COOPERATIVE GROUNDWATER REPORT 7
ILLINOIS STATE WATER SURVEY
ILLINOIS STATE GEOLOGICAL SURVEY
Champaign, Illinois 61820
PROCEDURES FOR THE COLLECTION OF
REPRESENTATIVE WATER QUALITY DATA
FROM MONITORING WELLS
James P. Gibb, Rudolph M. Schuller,
and Robert A. Griffin
Prepared in cooperation with
Municipal Environmental Research Laboratory
U.S. Environmental Protection Agency
STATE OF ILLINOIS
DEPARTMENT OF ENERGY AND NATURAL RESOURCES
1981

-------
PROCEDURES FOR THE COLLECTION OF REPRESENTATIVE
WATER QUALITY DATA FROM MONITORING WELLS
James P. Gibb, Rudolph M. Schuller, and Robert A. Griffin
STATE WATER SURVEY	STATE GEOLOGICAL SURVEY
In Cooperation with
Municipal Environmental Research Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
COOPERATIVE RESOURCES REPORT
CHAMPAIGN, ILLINOIS
1981

-------
STATE OF ILLINOIS
HON. JAMES R. THOMPSON, Governor
DEPARTMENT OF ENERGY AND NATURAL RESOURCES
FRANK H. BEAL, M.U.P., Director
BOARD OF NATURAL RESOURCES AND CONSERVATION
Frank H. Beal, M.U.P., Chairman
Walter E. Hanson, M.S., Engineering
Laurence L. Sloss, Ph.D., Geology
H. S. Gutowsky, Ph.D., Chemistry
Lorin I. Nevling, Ph.D., Forestry
William L. Everitt, E.E., Ph.D.,
University of Illinois
John C. Guyon, Ph.D.,
Southern Illinois University
STATE WATER SURVEY
Stanley A. Changnon, Jr., Chief
STATE GEOLOGICAL SURVEY
Robert E. Bergstrom, Acting Chief
Printed by authority of the State of Illinois/3000/1981

-------
FOREWORD
This publication summarizes the results of research conducted by the
State Water Survey and State Geological Survey. The research was funded (Con-
tract No. R-806304-010) in part by the Solid and Hazardous Waste Research
Division (SHWRD), Municipal Environmental Research Laboratory, Cincinnati,
Ohio. Project officers were Donald Sanning of SHWRD, and Marion R. Scalf of
the R. S. Kerr Environmental Research Laboratory, U.S. Environmental Protec-
tion Agency, Ada, Oklahoma.
The report has been reviewed by the U.S. Environmental Protection Agency
and released for publication as a combined report. Approval does not signify
that the contents necessarily reflect the views and policies of the U.S.
Environmental Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use by any of the re-
spective agencies.
The research summarized in this report is one component of the technical
basis required to operate meaningful groundwater monitoring systems. As such,
it should be useful to regulatory agencies, consultants, industry, and other
researchers. Additional information on this study can be obtained by contact-
ing the project officers or authors.
James P. Gibb
Head, Groundwater Section
Illinois State Water Survey
Rudolph M. Schuller
Assistant Geochemist
Illinois State Geological Survey
Robert A. Griffin
Head, Geochemistry Section
Illinois State Geological Survey
i i i

-------
ABSTRACT
Data collected from six monitoring wells at waste disposal sites in Illi-
nois were used to evaluate procedures for the collection of "representative"
water samples from monitoring wells. The effects of four types of pumping
mechanisms, the extent of well flushing, the rate and time of pumping, and
storage and preservation techniques on chemical composition of water samples
were studied. Pump tests and multiple sample experiments provided data on
which recommended sampling protocols and sample preservation, preparation, and
storage procedures are based.
The selection of the type of sampling device; sample preparation, preser-
vation and storage; and sampling procedures all must be tailored to the size
and accessibility of the individual well, its hydrologic and chemical charac-
ter, the chemical constituents of interest, and the purpose for monitoring the
site. Generally recommended sampling procedures include: 1) using peristal-
tic or submersible diaphragm type pumps when possible; 2) conducting pump
tests prior to sampling to determine sampling frequency and time and rate of
pumping; 3) flushing the well by pumping four to six well volumes; 4) measur-
ing key parameters such as pH, specific conductance, oxidation-reduction po-
tential, and alkalinity at the time of sample collection; and 5) filtering
samples immediately on site through a 0.45 ym pore size filter and then pre-
serving the samples immediately after filtration, according to U.S. EPA (1979)
recommended procedures.
iv

-------
CONTENTS
Page
Foreword		ii
Abstract	iii
Figures 		v
Tables	vii
Acknowledgments 		viii
1.	Introduction						i
2.	Conclusions		5
3.	Recommendations		7
4.	Site Selection and Descriptions		9
5.	Pumping Equipment		17
6.	Sample Collection and Preparation. ..............	ig
7.	Chemical Analytical Methods		21
8.	Factors Controlling Groundwater Quality		22
9.	Results		26
Pump test analyses		26
Effects of pumping mechanisms on chemical composition . .	35
Effects of well flushing on chemical composition		43
Effects of sample preparation, preservation and
storage on chemical composition 		53
References				60
v

-------
FIGURES
Number	Page
1	Generalized stratigraphic columns for sites 1-SDV, 2-ELG,
3-FLR, 4-TYL, and 6-DUP 	 10
2	Typical well and air-lift pumping mechanism (after Walker,
1974)	 18
3	Distribution of species for the C02"HC03~-C03~2 system
in water (Manahan, 1972)	 24
4	Site 1-SDV: time-drawdown data and theoretical curve 		30
5	Aquifer yield curves for all six sites	30
6	Site 2-ELG: time-drawdown data and theoretical curve 		31
7	Site 3-FLR: time-drawdown data and theoretical curve 		31
8	Site 4-TYL: time-drawdown data and theoretical curve 		31
9	Site 6-DUP: time-drawdown data and theoretical curve 		31
10	Percent of aquifer water versus time for different
transmissivities	34
11	Percent of aquifer water versus time for different well
casing diameters	34
12	Effects of pumping mechanism on iron concentrations at
site 5-BRD as a function of well volumes pumped	38
13	Effects of pumping mechanism on iron concentrations at
site 2-ELG as a function of well volumes pumped	39
14	Effects of pumping mechanism on iron concentrations at
site 6-DUP as a function of well volumes pumped	40
15	Site 1-SDV: magnesium, cadmium and manganese concentrations
versus volumes pumped (peristaltic pump)	44
16	Site 1-SDV: selenium, arsenic, boron and copper concen-
trations versus volumes pumped (peristaltic pump) 	 45
vi

-------
Figures (cont'd)
Number	Page
17	Site 2-ELG: magnesium, potassium, sodium, and iron con-
centrations versus volumes pumped (peristaltic pump)	46
18	Sites 3-FLR and 4-TYL: sodium, calcium, magnesium and
manganese concentrations versus volumes pumped (peristaltic
pump)	47
19	Site 5-BRD: potassium, magnesium, manganese, zinc, and
iron concentrations versus volumes pumped (peristaltic
pump)	48
20	Site 5-BRD: iron and zinc concentrations versus volumes
pumped for two sampling periods (peristaltic pump)	49
21	Site 6-DUP: iron concentrations versus volumes pumped for
two sampling periods (peristaltic pump) 	 52
22	Sites 4-TYL, 5-BRD, and 6-DUP: chloride concentrations
versus volumes pumped (peristaltic pump)	54
vii

-------
TABLES
Number	Page
1	Typical detection limits for ICP	 21
2	Redox processes in an open system	25
3	Values of the function F(u,a) 	 27
4	Average pH values for the sixth, eighth, and tenth well
volumes collected at each site by each pumping mechanism	36
5	Analysis of samples from sites 5-BRD and 6-DUP collected
by four pumping mechanisms immediately after flushing ten
well volumes	42
6	Analysis of samples from site 2-ELG where the well was
placed in geologic materials of low hydraulic conductivity. ... 55
7 Analysis of samples from site 4-TYL where the well was
placed in geologic materials of low hydraulic conductivity. ... 55
8	Sites 2-ELG, 3-FLR, 5-BRD, and 6-DUP: calcium, magnesium
and sodium concentrations of samples filtered through
different pore size membranes	57
9	Sites 5-BRD and 6-DUP: iron and zinc concentrations of
samples filtered through different pore size membranes	58
10 Sites 5-BRD and 6-DUP: analysis of the tenth well volume
sample used for storage study	59
vi i i

