can assist the engineer or scientist in the selection of design
parameters before installation for WWF storage and treatment facilities. Settling columns can


                                           1-6

-------
help determine the settling velocity distributions for local conditions, e.g., silty WWF may
require larger facilities for a desired percent removal, while gritty waste streams could achieve
the same percent removals with much smaller facilities.  Onsite analysis of this overflow rate
derived from the observation of the actual settling velocity distribution is a better design
component than the assumption derived from Stoke's Law which only relies on the settling of
discrete particles.

In Field  Sampling
       Sampling devices must be able to capture the heavier SS or settleable solids and not
manifest  biased results due to stratification. For an automatic sampling device,  this means that
its intake velocities and ports must be greater than the mainstream velocity and be placed at
multiple levels in order to capture the heavier particles near the channel invert, respectively.

       The importance of in-field sampling is related to the change in settling properties due to
storage and transport.  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 hr for the second test. This difference in stored samples was also confirmed by CERGRENE
(Aiguier  et al., 1995), when a fresh sample was compared to the settling rates of three samples
stored for 24-hr at different temperatures (room, refrigerated and frozen).  Each sample, all
collected at the same time, had different settling distributions.

Field Test 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), the
number and volume 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 CSO's. 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, New Jersey 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 to a regional wastewater treatment
plant 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. A CSO
regulator is located about 6 m (20 ft) below the Sadowski Parkway with an overflow weir and 2
m (7 ft) diameter CSO tide-gated outfall to the mouth of the Raritan River.
       The pumping station inflow is from the interceptor that discharges into one of two
wet-wells each equipped with a mechanical coarse bar screen for removing large debris and
                                          1-7

-------
protecting the sewage pumps.  The screen chamber inflow may be utilized for settling studies,
since the inflow is a part of the CSO and will have the same characteristics at the outfall point
during storm events. The wet-well is approximately 9 m (30 ft) deep. This is an enclosed
facility with 24 hr access and a person on duty. Grit which accumulates in the wet-well is
removed from the facility once every three months. A winch and two pumps are also available.
This location is approximately five miles away from the UWMB in Edison, NJ.
                                         1-8

-------
Field Sampling Review
       As background for this project John Meunier Inc. reviewed and wrote an internal report
(Champigny et al, 1997) on the state-of-the-art of 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 objective 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 are 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 has been observed by some researchers.
•      Sediments found at the bottom of the channels interact with the SS  and have to be
       included in the sampling.

       Two separate sampling systems were recommended:
1.     Sampling a complete section of the flow from bottom to top, or
2.     Placing sampling port intakes at two points.

       This second method would mount one sampling point just above the level of the dry
weather flow, near the pipe walls. The second sampling point would be maintained at 60% of
the total water level throughout a WWF event.
                                         1-9

-------
                           2.  MATERIALS AND METHODS

       The design and fabrication of the columns was conducted in the EPA UWMB facility in
Edison, NJ and in the offices of John Meunier, Inc. in Montreal, Quebec, Canada.
Column Description and Delivery
       The Long and CERGRENE columns were fabricated and preliminary evaluations were
conducted by the EPA and the John Meunier, Inc., respectively. In phase I, different types of
particles were tested to select the best media to be used in the benchtop laboratory studies and to
answer other questions including sampling and analysis procedures, and experimental parameters
and number of samples.  In phase II, a set of 15 experiments was conducted in John Meunier's
Laboratory from June 9, 1998 through June 17, 1998.

Long Column
       The accepted settling-velocity distribution determination methods in the United States
have commonly employed 1.8 to 2.5 m (6 to 8 ft) columns to study settling characteristics of
solids.  The EPA designed an eight-foot modular column fashion which allows for several
assembly and sampling configurations (Figure 2.1).  The column is made from cast acrylic
tubing with an 203 mm (8 in.) outer diameter, a wall of 6.5 mm (0.25 in.) and 190 mm (7.5 in.)
inner diameter. The column consists of four separate modules: a base section, a 1.2 m (4 ft)
section, and two 0.6 m (2 ft) section and allows the column to be set up in a 1.2., 1.8 or 2.5 m (4,
6, or 8 ft) configuration.  The 1.2 m (4 ft) section must be attached to the base. The volume of
the column is approximately 70 L (18 gal).  The modules are attached by acrylic flanges with
foam gaskets to eliminate leakage. A cap is also available to prevent foreign  material from
entering the top of the column.