-------
ACKNOWLEDGMENTS
This project was funded in part by the U.S. Environmental Protection
Agency, Solid and Hazardous Waste Research Division, Municipal Environmental
Research Laboratory, Cincinnati, Ohio (Grant No. R-806304-010). Special
thanks go to Don Sanning and Dick Scalf, Project Officers, for their assist-
ance during the project.
The authors wish to express their gratitude to the operators of the four
landfills and two industries that permitted studies to be conducted on their
property.
Special thanks go to Michael O'Hearn, Assistant Engineer at the Illinois
State Water Survey, for his assistance in conducting pumping tests, collecting
samples, and his work with the various types of pumps used in the study. Bill
Bogner of the Illinois State Water Survey and Barry Fisher, Bill Roy, and Ivan
Krapac of the Illinois State Geological Survey also assisted on various sam-
pling runs. The authors also wish to thank Beverly Herzog and Keros Cart-
wright for contributing the geologic descriptions of the sites studied during
this project.
Use of the inductively coupled argon plasma emission spectrograph was
under the general supervision of Dr. Ken Smith, formerly of the Illinois State
Natural History Survey, and is gratefully acknowledged.
ix

-------
SECTION 1
INTRODUCTION
Regulatory agencies are charged with the assignment of regulating the
disposal of waste to insure that the environment is not adversely affected.
To accomplish this task, it is necessary for these agencies to set design and
operational standards based on available technology to minimize potential pol-
lution. The operating disposal facilities must then comply with these stand-
ards. They also must monitor the effects of their operation on the surround-
ing environment. The use of wells or piezometers for collecting water samples
and water level data has been, and probably will continue to be, the method
for monitoring the effects of waste disposal facilities on groundwater.
Considerable research has been conducted to develop analytical laboratory
techniques to detect the low levels of various constituents set forth in water
quality standards. Water sample collection and preservation techniques have
been established by several different laboratories and agencies in an attempt
to insure that water samples delivered to the laboratory are chemically repre-
sentative of the water sampled. However, there is considerable controversy
among laboratories, agency policies, and researchers concerning proper tech-
niques of sampling from monitoring wells and appropriate procedures for pre-
serving the original chemical character of the samples. If monitoring wells
and water samples are to provide the performance yardstick for disposal facil-
ities' design and operation, the significance of the various sampling proce-
dures and preservation techniques must be determined.
PREVIOUS WORK
Definite protocols can be established for collecting "representative"
water samples from monitoring wells. First, it is necessary to select a pump-
ing mechanism that is suitable to the physical limitations of the well (depth,
diameter, and yield potential) and that not alter the chemical properties of
the water during pumping. It also is necessary to determine when the water
being pumped is representative of water contained in the aquifer and is not a
mixture of aquifer water and unrepresentative "stagnant" water removed from
storage in the well casing.
Once the sample has been taken, it must be decided whether key parameters
will be measured on-site at the time of collection or in the laboratory some
time later. If the sample is to be returned to the laboratory for analysis,
the appropriate techniques of preparation, preservation, and storage must be
selected to prevent the chemical composition of the sample from changing be-
tween the time of collection and the time of analysis. If the samples are to
1

-------
be returned to the laboratory without preparation or preservation, the changes
in chemical composition occurring during sample storage must be determined.
The choice of pumping mechanism usually is dependent on the diameter of
the well, volume of water in the well, depth to water, yield potential of the
well, and accessibility of the site. Mooij and Rovers (1975) suggested that a
bailer was the best overall water sampling device. However, as Fenn et al.
(1977) pointed out, removing a large volume of stagnant water from a well with
a bailer can become tedious. It also can become difficult to obtain and meas-
ure a steady pumping rate when using a bailer. Using a peristaltic pump elim-
inates these problems but requires a power source which could cause difficul-
ties when sampling in more remote areas. Peristaltic pumps also are limited
in their lift capacities to a theoretical maximum of 9.75 m (32 ft), making
sampling from wells with deeper water levels impossible. Although submersible
pumps are designed for sampling at greater depths, these are not usually con-
structed to fit in a small-diameter monitoring well (diameter <2 in.). Walker
(1974) designed a sampling device that uses an air compressor or bottled com-
pressed gas to lift a water sample from a monitoring well at greater depths.
Sommerfeldt and Campbell (1975) and Trescott and Pinder (1970) have used simi-
lar pumping mechanisms employing gasoline-powered air compressors to sample
from as deep as 190 feet. Mooij and Rovers (1975) stated that the effect of
the air-lift pumping mechanisms on the chemical composition of a sample col-
lected with an air-lift system was unknown. Wood (1970) and Summers and
Brandvold (1967) have, however, cautioned against the exposure of groundwater
samples to air. Wallich (1977) showed that exposure of groundwater to the
atmosphere resulted in erroneous values for pH, alkalinity, and iron in
groundwater samples. Although several studies have been directed toward
studying the effects of exposing a groundwater sample to atmospheric condi-
tions, little has been done toward studying the effects of sample collection
systems on the chemical composition of groundwater samples.
After selecting the type of pumping mechanism to use, the extent of well
flushing needed to clear the stagnant storage water from the well must be
determined. Summers and Brandvold (1967) were among the first to document the
need to flush a well before collection of a groundwater sample. Although
their work involved high volume flowing wells and not shallow monitoring
wells, the results showed that pH, temperature, and specific conductance grad-
ually changed with increasing volumes of water pumped until steady values were
obtained. Schmidt (1977) also established the need for flushing large capac-
ity wells, but like the previous researchers ventured no estimate of the vol-
ume of water to be removed before a representative sample of aquifer water
could be taken.
Several investigators have attempted to establish sampling routines for
shallow monitoring wells. Hughes et al. (1968) noted the need for flushing a
monitoring well before collection of groundwater samples and recommended that
two well volumes be pumped. (Well volume constitutes the initial volume of
water in the well casing prior to pumping.) Gilkeson et al. (1977) found that
reductions in Fe, Pb, Zn, Cu, Cd, and Cr concentrations occurred in collected
samples after well flushing. For wells which cannot be pumped dry, Mooij and
Rovers (1975) recommend pumping five well volumes before collecting a sample.
For wells installed in materials of low hydraulic conductivity, they further
2

-------
recommend that the well should be pumped dry arid allowed to refill before
sampling.
Fenn et al. (1977) developed a comprehensive scheme for well flushing,
dependent upon the yield potential and the positioning of the pump intake, for
wells that could not be pumped dry. They recommended a minimum flushing of
one well volume, but preferably three to five well volumes. Mooij and Rovers
(1975) and Fenn et al. (1977) recognized the need for well flushing and the
dependence of its magnitude on the hydrogeology of the site.
Other considerations for proper sampling are the procedures for sample
preparation, preservation, and storage. Hughes et al. (1968), Brown et al.
(1970), Chian and DeWalle (1975), and Jackson and Inch (1980) all suggested
that certain parameters should be measured on-site at the time of sample col-
lection, since parameters such as pH, specific conductance, temperature,
oxidation-reduction potential, and dissolved oxygen are all likely to undergo
change before the samples can be transported to the laboratory for analysis.
Once key or target parameters have been determined, the samples must be
prepared and returned to the laboratory for analysis. The United States Envi-
ronmental Protection Agency (1979) recommends filtering groundwater samples
through a 0.45 pm pore size membrane on-site immediately after collection
unless field filtration is impractical. Field filtration, like measurement of
the key parameters, requires transporting additional equipment to the sample
collection site. If returning the samples to the laboratory without filtra-
tion and preservation would result in no chemical changes, there would be a
considerable savings in time and expense. This savings, however, must not be
made at the expense of the chemical integrity of the samples. Sample preser-
vation by acidification before filtration is well recognized as causing in-
creased concentrations of clay-related and other colloidal constituents.
The use of the 0.45 ym pore size membrane for filtration of environmental
water samples has been universally accepted. However, Kennedy et al. (1974)
demonstrated that clay sized particles could pass through 0.45 urn pore size
membranes, resulting in increased concentrations of Al, Fe, Mn, and Ti. In a
similar study, Wagemann and Brunskill (1975) saw an increase in Al and Fe con-
centrations in solutions filtered through 0.45 ym pore size membranes, but no
changes in Mn and Ti. The use of smaller pore sized membranes would reduce
the passage of clay sized particles, but also could greatly increase the fil-
tration time for turbid samples.
Once the samples have been filtered they must be split into aliquots and
preserved. Struempler (1973), Rattonetti (1976), and Subramanian et al.
(1978) found it necessary to acidify aqueous samples to a pH <2.0 to avoid
loss of cations from solution by precipitation and adsorption on container
walls. The U.S. EPA (1979) has established techniques for the preservation of
most constituents found in waters and waste waters. Struempler (1973) and
Shendrikar et al. (1975) were among many others who have studied the adsorp-
tion characteristics of materials used for sample containers. They found that
linear polyethylene and glass were the best storage vessels for aqueous sam-
ples for which inorganic constituents were to be determined.
3