       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.) are located every foot, on either side of the
column, for a total of 16 ports at eight depths. Sampling  from both sides of the column is 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 the sample flows are highest than at the top where
samples flows are smaller due to decreased head. The base section includes a one inch diameter
drain which is connected to  a three-way valve.  This valve is used for filling and draining the
column. A conical plastic piece (funnel) is installed above the drain inside the column to
facilitate resuspension of solids during the filling process and aid clean out between tests.  A
wooden baffle screwed into  the cone disperses the flow and keeps the influent well-mixed.

       Filling is accomplished by pumping from a mixing basin through the bottom valve.
Before filling, the pump is primed, and the bottom valve turned to "fill."  Prior to and during
filling, the mixing basin (described later) stirs the solids to keep them suspended. After filling,
the pump is turned off and the bottom valve is turned to the middle position.  Sampling from the
side ports can then proceed.


                                         2-1

-------
                           2.  MATERIALS AND METHODS

       The design and fabrication of the columns was conducted in the EPA UWMB facility in
Edison, NJ and in the offices of John Meunier, Inc. in Montreal, Quebec, Canada.
Column Description and Delivery
       The Long and CERGRENE columns were fabricated and preliminary evaluations were
conducted by the EPA and the John Meunier, Inc., respectively. In phase I, different types of
particles were tested to select the best media to be used in the benchtop laboratory studies and to
answer other questions including sampling and analysis procedures, and experimental parameters
and number of samples.  In phase II, a set of 15 experiments was conducted in John Meunier's
Laboratory from June 9, 1998 through June 17, 1998.

Long Column
       The accepted settling-velocity distribution determination methods in the United States
have commonly employed 1.8 to 2.5 m (6 to 8 ft) columns to study settling characteristics of
solids.  The EPA designed an eight-foot modular column fashion which allows for several
assembly and sampling configurations (Figure 2.1).  The column is made from cast acrylic
tubing with an 203 mm (8 in.) outer diameter, a wall of 6.5 mm (0.25 in.) and 190 mm (7.5 in.)
inner diameter. The column consists of four separate modules: a base section, a 1.2 m (4 ft)
section, and two 0.6 m (2 ft) section and allows the column to be set up in a 1.2., 1.8 or 2.5 m (4,
6, or 8 ft) configuration.  The 1.2 m (4 ft) section must be attached to the base. The volume of
the column is approximately 70 L (18 gal).  The modules are attached by acrylic flanges with
foam gaskets to eliminate leakage. A cap is also available to prevent foreign  material from
entering the top of the column.

       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.) are located every foot, on either side of the
column, for a total of 16 ports at eight depths. Sampling  from both sides of the column is 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 the sample flows are highest than at the top where
samples flows are smaller due to decreased head. The base section includes a one inch diameter
drain which is connected to  a three-way valve.  This valve is used for filling and draining the
column. A conical plastic piece (funnel) is installed above the drain inside the column to
facilitate resuspension of solids during the filling process and aid clean out between tests.  A
wooden baffle screwed into  the cone disperses the flow and keeps the influent well-mixed.

       Filling is accomplished by pumping from a mixing basin through the bottom valve.
Before filling, the pump is primed, and the bottom valve turned to "fill."  Prior to and during
filling, the mixing basin (described later) stirs the solids to keep them suspended. After filling,
the pump is turned off and the bottom valve is turned to the middle position.  Sampling from the
side ports can then proceed.


                                         2-1

-------
                      Col urnn  Conflguratio 11
               Stmipiiuy: 1'urls _. -•••
                 (c v cry 1" i
                   Sample

                     Fill
                      Figure 2.1 Long Column General Arrangement

       Sampling tubes are attached to a male quick disconnect fitting at each sampling port. For
two-sided sampling, two tubes attached to male quick disconnect fittings lead to a T-connector
which is attached to a short tube. This short output tube brings the sampling streams together.
Sampling from the end of this output tube is controlled by attaching another set of quick
disconnect couplings.