-------
PURPOSE AND SCOPE OF STUDY
The three principal purposes of this study were: 1) to determine if cur-
rent sampling methods produce samples that are representative of water con-
tained in the aquifer or water-bearing strata being monitored; 2) to determine
if groundwater samples collected in the field must be treated (filtered and
acidified) on location, or if they can be brought back to the laboratory for
treatment without altering their chemical nature; and 3) to determine which
sampling and preservation techniques should be accepted as standards for moni-
toring well sampling. In order to fulfill these purposes, the following spe-
cific objectives were set:
1)	Determine the hydrologic properties of the materials tapped by each
monitoring well studied.
2)	Determine a pumping scheme for each well to obtain water samples
representative of the aquifer or water-bearing strata being moni-
tored .
3)	Collect a series of samples from each well using four different pump-
ing methods.
4)	Determine the effects that the pumping mechanism, time and rate of
pumping, and preservation techniques have on the chemical composition
of the samples collected.
5)	Recommend monitoring well sampling procedures and sample preservation
techniques for specific chemical constituents.
Because of the enormous quantity of data produced by the above scheme,
certain sampling practices had to be accepted. All of the samples were pre-
served according to the U.S. EPA's (1979) recommended procedures. Also, all
of the samples were stored in linear polyethylene bottles. These practices
were well documented in the literature as not being detrimental to a sample's
integrity for short time periods. During the collection, preparation, preser-
vation, and storage of the samples, no new procedures or equipment were used.
The purpose of this project was to analyze existing procedures and equipment
that were available.
4

-------
SECTION 2
CONCLUSIONS
The results of this study make it apparent that collecting "representa-
tive" water samples from monitoring wells is not a straightforward or easily
accomplished task. Each monitoring well has its own individual hydrologic and
chemical character that must be considered when planning a sampling protocol.
The selection of the type of sampling device; sample preparation, preserva-
tion, and storage; and sampling procedures all must be tailored to the size
and accessibility of the individual well, its hydrologic and chemical charac-
ter, the chemical constituents of interest, the time of year, and the purpose
for monitoring.
It has been demonstrated that meaningful pump tests can and should be
conducted on monitoring wells. The analyses of time-drawdown data from con-
stant rate pump tests using the equations developed by Papadopulos and Cooper
(1967) yielded realistic transmissivity values for the materials tapped by the
monitoring wells studied. The determined transmissivities were used to pro-
ject time-drawdown relationships at various pumping rates and to determine the
percent of "aquifer water" contained in the total amounts pumped at any given
time. These relationships should be used as guides for obtaining "representa-
tive" water samples from monitoring wells.
The transmissivity values are also invaluable when attempting to deter-
mine times of travel of groundwater within the aquifer or materials being
monitored. By applying appropriate hydraulic gradients, porosities, and
transmissivities, the Darcy and real velocities of groundwater movements can
be estimated. These velocities can be used to project the rate of migration
of pollutants detected in the monitoring well and to describe the geometry and
dynamics of the contaminate plume. The velocity of regional groundwater flow
should be used to determine realistic and economic frequencies for sample col-
lection. In very "tight" materials (those with low transmissivities) the
rates of groundwater movement may be on the order of 0.2 to 1.5 meters (1 to 5
feet) per year. If monthly sampling is prescribed in a case such as this, the
same "slug" of water may be sampled for 2 or 3 months consecutively.
Chemical data from samples collected with the four types of pumps used in
this study indicated that peristaltic pumps and bailing yielded comparable
data and resulted in the least changes in chemical quality of water delivered
to the surface. Air- and nitrogen-lift pumping mechanisms increased the pH of
water samples during pumping and altered the concentrations of several chemi-
cal constituents. Iron and zinc were shown to be particularly sensitive to
the use of these types of pumps.
The effects of flushing or pumping a well for a period of time to insure
collection of a "representative" sample also have been effectively documented.
5

-------
In most cases, the water stored in the well casing was of different chemical
quality than that contained in the aquifer to be sampled. Usually, the oxi-
dizing environment in the well was sufficiently different from that of the
aquifer to create a shift of chemical species in solution.
To insure that a "representative" water sample is collected, the well
should be pumped until a high percentage of "aquifer water" is obtained. The
length of time of pumping depends on the rate of pumping, well diameter, and
transmissivity of the aquifer being sampled. Given these factors, the aquifer
water percentages with time can be calculated from the examples in this report
and used as a guide for determining the appropriate pumping time before a
sample is collected. Monitoring pH while pumping with a peristaltic pump or
bailer appears to be a reasonable field check for assuring that a "representa-
tive" water sample is being collected. For monitoring wells that can easily
be pumped dry, the limited data collected in this study suggest that repre-
sentative samples can be taken by pumping the well dry and then collecting the
sample as the well refills. To insure that the collected samples are truly
"fresh" aquifer water flowing into a dewatered well, it is suggested that the
water levels in the well be monitored and the sample collected while the water
levels are still rising.
Data collected from the same monitoring wells at different times of the
year show that significant seasonal variations in chemical quality can occur.
These changes can be related to varying rates of recharge and changes in
oxidation-reduction conditions in relatively shallow aquifer systems. Aware-
ness of the possibilities of these types of seasonal variations is essential
to understand and properly interpret the significance of changes in water
quality with time.
Changes in field-measured pH or specific conductance values from one sam-
pling period to another should be expected and are not causes to abandon
established sampling procedures. To insure that "representative water sam-
ples" are obtained, the time and rate of pumping should be the same each time
a particular well is sampled.
Data from the sample preparation, preservation, and storage portions of
the study show that when chemical concentrations for certain constituents are
desired, samples definitely should be filtered in the field at the time of
collection. Chemical constituents sensitive to pH changes can be affected
within 7 hours if not filtered and preserved at the time of collection. The
constituents found to be most sensitive were Fe and Zn.
In addition, the selection of filter pore sizes used to filter samples
can affect the results of the chemical analyses. Mineral constituents asso-
ciated with clay and colloidal particles increased significantly when the sam-
ples were filtered through 3.0 ym pore size membranes. Samples filtered
through 0.45 ym pore size membranes had only slightly higher clay and colloi-
dal particle-related mineral values over those filtered through 0.22 ym pore
size membranes. For practical purposes, the differences in chemical composi-
tion were small and the use of the 0.45 ym pore size membrane appears satis-
factory. Use of the smaller pore size filters was more time consuming and
resulted in filter clogging problems, particularly for turbid samples.
6

-------
SECTION 3
RECOMMENDATIONS
On the basis of this study the following recommendations are made:
1)	A brief 2 or 3 hour pumping test should be conducted on each monitor-
ing well to be sampled. Analyses of the pump test data and other
hydrologic information should be used to determine the frequency at
which samples will be collected and the rate and period of time each
well should be pumped prior to collecting the sample.
2)	The general rule of thumb of pumping 4 to 6 well volumes will in most
cases produce samples representative of aquifer water. For aquifers
with unusually high transmissivities, pumping for periods long enough
to remove the "stagnant" water column may induce migration of water
from parts of the aquifer remote from the monitoring well. The cal-
culations of percent aquifer water with time provide a more rational
basis on which the length of pumping can be determined. Samples
should be collected in the minimum time required to produce water
representative of the aquifer.
3)	A controlled sampling experiment, similar to those in this study,
preferably using a peristaltic or submersible diaphragm type pump,
should be conducted to accurately determine the chemical quality of
the aquifer water and to verify the response of the monitoring well
to pumping as predicted from the pump test data. Once the chemical
character and responses of the monitoring systems have been deter-
mined, key chemical constituents for routine sampling can be select-
ed.
4)	Based on the sensitivity of the selected chemical parameters, a
choice of pumps for routine sampling can be made. The use of air- or
nitrogen-lift pumping mechanisms should be restricted to chemical
constituents insensitive to oxidation-reduction reactions and changes
in pH. Although this study dealt with inorganic constituents, the
data suggest that these types of pumping mechanisms probably would
also strip volatile organic compounds from the water during pumping.
The peristaltic or submersible diaphragm pumps and the bailer are
recommended for most applications. If a bailer is to be used, the
procedures outlined in the results sections of this report should be
fol 1 owed.
5)	The monitoring well should be pumped at a constant rate for a period
of time that will result in delivery of at least 95 percent aquifer
7

-------
water. The rate and time of pumping should be determined on the
basis of the transmissivity of the aquifer, the well diameter, and
the results of the sampling experiment.
6) Measurements of pH, Eh, and specific conductance should be made at
the time of sample collection. These measurements should be made
within a closed cell, which will prevent the sample from coming into
contact with atmospheric conditions. All samples should be promptly
filtered through a 0.45 pm pore size membrane and preserved according
to recommended U.S. EPA procedures for the chemical constituents of
interest.
In addition to the specific recommendations for establishing sampling
procedures and sample preparation and preservation protocols noted above, the
following general recommendations and suggestions for additional research
needs are offered:
1)	Studies should be conducted to develop optimized sampling procedures
for monitoring groundwater for organic compounds.
2)	Additional studies are needed to develop sampling procedures for geo-
logic strata and materials of very low hydraulic conductivity and for
monitoring wells that can easily be pumped dry.
3)	Studies also should be undertaken to investigate the effects of sea-
sonal and other natural variations in chemical quality of groundwater
on sampling protocols and the interpretation of data from monitoring
wel1s.
4)	Studies are also needed to develop in-situ sampling procedures and
apparatus to minimize the possible artificial introduction of contam-
inants into the well during the monitoring procedure itself.
8

-------
SECTION 4
SITE SELECTION AND DESCRIPTIONS
Monitoring wells at six sites in the State of Illinois were selected for
study. Two sites are active sanitary landfills; two are inactive landfills;
one is a stack scrubber lagoon for a secondary zinc smelter; and one is an
anaerobic digestion lagoon for a hog processing plant. The wells at all sites
were installed prior to this study, and all but one were constructed by per-
sons other than the authors. Five are cased with PVC pipe, either 3.81 or
5.08 cm (1-1/2 or 2 in.) diameter. The sixth well is cased with 5.08 cm
(2 in.) diameter galvanized iron. The well depths range from about 5 to 10 m
(16 to 30 ft) and have nonpumping water levels from 0 to 5 m (0 to 16 ft)
below land surface.
Factors considered in selecting the sites included: 1) accessibility of
the monitoring well to vehicles so that pumping tests and sampling could be
accomplished with equipment using generated electricity without undue hard-
ship; 2) the physical characteristics of the monitoring well and geology of
the location; 3) the potential yield capability of the materials tapped by the
monitoring well; and 4) the chemical quality of water obtained from the moni-
toring wel 1.
SITE 1 - SANDOVAL (1-SDV)
Site 1-SDV is a secondary zinc smelter located in south-central Illinois.
It has been in operation since about 1885. In compliance with air pollution
control regulations, a scrubber was installed on the emission stack at the
plant in 1970. The scrubber waste water is disposed of in a surface impound-
ment built on a cinder fill derived from early smelting operations. A series
of monitoring wells were constructed on the plant property in 1974 and 1975
(Gibb et al., 1978). The monitoring well chosen for use in this study was
well no. 12, a 5.08 cm (2 in.) diameter well 4.25 m (14.0 ft) deep. The well
is cased with PVC pipe and has slots sawed in the bottom 61.0 cm (2 ft). The
well is located about 4.6 m (15 ft) downgradient from the scrubber waste dis-
posal pond. Figure 1 shows the material encountered at this site.
The surficial materials at the site are wastes from early smelting opera-
tions. These wastes are cinders and wind-blown ash which are rich in zinc and
other heavy metals and which range in thickness from 0.305 to 3.05 m (1 to
10 ft) on the Sandoval property.
Glacial material that underlies the cinders is about 18.3 m (60 ft) at
the well and thickens to 22.9 m (75 ft) on the west side of the smelter prop-
9

-------
rO 0/>
b°£&
=> 'OyO
->q
o O •
3o''»
'^f.0
K Q<».'
)°0"
r---;
0 °°/5 ^
Ci y •
Site 1 - SDV
Cinder Fill
Peoria Loess
Roxana Silt
Berry Clay Member
Hagarstown Member*
Glasford Till Member
Lierle Clay Member
Banner Formation Till
Bond Formation
z-r.oQ^
Oa-0„
°.^°6
yo.:» ^
cfifi ?:¦<$!
eo;o--:'
m
7 7 7
XX
Site 2 — ELG
Clean fill
Mackinaw Member
Tiskilwa Till Member
Outwash*
Silurian dolomite
*0 « ,
•K°C$
C>°oc
'
0 ®0o
•Oo«j
A'C^0

Site 3 — FLR
Peoria Loess
Sangamon Soil * (?)
Vandalia Till Member
Pennsylvanian sandstone
»V.
¦ oo •
rti
ffr
Site 5 — BRD
Surface soil
Wisconsinan outwash*
Sankoty Sand
Keokuk—Burlington
Limestone
S T *
R*>;
j \
v
'a-
i
l.' -PJ. °
Site 6 — DUP
cover material
Equality Formation*
Yorkville Till Member
Maiden Till
* Stratigraphic unit in which the monitoring wells are finished
Figure 1. Generalized stratigraphic columns for sites 1-SDV,
2-ELG, 3-FLR, 5-BRD, and 6-DUP
10

-------
erty. The stratigraphic units recognized in the drift are essentially flat
lying, uniform in character, and mostly of low hydraulic conductivity.
The uppermost stratigraphic unit in the glacial drift is 1.2-1.8 m (4-6
ft) of a brownish-gray clayey silt called the Peoria Loess. Expandable clay
(montmorillonitic) material content averages nearly 85 percent in the loess.
The remaining material is predominantly illite and sand. A modern soil pro-
file has developed in the loess.
Less than 1.5 m (5 ft) of Roxana Silt, a dark-brown clayey silt, under-
lies the Peoria Loess. While the clay mineral content is similar to that of
the loess above, the sand content averages 20 percent and may be as much as
34 percent. The lower part of the modern soil profile is probably in the
Roxana Silt.
The Glasford Formation lies below the Roxana Silt and has three identifi-
able members in this area. The uppermost member, the Berry Clay Member, con-
sists of 0.9-1.5 m (3-5 ft) of dark gray silty sandy clay with a trace of
gravel. The clay mineral content is similar to that of the overlying loess
units, but the sand and gravel percentages increase toward the base. The clay
is rich in organic material and iron staining but is lacking in carbonates.
The Hagarstown Member, below the Berry Clay, is the most permeable zone.
It consists of 0.3-0.6 m (1-2 ft) of silty sand with some gravel. Its clay
content is similar to that found in the Glasford Formation Till below; the
sand content is variable and may be as much as 50 percent. The monitoring
well used for the study is finished in the Hagarstown Member, which is 45 cm
(18 in.) thick at this location.
The Glasford Formation Till comprises approximately 6.1-13.1 m (20-43 ft)
of gray to dark gray sandy and silty glacial till. Its thickness is approxi-
mately 9.8 m (32 ft) at the test site. Both illite and sand content decrease
with depth. Lenses of dark olive-brown leached clay are locally present, as
are discontinuous lenses of sand and silt. The Sangamon Soil was formed in
the two upper members of the Glasford Formation and in the upper part of the
Glasford Formation Till.
The lowest glacial unit present is the Banner Formation, which contains
the Lierle Clay Member and the Banner Formation Till. A dark olive-brown
silty clay, the Lierle Clay is less than 1.2 m (9 ft) thick. Sand content
averages 24 percent, while clay mineral composition is variable. The Banner
Formation Till is 3.0-8.8 m (10-29 ft) of gray to pinkish-gray sandy, silty
clay till with some gravel. Carbonates are present except locally at the top.
Shale fragments and discontinuous sand and silt lenses also occur.
The Bond Formation of Pennsylvanian age forms the bedrock at this site.
It is a green shale containing abundant mica. Kaolinite and illite content is
high, while montmorillonitic clay content is low.
11