       Sampling is initiated at the top of the column, and progresses downward for each time
interval.  Each sampling tube are purged before sampling. The output tube is placed in the
plastic bottle marked "Purge," and  a male pipe adapter is attached to an elbow hose barb.  The
"Purge" bottle is filled to a measured marking and then in midstream the tube is quickly moved
to a plastic storage bottle.  Each plastic storage bottle is marked with an individual identification
number which is recorded along with the corresponding port (1, 2....8) and sample time. After
sufficiently filling the plastic storage bottle, the male quick-disconnect adapter is removed from
the output tube.  Storage bottle size was nominally 250 ml which appeared to match the 10-20
mg target mass range for SS analysis for the media in Phase I.  Cold storage was not required for
Phases I and II as the samples contained inert material (e.g., sand and clay).  Storage
requirements that address  the specific types of analysis for combined sewer samples which will
be taken in Phase III are presented in the Work/QA plan.

CERGRENE Column
       CERGRENE, the Centre technique international de rAssainissement, 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 this study were to:
                                         2-2

-------
•      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 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 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 is constructed of 65 mm
(2.5 in.) inner diameter clear PVC and stands approximately  1 m (3 ft) tall.  The volume of this
column is approximately 2.2 L (0.55  gal). The column (Figure 2.2) has three valves located at
the top, the bottom and the middle. The middle valve, a 65 mm (2.5") inner diameter ball valve,
is approximately 2/5 of the length from the bottom and divides the column into two sections.
                     to vacuum
                               valve

                     clear PVC 	
                   65 mm
                   inner diameter
                   ball valve
                                  valve
                from mixing basin
1 m
        Not to Scale
                     Figure 2.2 CERGRENE® Column Configuration

       The column is filled by vacuum aspiration to minimize the variation of SS concentration
between the bottom and top sections. An advantage of a vacuum pump is that it doesn't break
SS apart as readily as a typical positive displacement water pump which tends to grind up and
shear particles thereby changing the settling characteristics.  After filling, the top and bottom
                                        2-3

-------
valves are closed. At the specified sampling time, the two sections are separated by turning the
middle ball valve. The bottom portion is the sample. This volume of approximately 1 L (0.25
gal) is drained and SS analysis is performed. As previously discussed, the major premise for
performing this comparison is to determine whether the CERGRENE column provides truer
representation of high-settling velocity SS at the test starting time, to, than the Long column.
       Each CERGRENE column is sampled independently at discrete times and represents one
sample for settling-velocity distribution analysis; this differs from the Long column which
requires multiple samples to be taken per time interval.  Filling of the columns is conducted
sequentially.  Results from Chebbo et al. (1995) indicated that sequential filling of the columns
did not significantly impact results.

Mixing Basin
       The mixing basin is 0.66 m (2 ft) in diameter and 1.3 m (4 ft) high, and has a 300 L (80
gal) holding volume. Four vertical baffles were inserted at 90° intervals to prevent the formation
of a vortex (Dickey and Hemrajani,  1992; and Etchells et al., 1992).  The mixer is mounted on a
sawhorse above the basin. The mixer shaft is in the middle of the basin and two impellers are
used, i.e., a marine impeller at the bottom and 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;  and 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 withl/3 hp of
power, has a constant mixing velocity of 350 rpm.

       The use of a  Rushton impeller is 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
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  of 200 rpm and 600 rpm were adequate to
generate complete mixing in the basin and achieve the similar results (Chadirat et al., 1997).  The
mixer was turned on 15 min prior to and stayed on throughout sampling.

       During the experiments, the mixing basin was filled  with an initial SS concentration of
300 mg/L.  On the sixth side-by-side experiment, the volume in the mixing basin was changed
from 200 L 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/L. 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

-------
Sampling
       The Long column and CERGRENE system were filled from the mixing basin in an
alternating sequence. The point of withdrawal was 7 to 10 cm (3 to 4 in.) from the wall, set
halfway between the baffles and at an alternating height of 18 to 35 cm (7 or 14 in.) from the
bottom which maintained a volume of water in the basin throughout a test.  The alternating
sequence for filling and height of withdrawal were intended to reduce bias in the experiment as
described in Experimental Design.

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 up flow velocity of water in the 10 out of 15 experiments where both height and
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
       The vacuum pump filled the CERGRENE columns in an average of 8.2 s. Using a
standard height value of 0.91 m (3 ft) for the columns the uplfow velocity 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 for this scenario with a worse case sample height of 0.76 m (2.5 ft), the
up flow velocity in the CERGRENE would exceed 6.6 cm/s. The up flow velocity in the
CERGRENE column is at least twice as fast as the Long column.