-------
SITE 2 - ELGIN (2-ELG)
Site 2-ELG is an inactive landfill now used as a county forest preserve.
The site originally was a gravel pit, and filling began in 1948 as an open
burning dump. In later years controlled filling and covering was undertaken
until the site was closed in about 1970. Forty percent of the fill material
reportedly is household and garden refuse; 60 percent is industrial waste
(Hughes et al., 1971).
The well chosen for use in this study is LW1B constructed in 1967 by
Hughes et al. (1971). The well is 3.81 cm (1-1/2 in.) in diameter and 7.92 m
(26 ft) deep). The well is cased with PVC pipe and has a 91.4 cm (3 ft)
screen at depth between 7.01 and 7.92 m (23 and 26 ft). The well is located
about 76.2 m (250 ft) downgradient from the eastern edge of the old landfill
and within 15 m (50 ft) of the Fox River.
The materials encountered at this site include clean fill and glacial
drift. These materials dip eastward west of the monitoring well, but are
relatively flat-lying at the site. Depth to bedrock is approximately 15 m
(49 ft). Figure 1 shows the sequence of materials encountered.
The surficial material (Mackinaw Member) consists of 0.6-0.9 m (2-3 ft)
of clean fill material. The fill is a non-calcareous, clayey to sandy silt.
It was most likely derived from the brown to black silt and fine sand that
originally formed the top soil in the area.
Below the Mackinaw Member is 3.7-6.4 m (12-21 ft) of the Tiskilwa Member
of the Wedron Formation. The Tiskilwa Till Member is a calcareous, pinkish-
gray to reddish-brown clay loam till. Gravel is present at scattered loca-
tions. At the base of the till, 1.8-4.6 m (6-15 ft) of medium to coarse
gravel outwash with abundant dolomite is present at most locations. The moni-
toring well used in this study is finished in this outwash at the base of the
Tiskilwa Till Member above the bedrock.
Silurian dolomite forms the bedrock in the Elgin area. It is fractured
and used as an aquifer in the area.
SITE 3 - FLORA (3-FLR)
Site 3-FLR is an active landfill located in the clay pan area of southern
Illinois. Landfilling operations began in about 1976. The materials being
buried are principally household and garden refuse. No significant industrial
waste is permitted for disposal at this site.
The well chosen for study is the northeast well being used for routine
monthly monitoring in accordance with Illinois EPA guidelines. The well is
5.08 cm (2 in.) in diameter, is 5.88 m (19.30 ft) deep, and has a 61.0 cm
(2 ft) screen between depth of 5.27 and 5.88 m (17.30 and 19.30 ft). It is
cased with PVC pipe.
12

-------
No driller's log is available for the well and there are few well logs
from the area to show the geology of the site. The well logs available for
the surrounding area are driller's logs and are very sparse in detail.
Available data indicate that the area is characterized by a loess and a
till unit overlying a sandstone bedrock. The total drift thickness is less
than 15 m (50 ft) thick. Figure 1 shows the materials found in the vicinity
of the site.
The surficial deposits consist of an average of about 4 m (13 ft) of
wind-blown Peoria Loess. It is described by drillers as yellow clay. Region-
al evidence suggests the loess i,s massive and contains the modern soil pro-
file.
Below the Peoria Loess there is assumed to be an average of 1.5 m (5 ft)
of Sangamon Soil. The Sangamon Soil is a weathered zone with a range of char-
acteristics similar to, but more highly developed than, a modern soil. It is
a sandy clay but may be coarser or finer-grained. The landfill operator re-
ported the monitoring well used in this study to be finished in a thin sand
layer (probably Sangamon Soil) overlain by clay and till.
The lowest geological unit present is the Vandalia Till Member of the
Glasford Formation. Near Flora, it is a medium-grained, relatively silty,
compact till. The density increased greatly toward the base. Thin sandy and
pebble zones are common. On the driller's logs, this unit is commonly re-
ferred to as blue clay or shale.
Pennsylvanian aged sandstone forms the bedrock in this area. The sand-
stone is the main source for small domestic wells in the area.
SITE 4 - TAYLORVILLE (4-TYL)
Site 4-TYL is an active landfill located in central Illinois. Landfill-
ing operations began in about 1975. The materials being buried reportedly are
about 70 percent household and garden refuse and about 30 percent light indus-
trial wastes.
The well chosen for use in this study is the "northwest" monitoring well
being used for monthly monitoring. The well is 5.08 cm (2 in.) in diameter
and 12.01 m (39.4 ft) deep. The well is cased with PVC pipe and has a 61.0 cm
(2 ft) screen between the depths of 11.40 and 12.01 m (37.4 and 39.4 ft).
The landfill is located in two geological terrains: the Shelbyville
Moraine and the edge of the flood plain of the South Fork of the Sangamon
River. The valley is eroded glacial till that has been filled in with outwash
and alluvium. The valley fill thickness is variable and contains large
amounts of sand. The sands generally lie between finer-grained sediments
above and older tills on bedrock below.
The upland area is mostly till covered by loess. Since the area is near
the terminus of the moraine, soils are erratic both in type and distribution.
13

-------
The soils encountered vary from mixtures of clays, silts, sand, and fine
gravel (glacial tills) to fine, silty sands of alluvial origin. The material
encountered in nearby borings all had a low permeability. Depth to bedrock
beneath the moraine ranges between approximately 7.6 and 46 m (25 and 150 ft).
The bedrock is a soft, dark shale of Pennsylvanian age.
No driller's log is available for the well and only one driller's log
showing the unconsolidated material was on record for the quarter section.
The recorded well was drilled in 1941 and was located on the moraine in the NE
1/4 SE 1/4 SE 1/4 Section 13, T. 13N., R.3W. The upper portion of its log
reads as follows:
From top to bottom these units most likely are Peoria Loess, Sangamon
Soil, Vandalia Till (both units described as hardpan) and Pennsylvanian shale.
The landfill operator reported the well to be finished in a slightly sandy
clay zone 12.01 m (39.4 ft) deep (probably Vandalia Till) overlain by clay and
till (probably Sangamon Soil), as illustrated by the sequence for site 3-FLR
shown in Figure 1.
SITE 5 - BEARDSTOWN (5-BRD)
Site 5-BRD is a hog processing plant located in western Illinois. Liquid
slaughter and processing wastes are routed through a 3-stage lagoon system and
then either irrigated onto grain crops or discharged to the Illinois River.
The well chosen for use in this study is well 11, used by the plant per-
sonnel to monitor the effects of their disposal operation on groundwater. The
well is 5.08 cm (2 in.) in diameter and 8.43 m (27.65 ft) deep. The well is
cased with galvanized iron pipe and reportedly has a 61.0 cm (2 ft) screen be-
tween the depths of 7.82 and 8.43 m (25.65 and 27.65 ft). The well is located
approximately 61 m (200 ft) downgradient from the first stage anaerobic la-
goon.
Driller's logs for the area indicate that the bedrock is more than 30 m
(100 ft) below the ground surface. The predominant unconsolidated deposits
are about 15 m (50 ft) of sand extending from the surface and 15 m (50 ft) of
gravel below. The character of the subsurface deposits is relatively uniform
and highly permeable. Figure 1 shows the materials expected at Beardstown.
Top soil was noted on only one log in the Beardstown area. No descrip-
tion was provided of this soil, although regional data suggest that it is
loess.
Thickness
Depth
Clay, yellow
Sand, gravel and clay
Hardpan and sandy clay
Hardpan
Shale, soft, dark
6.1 m (20 ft)	6.1 m (20 ft)
3.05 m (10 ft)	9.15 m (30 ft)
11.90 m (39 ft)	21.05 m (69 ft)
3.36 m (11 ft)	24.40 m (80 ft)
1.53 m (5 ft)	25.95 m (85 ft)
14