        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 is 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 3.1.2 and specific
recommendations and a CERGRENE institutional modification are presented in Chapter 4.
                                        2-5

-------
Suspended Solids Analysis
       Suspended solids was the critical measurement of these experiment. Table 2.1 presents
the summary of Standard Methods used. The selection of analytical methods is based on the
following priority:
       1.     Standard Methods, 19th Edition.
       2.     EPA Method

                 Table 2.1 Summary of Standard Methods and Procedures
Parameter
Suspended Solids
Settleable Solids
Samp le
Type
Water
Water
Method
No.
2540 D
2540 F
Metho d Title
Total SS Dried at 103-105 °C
Settleable Solids
Source
Standard Methods
Standard Methods
       The upper limit for SS sample sizes is 200 mg of residue. The lower limit for SS is 4
mg/L as specified by EPA method 160.2.

       SS was calculated by the following procedure:
             mg  suspended solids/L =
                                            (A- B)X 1000
where:
                            sample  volume, mL
A = weight of filter and dried residue, mg and
B = weight of filter, mg.
                                                                     (2-1)
       The grain size of the particles for the laboratory experiments exceed the filter paper pore
size (1.5 |im, Whatman 934-AH) of the filters being used for the SS analysis. For the laboratory
experiments, 47 mm diameter filters were used; this may be switched to 70 mm diameter filter
for the raw wastewater samples to be taken in the field as recommended by Standard Methods.

       To perform settleable solids, a graduated cylinder was filled from the pump used for the
Long column.  After one hour, a 250 mL sample was siphoned from the approximate center of
the graduated cylinder and was analyzed for SS. This was the nonsettleable solids. This
concentration is subtracted from the initial SS concentration as derived from separate samples.
As settleable solids is an extension of SS analysis, the same limits apply.
       Settleable solids is calculated by the following method:

mg settleable solids / L = mg suspended solids / L - mg nonsettleable solids / L
                                                                     (2-2)
Identification of Experimental Materials
       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.
                                         2-6

-------
Microsand was chosen to test the mixing basin and for the laboratory experiments. The
characteristics of the microsand used for the testing were :
              d10 = 85
              d60 = 125
              p = 2.62 g/cm3

The calculated settling velocities for these parameters using Stokes Law (ideal sphere assumed)
are <1 cm/s, which is less than the upfiow velocities measured in the columns.

      A surficial soil  excavated from near Princeton, New Jersey, was also used as an
additional reference. The soil (Neshaminy) is a silty clay loam, containing 17% sand, 46% silt,
and 37% clay. Although this soil has a greater proportion of very small particles, it also has a
wider variation, with d10 = 75 |im and d60 = 700 |im. (Fischer, 1995)

      An analysis of the microsand by a Coulter® LS Particle Analyzer determined the
following particle diameters:
             d10 = 156.2
             d50 = 232.8

This analysis was performed in January, 1999 after the Phase II was complete.
Experimental Design
       Table 2.2 shows the original experimental design of the side-by-side test to account for
differences due to the order of filling the columns and variations in the height of the mixing
basin.  One duplicate was performed for each medium (microsand, Neshaminy and mixture) for a
total of fifteen runs. The run order was randomized to reduce bias due to one soil type being
repeated and to reduce influence of "learn as you go" increased precision.

       Table 2.3 provides a prototype sampling strategy for microsand. A set of samples from
the Long column will be withdrawn at specified intervals (e.g., 1 min, 3 min, 5 min, 10 min) with
time measurements to the nearest second for the individual sample end-times. One set of
samples for the Long column comprises four samples. For each CERGRENE column, the valve
is turned to capture SS at a time corresponding to one time "set" for the Long column. The initial
time for the individual CERGRENE column shall be  designated t0; where i represents the
number of the column. The time for the CERGRENE column is the time from the moment each
column is filled to the time the sample is isolated by turning the valve of each column. The Long
column t0 was the time that filling was completed, and initial concentration was assumed to be
background or "recycle" concentration as taken from  the mixing basin by the pump as explained
in Chapter 3.
                                         2-7

-------
                    Table 2.2 Experimental Design for Laboratory Test
Media
Microsand
Microsand
Microsand
Microsand
Microsand
Clay soil
Clay soil
Clay soil
Clay soil
Clay soil
Mixture
Mixture
Mixture
Mixture
Mixture
Run#
1
2
3
4
Duplicate - 13
5
6
7
8
Duplicate - 14
9
10
11
12
Duplicate - 15
Randomized
Run # 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
          Table 2.3 Typical Critical Time Measurements for One Run of Microsand
CERGRENE Column
Initial Time
V
tfl,2
t«,3
t0,4
to,5
Duplicate - ^ x
Time
V
1 min - 10 2
3 min - 10 3
5 min - t0>4
10 min-t05
x min - t0;X
Long Column
t0 - initial time
< 1 min
< 2 min
~3 min
~5 min
-10 min
Port 3
ti-to
Vt0
Vt0
tl3- to
tl7- to
Port5
t2-t0
Vto
MO" tfl
tl4- to
M8~ tfl
Port?
Vto
Vt0
tn- to
tl5- to
M9~ tfl
Port8
Vt0
Vt0
tl2- to
tie" to
tao" to
Cannot perform Duplicate on Long column
       A similar table to Table 2.3 could be constructed for the Neshaminy soil and other
particle mixtures, with the time of the samples extended out to one hour. The estimated settling
rate limit of Microsand d10 80|im < 15 min Long and <10 min for CERGRENE.
                                        2-8

-------
       Since only four CERGRENE columns were available for the five or six required tests for
each run, the first two CERGRENE columns were sampled, rinsed out and refilled from the
mixing basin during Phase II.  This procedure was used for the first nine test when it was
discovered that the CERGRENE columns were producing random settling results. Then only
one CERGRENE column was used for the remaining six experiments, and the column rinsed out
after each individual test.

       Ideally, any variation caused by the mixing and separation of flows was not expected to
exceed values already determined in Phase I. Figure 2.3 shows the lab setup.
                                 Monarch Type A Pump
                                                                       Not to Scale
Vacuum Pump
              Rushton impeller
           Marine impeller
             John Meunier Miner
                 Figure 2.3 Configuration for Side-by-Side Field Analysis
                                        2-9

-------
                                      3. RESULTS
       All figures for this section 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 testing procedure were incorporated to address quality control in the
Phase II analysis.

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 the cleanliness of SS analysis. All blanks
showed concentrations below 4 mg/L (Figure A-l).

	The accuracy of the SS procedure was determined from the analysis of laboratory control
samples whose true values are known 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. In phase III, diatomaceous-silica will be used as the SRM.
Table 3.1 specifies the quality control standard concentrations and the expected recoveries for
diatomaceous-silica.

                        Table 3.1 QA Objectives for Measurements
Measurement
Suspended Solids
Settleable Solids
Method
2540D
2540F
Reporting
Unit
mg/L
mg/L
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
       Accuracy is expressed as percent recovery. The formula used to calculate this laboratory
QC values for a SRM is:

%R = 100% x Cm / Csnn                                                              (3-1)

where: %R  = percent recovery
       Cm    = measured concentration of SRM
       Csrm   = actual concentration of SRM

Excluding two extremes, the average percent recovery was 83% with media results of 76%,  83%
and 88% for sand, mixture and clay, respectively.
                                         3-1

-------
       Figures A-2 through A-5 show an analysis 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
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 expected 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 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, while Long column sample were 250 to 300 mL and CERGRENE samples
were 250 mL 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 percent is expected to ensure that sufficient valid data are collected to evaluate
the settling velocity distributions. Table 3.2 shows the completeness for critical SS
measurements made for all 15 test.  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

       The Recycle and the Standard Methods  samples are the only sample types that do 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 3.2, only  17 of the 340 long column samples were disqualified for
faulty measurements (e.g., sample volume not noted). Additionally, three microsand samples fell
below the 4 mg/L limit of detection for the EPA SS method, though these samples appear in the
graphs as approaching zero concentration, or 100% removal.