-------
Wisconsinan outwash, 10.7-24.3 m (35-80 ft) thick, is present beginning
at or near the surface throughout the area. The outwash is mainly medium to
coarse sand, although fine sand has been noted near the top or base of the
deposit. A discontinuous, relatively thin, clayey silt layer is found at
depths of 1.5 to 4.6 m (5-15 ft) below the land surface in much of the area.
The monitoring well used in this study is finished at a depth of 8.43 m
(27.65 ft) in the Wisconsinan outwash.
The Sankoty Sand beneath the outwash is at least 6.7-18.3 m (22-60 ft)
thick. No wells on file have penetrated the base of this sand in this area.
In many locations the sand is actually a fine to coarse gravel deposit with
some discontinuous lenses of fine to coarse sand. Large capacity wells (3790
to 5685 Lpm, or 1000 to 1500 gpm) are commonly constructed in this unconsoli-
dated aquifer system.
Mississippian-age Keokuk-Burlington Limestone forms the bedrock in the
area. Although no data are available on the limestone in this area, it is
generally crinoidal limestone interbedded with fine-grained limestone, argil-
laceous dolomite and calcareous gray shale.
SITE 6 - DU PAGE (6-DUP)
Site 6-DUP is an inactive landfill located in northeastern Illinois.
Filling by the trench and fill method began in 1952 and was completed by the
end of 1966 (Hughes et al., 1971). Household and garden refuse was the major
component of the fill, but small amounts of spent battery acid, construction
debris, and sewage sludge were also buried at the site. The maximum thickness
of this debris is about 6 m (20 ft).
The well chosen for study is MM 63, constructed in 1968 by Hughes et al.
(1971). The well is 3.81 cm (1-1/2 in.) in diameter and 5.12 m (16.8 ft)
deep. It is cased with PVC pipe and has a 30.5 cm (1 ft) screen at depths
from 4.82 to 5.12 m (15.8 to 16.8 ft).
The DuPage site lies on a flat upland area between two moraines. The
area was originally swampy and is drained by tiles into a creek flowing along
the east side of the property. The general sequence of earth materials at the
site consists of an upper surficial silty sand overlying till, which lies on
dolomite bedrock. The materials slope gently to the east. Figure 1 shows the
expected materials at DuPage.
The surficial material at the landfill site consists of 0.6-0.9 m (2-3
ft) of cover material comprised of clay loam, clay, silty clay loam and silt
loam. Around the landfill, silty clay loam and clay loam of the Richland
Loess constitute the surficial deposits.
Below the landfill is Equality Formation sand, which is probably outwash
from the eastern moraine. The upper sand is a sandy silt to a silty sand and
is up to 6.4 m (21 ft) thick below the landfill. This thickness decreases
toward all edges of the landfill. A sand and gravel bar is present in the
southeast corner of the landfill, and other bars are scattered in the field
15

-------
south of the till. The monitoring well used in this study penetrates the
landfill refuse and is completed in a 1.22 m (4 ft) thickness of the Equality
Formation sand immediately beneath the refuse.
Two tills of the Wedron Formation occur below the Equality Formation.
The upper till formation, the Yorkville Till Member, is a clayey, silty till
from 1.5 to 7.5 m (5-25 ft) thick. This deposit is similar to the predominant
surficial deposit present throughout northeastern Illinois. The Maiden Till
Member, below the Yorkville Till Member, has interbedded sand lenses between
two till units. The upper till is about 3.7 to 6.1 m (12-20 ft) of sandy
silty till. It is discontinuous beneath the landfill site.
Interbedded lenses of sand below the upper till consist of 0.5-1.5 m
(1.5-5 ft) of sand and fine gravel, and are also discontinuous at the site.
The lower till is a silty till, approximately 6 m (20 ft) thick, at the base
of the section.
Silurian-age fractured dolomite is the bedrock at the site. The dolomite
is the major aquifer in the area.
16

-------
SECTION 5
PUMPING EQUIPMENT
Four types of pumping methods were selected for collecting samples from
the monitoring wells. These methods included a peristaltic pump (and/or dia-
phragm type pump), an air-lift system, a nitrogen-lift system, and a bailer.
Most monitoring wells are commonly constructed using 3.81 or 5.08 cm (1-1/2 or
2 in.) diameter PVC pipe. These small diameters create severe limitations on
the selection of a pumping apparatus.
In wells where pumping water lifts were within suction lift capabilities,
a Masterflex 7545 variable speed drive unit equipped with a 7015 peristaltic
pump head was used. When operated on 115 volts (a portable generator), it is
capable of producing from about 50 to 1000 mL/min (0.013 to 0.264 gal/min).
For wells where pumping lifts were beyond suction lift capabilities, a
diaphragm type pump modeled after the Middleburg pump was constructed and
used. This type of pump is available cotmiercially, but on a limited basis.
The flow rate and lift capability from this type of pump is controlled by
varying the frequency of alternately applied and released pressure on the dia-
phragm and by regulating the pressure at which it operates. The pump con-
structed for this project is 45.72 cm (18 in.) long, is 3.18 cm (1-1/4 in.) in
diameter, and delivered water from 12.19 cm (40 ft) at a rate of 3500 mL/min
(0.92 gal/min). The pump was not tested at greater depths but should be lim-
ited only by the operating pressure.
For the air-lift and nitrogen-lift pumping systems, an apparatus similar
to that shown in Figure 2 was used. For some sampling runs, a 1.27 cm
(1/2 in.) diameter rigid PVC discharge pipe and a 0.635 cm (1/4 in.) plastic
airline were used. For others a 0.952 cm (3/8 in.) diameter flexible poly-
vinyl discharge line and 0.635 cm (1/4 in.) plastic airline were used. For
all air-lift runs, a four-cylinder electric driven air compressor was used.
For the operation of the nitrogen-lift pumping mechanism, cylinders of
compressed nitrogen gas were used to pressurize the identical apparatus used
in the air-lift pumping experiments. A pressure regulator and flow-meter were
used to control the pressure and flow rate to the gas line. Physical dimen-
sions affecting the operation and efficiency of the air-lift or nitrogen-lift
pumping mechanisms include: 1) the operating pressure should be 1 or 2 psi
greater than the hydrostatic pressure at the pump intake; 2) the rate of air
or nitrogen delivery to the system is dependent on the diameter of the dis-
charge pipe; and 3) this type of system generally will not operate efficiently
at a submergence of less than 1/3 of the total lift to be overcome.
17

-------
1/2" cap
(1 27 cm)
1/2" Tee
(5.08cm) 2" cap—~
2" casing —~
1/2" discharge
pipe
7" bore hole-
(17 78 cm)
slotted PVC
well casing"
1/2" pipe
1/2" cap
vent hole
Shrader valve

o o
o o
O 0
o o
°°«
o °
(7.62 cm) 3"
ooo
O O ®
. 0 o
0 o o
o o
o o
°n°fl
0 °#1
bentonite
slurry
1/4" airline
(0 635 cm)
6" sand (15 24 cm)
12" (30.48cm]
-gravel
Figure 2. Typical well and air-lift pumping mechanism
(after Walker, 1974)
For bailing, a 2.54 cm (1 in.) diameter stainless steel bailer, 0.914 m
(36 in.) long, was constructed. The bailer retrieved a sample of about
300 mL (0.08 gal).
For purposes of comparison, the vertical point of sample collection from
within the well for all pumping mechanisms was the mid-point of the well
screen. The pump intakes for the peristaltic, air-lift, and nitrogen-lift
systems were set in each well at the mid-point of the screen. When bailing,
the bailer rope was marked such that the bottom of the bailer would be lowered
to the same point in the well (the mid-point of the screen) during each suc-
cessive bai 1.
Controlled pumping tests were conducted on the six wells to determine the
hydraulic characteristics of the materials in which the wells were finished.
Pumping tests were conducted using a peristaltic pump. The pump rates were
determined by a graduated cylinder and stop watch. Water levels were measured
with a steel tape.
18