                                         3-2

-------
                   Table 3.2 Completeness of Suspended Solids Analysis
Type or location
Blanks
CERGRENE
Long
Recycle
Standard Methods
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 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
became more critical during analysis of the settling velocities 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 test proceeded, better control of the level in the
CERGRENE columns was exhibited.  For experiments 10 through 15, which are the only
acceptable CERGRENE experiments, 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 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. This was not as apparent with the mixture, as the
Neshaminy may have clung to some of the sand and reduced water tension of the sand to the
filter housing.
                                        3-3

-------
       Neshaminy soil tended to clog the filters for large 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 formicrosand  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 colloidal mixture.  Sample loading for the filters exceeded
200 mg for the CERGRENE columns. Thus the larger sand and mixture samples tended to
overload the filters, which may have introduced considerable error in microsand CERGRENE
runs.
Mixing Basin - Recycle
       Three "recycle" samples were obtained immediately prior to filling the Long column.
The known concentration in the mixing basin was 300 mg/L, so recycle concentrations should
have centered around that number. The recycle concentration of each test 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.  The recycle
concentration was thought to better represent the concentration delivered to the Long column.
The average recycle concentration for all 15 test 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/L, 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/L 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.  However, an analysis of variance showed no
significant difference in recycle concentrations between media type, order of filling or intake
depth.
                                         3-4

-------
Settleable Solids
       The gravimetric settleable solids analysis (SM 2540F) entails first performing a SS
analysis on a representative sample. In this experiment, 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
half the mass of Neshaminy soils. An interesting result, however, is the tighter distribution of the
mixture results, possibly due to a fiocculent 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 an increasing pattern 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
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 fifteen 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 re liable 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 to
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.


                                         3-5

-------
       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 to 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 have lower specific gravities than sand and typically are not
spherical in nature. The Neshaminy also took longer to settle with significant concentrations
after one hour while the sand had settled out within five minutes. This problem has long been
recognized as a shortcoming of the Long column, and causes the scatter that can be seen in the
early measurements of more in-depth analysis.

       In fact, concentrations for the lower Ports 5, 7 and 8 , especially 7 and 8, during the test
using the Neshaminy soil were probably demonstrating hindered zone settling and could even
have been displaying compression zone settling. Concentrations were relatively flat throughout
the test, except for an initial dip, and even began to exceed the known delivered concentration of
300 mg/L for ports 7 and 8 after one hour as demonstrated in Figure  B 31.

CERGRENE Shortcomings - Lack of Rep eatable Results
       The precision of the 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
             C2 = the smaller of the two observed values

       When three or  more replicates are available, the relative standard deviation (RSD),
instead of the RPD, was used as follows:

RSD = (s/y)xlOO%                                                                   (3-4)

where:       s  = standard deviation, and
             y = mean replicate analysis.
                                          3-6

-------
The standard deviation is defined by:
s =
                                      n- 1
                                                                                 (3-5)
where:       s = standard deviation
             y; = measured value of the z'th 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 3.3.

      Table 3.3 Duplicate Analysis for Recycle Concentration and CERGRENE Columns
Test
#
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
QA
Run#
8
12
11
3
5
14
15
1
9
6
10
13
4
7
2
Withdrawal
Height (cm)
18
36
18
36
18
18
36
18
36
36
18
36
18
36
18
Filling
Sequence
L/C
L/C
C/L
C/L
L/C
L/C
L/C
C/L
C/L
L/C
L/C
L/C
C/L
C/L
C/L
Media
Neshaminy
Mixture
Mixture
Neshaminy
Neshaminy
Microsand
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
14.1
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
       This data indicates that the variation of the recycle concentration for each test was
random and not media driven (microsand, Neshaminy and mixture).  This may due in part to the
                                        3-7

-------
size of the samples taken (about 250 mL) and the force with which the sample bottles were filled
by the pump (the same pump used to fill the Long column).

      The duplicate analysis for the CERGRENE column tells a different story. During the
first nine tests, four CERGRENE columns were used. For the remaining six tests, only one
column was used. Here the CERGRENE columns behave randomly for the first nine tests, and
then duplicate analysis improves dramatically as distinct patterns can be interpreted from the
media being tested. The Neshaminy and mixture have better duplicates than the microsand.
While the amount of data may not be large enough to state this finding with statistiscal validity,
this observable result is expected.  The percent recovery analysis of  the "Standard Reference
Material" indicated the microsand had the largest variation while the Neshaminy had the least
with the mixture somewhere in between. That the performance of the CERGRENE columns in
the last six test shows that the duplicate were better than the expected results from the SRM
analysis may be due in part to the CERGRENE columns using larger volumes of sample.

      As the recycle concentrations continued to behave randomly, the improved CERGRENE
performance in the latter test was not due to enhanced technique of the testers as the experiment
progressed.  One column, instead of a several columns each with their own idiosyncracies,
produced repeatable results. This duplicate analysis only compared the SS concentration and not
settling rates.

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 in the Long column 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. Had a plug flow been introduced in the top of the column,  the settling
velocity computation would be more straightforward.

      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 were obviously not equal to the recycle concentrations (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 until the top water level) by the specific 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

                                         3-8

-------
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.

       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 flaws are apparent in the sand
experiments (12,13, and 15). The Long column is overstating overflow rates for the fast settlers,
as the right side of the Long column graphs should approach zero for increasing settling rates,
and the CERGRENE columns are under estimate removal rates, as they should approach 100%
on the left side of the graphs. For the Long column, this shortcoming is due to the lack of
adequate initial mixing. The assumption of C0 being equivalent to average recycle concentration
yields 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 vave settled out at five and ten minutes and percent recovery should have been  100% .
Losses of mass result in prediction of lowere concentrations. Besides the problems with the SS
anlysis already discussed under Completeness, some sand particles may have been trapped in the
ball valve mechanism. This is currently under investigation.

Matrix Iteration Process for CERGRENE Columns
       The CERGRENE group of France (Lucas-Aiguier et al., 1997) developed a methodology
to use the data from the small columns to produce settling velocity distributions.  A spreadsheet
application, "VICTOR",  was developed which utilizes an iterative method to solve simultaneous
equations, resulting in a matrix M(i,j) which contains mass removed for each particular time
interval i and pollutant/ Based on this matrix, a distribution of settling velocities may be
constructed. A more complete description of the derivation is in Appendix G.

       The nature of the CERGRENE process sometimes results in certain points being in error.
Thus a graph of M(i,j) 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. The software allows the user to choose either analysis of all points, which includes all
data in the computation of the velocity distribution, or analysis with "suppression", which
excludes points which do not show the expected increasing concentration in time. The choice  of
suppression or non-suppression is based upon several factors, but unless the points in question
are significantly skewed to one direction, it will not have a severe impact on the resultant
velocity distribution.

       As discussed earlier, further analysis of CERGRENE experiments 1 through 9 is not valid
due to the inconsistencies between columns.  Figure D-l illustrates experiment 9, where the
column order was randomized, and no consistent pattern can be seen in either graph.
Suppression is not possible with this data, as no pattern can be inferred from the mass removal
graph.  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, as well as a more varied velocity distribution.

                                         3-9

-------
       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 second
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. 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 proves a useful tool for  computing a settling velocity distribution for the
CERGRENE columns. Care must be taken, however, in trying to compare these velocity results
to the results  from other types of analysis, which use different assumptions and computational
techniques, and even different methods for deriving settling velocity distribution. Additionally,
VICTOR would appear to work better with higher numbers of samples and time intervals. The
capability of the software to track several pollutants could be a very useful for partitioning
experiments,  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 flocculent 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 or compaction zone settling. During the
sand and mixture experiments the sand settled within a five minute time frame and exhibited
properties of discrete 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
shows some similarities, though CERGRENE analysis is complicated by the lack of sufficient
data points.
                                         3-10

-------
       The calculated Stoke's Law settling velocities for ideally spherical sand at 15 °C ranged
from 0.5 cm/s for 80 |im diameter sand to 12.6 cm/s for 400 |im. At 230 |im, the dso value as
calculated by the Coulter® LS Particle Analyzer for the microsand, the settling velocity was 9.3
cm/s.

      Table 3.4 Comparison Predicted Removal between 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
(cm/s) Longf
0.069 i 0.81
1
0.0077 |
I
0.00086 i
1
4.4 | 0.61
I
0.14 |
1
0.0046 i
I
100 i 0.64
1
18 i
I
3.1 i
1
120 i 0.65
i
20. !
1
3.4 i
I
0.014 i 0.43
1
0.00059 i
I
0.000024 i
1
220 i 0.66
i
30. !
1
4.0 |
Overflow Rate
(cm/s)
CERGRENE*
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
r2
(#pts)
0.96
(5)

0.99
(3)

0.22
(4)

0.92
(4)

0.82
(4)

0.90
(3)

f Based on average recycle concentration. Points with calculated %Removal<0 were deleted.
1 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

-------
                         4. Conclusions and Recommendations
       The Preliminary Results (Phase I) determined the expected homogeneity of the mixing
basins and the initial performance of the respective 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 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 from the top two ports.  Only ports 3, 5, 7 and 8
were used during the evaluation for determination of settling methods.  At this point analysis
had been performed independently by the EPA and John Meunier, Inc.