-------
SECTION 6
SAMPLE COLLECTION AND PREPARATION
Each of the six sites was sampled a maximum of once a month by one of the
four pumping mechanisms. This sampling scheme was used to study the effects
of the pumping mechanisms, well flushing, and filtration on sample composi-
tion. The frequency of sampling allowed ample time for the wells to recover
from the pumping that had occurred during the previous sample collection.
Samples were taken from the initial water in storage, designated as the
zero well volume, and at 1/2, 1, 1-1/2, 2, 4, 6, 8, and 10 well volume inter-
vals. One well volume is defined as the amount of water occupying the well
casing before pumping was initiated. The volumes of the samples collected
were determined by the amount of water available per well volume and the
amount needed for the analysis to be conducted. Measurements of pH, specific
conductance, and alkalinity were made on unfiltered aliquots of each of the
well volumes or partial well volumes that were sampled. These measurements
were conducted immediately after the aliquots were collected to minimize
opportunities for changes in chemical composition due to exposure with the
atmosphere.
The sample aliquots were subdivided into three portions bv filtering
through either a 3.0 um, 0.45 ym, or 0.22 pm pore size Millipore* membrane.
Satorius^ plastic filter holders fitted for 47 mm diameter filters were used
with compressed nitrogen gas over pressure to speed filtration. Each filtered
subsample was then further subdivided into samples for total organic carbon
(TOC), cation, and anion analysis. All filtering was conducted in the field
immediately after each sample was collected.
Samples for sulfate, chloride, ammonia, and nitrate determinations were
stored on ice while in the field and later at 4°C in the laboratory. Ali-
quots for total organic carbon (TOC) were preserved by addition of a few drops
of concentrated sulfuric acid, and the samples for cation analysis were acidi-
fied with nitric acid to a pH <1.5. Samples were acidified immediately after
filtration. Detailed procedures used for sample preservation are given by the
U.S. EPA (1979).
All samples were stored in linear polyethylene bottles that had been
washed with dilute nitric acid and rinsed with dionized water. The preserva-
tion techniques and use of linear polyethylene bottles have been well docu-
mented in the literature, as stated previously, and no variations of any of
these preservations procedures were attempted.
19

-------
To study the effects of sample storage on sample integrity and to further
study the effects of the pump systems on chemical composition, an additional
sampling scheme was conducted. At sites 5-BRD and 6-DUP, large volume samples
(4 L) were collected from the 0, 1, 2, 4, 6, 8, and 10th well volumes pumped
with the peristaltic pump. A small aliquot of each of the collected samples
was immediately filtered and preserved. After pumping ten well volumes, the
pump was withdrawn from the well and replaced with the air-lift mechanism, and
a second 4-L sample was taken using compressed air. Nitrogen gas was then
substituted for the air and used to collect another 4-L sample. The air-
nitrogen-lift mechanism was then removed from the well and a 4-L sample was
collected using a bailer. On-site pH measurements were made on each of the
unfiltered samples, after which small aliquots of each were filtered and pre-
served. The remaining large volume samples were returned to the laboratory
unfiltered, without preservation. Samples were taken in the laboratory the
same day as field sampling and every succeeding day for a period of 4 days.
For this part of the project only the 0.45 pm pore size membranes were used
for filtration.
20

-------
SECTION 7
CHEMICAL ANALYTICAL METHODS
The Jarrel1-Ash^ Atom Comp 975 inductively coupled argon plasma source
direct reading emission spectrophotometer (ICP) was used for the determination
of A1, As, B, Ca, Cd, Cr, Co, Fe, Mg, Mn, Pb, Se, and Zn. Table 1 lists the
typical detection limits for the ICP used during this study. Na and K were
determined by atomic absorption, while CI , F^, and pH were measured electro-
metrically using specific ion electrodes. Sulfate was determined turbidimet-
rically. Nitrate analysis was accomplished by the cadmium reduction method,
and TOC (total organic carbon) was measured using an Oceanographic Interna-
tional 0524B total carbon system.
Field measurement of pH was conducted with an Orion^ 407A field model
pH meter. Specific conductance (EC) was measured with a YSIK model 33 S-C-T
meter, and alkalinity was determined by titration to the methyl orange end
point.
Table 1. Typical Detection Limits for ICP
(concentrations in mg/L)
A1
As
B
Ca
Cd
Cr
Co
Cu
Fe
Mg
Mn
Pb
Se
Zn
0.09
0.07
0.01
0.004
0.01
0.02
0.01
0.01
0.05
0.01
0.005
0.03
0.05
0.05
21

-------
SECTION 8
FACTORS CONTROLLING GROUNDWATER QUALITY
Prior to studying the effects of pumping mechanisms and well flushing on
groundwater quality, it would be instructive to briefly review the factors
controlling groundwater quality. An understanding of these factors allows for
a clearer definition of the mechanisms induced by sampling that are respon-
sible for changes in chemical composition of groundwater samples.
Each of the six monitoring wells sampled during this project was situated
either into or adjacent to areas being used for solid or liquid waste dis-
posal. Thus, the water quality at the sites was influenced by the types of
waste materials and the disposal methods used at the individual sites. A dis-
cussion of the transport mechanisms and mobility of contaminants in ground
water at these sites is beyond the scope of this report. Such a discussion
would require a detailed knowledge of the hydrogeology of each site, the dis-
posal methodology, materials disposed, and detailed background data on water
quality and atmospheric conditions. With this information, a mass balance
analysis of mineral weathering, dissolution of the wastes, and chemical reac-
tions responsible for the groundwater quality could be attempted. However,
studying the factors influencing the in situ groundwater quality at the dis-
posal sites was not the objective of this study. The objective of the project
was to develop a methodology for the sampling of groundwater from monitoring
wells at disposal sites that would produce samples representative of the water
in the aquifer under study. Therefore, the subjects to be reviewed here are
the factors that would induce changes in chemical composition of groundwater
samples by sampling, not those that affect ground water quality in the aqui-
fer. The discussion will be further limited to those mechanisms and reactions
that affected samples collected at the six sites studied during this project.
Groundwater quality is primarily a function of the mineralogy of the
medium through which the water flows, the residence time of the water within
the medium, and the mixing of waters of different quality. Within an aquifer
there are six major processes that may alter the chemical character of the
groundwater. These are:
1)	complex formation
2)	acid-base reactions
3)	oxidation-reduction processes
4)	precipitation-dissolution reactions
5)	adsorption-desorption reactions
6)	microbial processes
22

-------
All of these processes can be affected by the methods used to obtain a
water sample from an aquifer, which may result in production of a sample that
may not be representative of the water from the aquifer. Changes may occur
when the groundwater is removed from the aquifer and exposed to atmospheric
conditions where the temperature, pressure, C>2> and CO2 content may be
different from that in the aquifer. The degree of change in water quality
will often depend upon the magnitude of difference between the groundwater and
surface environments.
The basis for many acid-base reactions in groundwater is the availability
of carbonate sources. To a large extent, the presence of carbon dioxide in
water is the result of the decomposition of organic matter by bacteria. If
rain water percolates down through surface organic matter, it may dissolve a
significant quantity of CO£• Carbon dioxide exhibits limited solubility.
For the reaction
water
COo ^ COo
(9)	(aq)
the equilibrium constant at 25°C is:
[C0?/ 1
k =	(aq) = 3.4 x 10"2 mo1es
PCo2	L x atm
The carbonate system in water can be described by the following set of
reactions:
Dissolution of CO2 in H2O:
H20 + C02 * h2co3, K = 3.4 x KT"2
First Acid Dissociation:
H2C03 * H+ + HCO3, Ki = 4.43 x 10~7
Second Acid Dissociation:
HCO3 * H+ +	k2 = 4-66 x 10-11
Given the equilibrium constants (K, K^, and <2)> a distribution of
species diagram for the carbon dioxide, bicarbonate ion, and carbonate ion
system can be prepared with pH as the controlling variable. According to
Figure 3, for most groundwaters the predominant species is HCO3.
The concentration of HCO3 may be enhanced in some groundwater systems
by the dissolution of calcite by dissolved CO2 in groundwater:
CaCO- A rn u n _ r,2+
U, + COo + HoO = CatT + HCOr
j(s)	c c	0
23