       The objective of the Phase II experiments was to compare, in side-by-side analysis, two
methods of measuring  settling velocity.  Lab 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 in the
mixing basin as well as the effect of filling order on experimental results and will be used in
Phase III.

       The original experimental design did not account for the three most important factors
affecting results during the Phase II of the experiment:
1. Individual behavior  of the CERGRENE columns
2. Overloading of the filter by  the samples
3. Wide variation in SS concentrations in the attempt to capture the micro-sand

       Initial conclusions from Phase II are:
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, as particles tend to stick to  analytical equipment and loss of mass easily occurs.
       This loss  was especially noticeable in the 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 as the analysis of the
       "standard reference material" was in the direction of a loss of sand.
2.     The performance of the CERGRENE columns was erratic for experiments  1-9. While the
       expected  results of a plot of concentration versus time would be an increase in the
       concentration with time, the columns in experiment 1-9 behaved almost randomly.  In
       contrast, experiments 10-15, where only one column was used, indicated a trend of
                                         4-1

-------
       increasing concentration versus time.  This seemed to point to a lack of consistency
       between columns.
3.      The long column was never fully mixed at to, especially for the microsand and mixture
       test.
4.      The CERGRENE columns were examined for defects and sizing specifications for
       volumes and height. No anomalies were found, though a small indentation in the ball
       valve could be trapping some solids and releasing them later.

Recommendations for the CERGRENE column are:
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 for this test. Starting at the same height would
       also allow for better duplicate analysis.
2.      The filling procedure should allow at least a Vz volume overflow to purge the lines.
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.
4.      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.
5.      Evaluations of other types  of valves should be conducted to attempt to minimize
       interference of the valve mechanism on solids settling.
6.      A mass balance should be performed around the center ball valve. The concentration of
       the top and bottom parts of the column should be measured at to, to ensure that the initial
       concentrations, before settling, are the same.

Long column recommendations are:
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 time of samples should be reduced to five
minutes while increasing the frequency samples. As previously discussed, the microsand
diameters are  larger than originally reported which increases the settling rate.

       It was a benefit from our standpoint to analyze "standards" using the same media that was
used in the experiments. This data confirmed that we should expect large error (or losses) for the
media 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.


                                         4-2

-------
                                  5. 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 mesure 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).

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.

-------
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).

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 and Helical 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 Burton, F.  "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,
1993.

Walker, D., Golden,  J., Bingham, D. and Driscoll, E. Manual: Combined Sewer Overflow
Control. EPA/625/R-93/007 (NTIS PB 93-144 649), Cincinnati, OH: U.S. Environmental
Protection Agency, 1993.

-------

-------
 4





 3
 1




 0




-1

-------

-------

-------
                            * **
            * *   *»        *
1


-------
400 -,
100
— > \J\J
900
Zj wvy
1 00
-Lwvy
0 -
* t
i 1
^
cA
400
300
200
100
  0
                   7"
 ^
T
                                   14"

-------
100 -,
80 -

60 -
40 -
20 -

o J
^^^^^^^^^^^^^~l
• |
ii
mik II
1
• n
t
* i I
I
ii
nesh II
	 II

-------

-------
*

-------

-------
T

-------
*

-------

-------

-------

-------

-------

-------

-------
-

-------

-------
-

-------
T

-------

-------

-------

-------

-------

-------

-------

-------

-------

-------

-------

-------

-------

-------

-------

-------

-------

-------

-------
/

-------

-------

-------

-------

-------
     V  =
      +          V  =
                                   V  =
V  =
      +       V  =
                                 V  =
                     +   I      v  =

-------
=  I       V  =

-------

-------

-------
60.65
60.85

-------

-------

-------

-------

-------

-------

-------

-------

-------
0.00

-------

-------

-------

-------

-------