-------
CO
CO
ID
a.
0.8
0.4
0.2
4
5
6
7
8
9
12 13
10
pH
Figure 3. Distribution of species for the COjj-HCOJ-CO^
system in water (Manahan, 1972)
An unstable condition is created when a groundwater sample, equilibrated
with carbonate minerals, is exposed to the surface environment. The sample
will de-gas or lose a quantity of CO? to the atmosphere. Since the partial
pressure of atmospheric CO2 is generally lower than that of groundwater, the
sample will begin to equilibrate to atmospheric conditions. A study by
Wallich (1977) revealed a similar condition where sample pH levels were ele-
vated by the loss of CO2 according to the following equations:
Ca+2 + 2 HCO3 - CaC03 i + H20 + C02 t
H+ + HCO3 - H20 + C02 t
H+ + CO32 ^ HCO3
Thus, the loss of C02 from the system decreases the hydrogen ion activ-
ity and increases pH. In Wallich's (1977) study, pH increases were as great
as 0.5 pH units. Interestingly, the loss of carbonate from the system was
further suggested by the precipitation of CaC03.
Oxidation-reduction reactions are also a major mechanism responsible for
the collection of nonrepresentative aquifer water. Jackson and Inch (1980)
reported a sequence of redox processes (Table 2) that would occur, according
to thermodynamic calculations, in reduced groundwaters exposed to dissolved
oxygen. This sequence represents such groundwater phenomena as the purifica-
tion of leachate by oxygenated groundwaters, and the oxidation of spring
waters. It also can represent oxidation reactions occurring in a groundwater
sample upon exposure to atmospheric conditions. According to Stumm and Morgan
(1970), the following redox processes should take place under these condi-
tions:
1)	oxidation of organics
2)	oxidation of sulfide to sulfate
3)	oxidation of ferrous iron and precipitation of Fe(0H)3
4)	oxidation of ammonium ion to nitrate
5)	oxidation of Mn and precipitation of Mn02 or similar hydrous oxide
24

-------
Table 2. Redox Processes in an Open System (Jackson and Inch,
1980, modified after Stumm and Morgan, 1970)
Reaction	Equation
(1)	Aerobic respiration Op , + CHpO = COp + H?0
(g)	(g)
(2)	Sulfide oxidation	Op + 1/2 HS_ = SO*-+ 1/2 H+
Mg)
(3)	Fe(II) oxidation	Op + 4 Fe^+ + 10 HpO = 4 Fe(0H)o + 8 H+
*(g)
(4)	Nitrification	02{ + 1/2 NHj = 1/2 NOj + H+ + 1/2 H20
(5)	Mn(II) oxidation	Op, + 2 Mn^+ + 2 HpO = 2 MnOo, + 4 H+
£(g)	(s)
The occurrence and magnitude of these processes depend on the groundwater
quality. Complex-ion formation may bind the Fe and Mn in the groundwaters in
a nonoxidizable form. The exposure of the reduced water sample to surface
atmospheric conditions would then have little if any effect on the Fe and Mn
concentrations in solution.
The kinetics of Mn oxidation are considerably slower than those for Fe
oxidation. It is possible to collect a water sample that is representative of
the aquifer water with respect to one constituent (Mn) and not another (Fe),
depending on how rapidly the sample is preserved after collection. Thus, the
extent of error induced by redox processes when measuring aquifer water qual-
ity is constituent specific.
Precipitation of hydrous Fe and Mn oxides presents another problem to the
sampler. It is well documented (Jenne, 1968) that hydrous Fe and Mn oxide
precipitates and coatings on sediment are a control on the sorption from solu-
tion of many heavy metals, including Co, Ni, Cu, and Zn. Thus, precipitation
of Fe and Mn may result in loss from solution of other constituents by adsorp-
tion or co-precipitation with hydrous oxides that were not directly subject to
oxidation themselves.
There are many changes that can take place in a groundwater sample be-
tween collection and preservation. The above discussion has treated the prin-
cipal mechanisms responsible for change in chemical composition as they are
thought to have affected water samples collected during this project. Samples
collected from monitoring wells at other sites may respond differently to the
same processes depending on the chemical composition of the groundwater at
each site.
25

-------
SECTION 9
RESULTS
PUMP TEST ANALYSIS
Traditional analyses of pump test data usually use the equations derived
by Theis (1935) and Jacob (1950). One of the basic assumptions made in deriv-
ing those equations is that all of the water pumped from a well during the
pumping test comes from the aquifer and that none comes from storage within
the well. Since this condition is not always fulfilled in practice, particu-
larly for low yielding wells commonly used for monitoring wells, Theis1 and
Jacob's equations are somewhat inappropriate for describing the behavior of
water levels during pumping for most monitoring wells.
Papadopulos and Cooper (1967) presented an equation describing the dis-
charge from a pumped well, which takes into account the volume of water re-
moved from casing storage.
The drawdown (s) inside the well is expressed as
s = (6.875 Q/T) F(u,a)	(1)
where
s = drawdown in the well after a given pumping time in meters
Q = pumping rate in L/sec
T = transmissivity in nr/day
F(u,a) = well function as defined by Papadopulos and Cooper (1967)
Table 3 presents values of the function F(u,a) as defined by Papadopulos
and Cooper. The values of u for each time increment can be calculated using
equation (2).
u = 360p2s	(2)
Tt
where
r = radius of the well in meters
S = coefficient of storage
T = transmissivity in nr/day
t = time in minutes
The value of a is synonymous with the coefficient of storage (S).
26

-------
Table 3. Values of the Function F(u,a)
10
1
5x10"1
2
1
5xl0"2
2
1
5xl0~3
2
1
5xl0"4
2
1
5xl0"5
2
1
5xl0~6
2
1
5xl0"7
2
1
5xl0~8
2
1
_ Q
5x10 3
2
1
a=10_1
9.755xl0"3
9.192xl0'2
1.767xl0_1
4.062
7.336
1.260x10°
2.303
3.276
4.255
5.420
6.212
6.960
7.866
8.572
9.318
1.024X101
1.093
1.163
1.255
1.324
1.393
1.485
1.554
1.623
1.705
1.784
1.854
1.945
2.015
_2
cx=10 c
9.976xl0"4
9.914xl0"3
1.974xl0"2
4.800
9.665
1.896xl0_1
4.529
8.520
1.540x10°
3.043
4.545
6.031
7.557
8.443
9.229
1.020X101
1.087
1.162
1.254
1.324
1.393
1.485
1.554
1.623
1.705
1.784
1.854
1.945
2.015
a=10~3
9.998xl0"5
9.991xl0"4
1.997xl0"3
4.989
9.966
1.989xl0"2
4.949
9.834
1.945xl0-1
4.725
9.069
1.688x10°
3.523
5.526
7.631
9.676
1.068X101
1.150
1.249
1.321
1.392
1.484
1.554
1.623
1.705
1.784
1.854
11945
2.015
ct=10"4
l.OOOxlO"5
l.OOOxlO"4
2.000
4.999
9.997
1.999xl0"3
4.995
9.984
1.994xl0"2
4.972
9.901
1.965X10"1
4.814
9.349
1.768x10°
3.828
6.245
8.991
1.174X101
1.291
1.378
1.479
1.551
1.662
1.714
1.784
1.854
1.945
2.015
a=10~5
l.OOOxlO"6
l.OOOxlO"5
2.000
5.000
l.OOOxlO"4
2.000
5.000
l.OOOxlO"3
2.000
4.998
9.992
1.997xlO'2
4.982
9.932
1.975X10"1
4.861
9.493
1.817x10°
4.033
6.779
1.013X101
1.371
1.513
1.605
1.708
1.781
1.851
1.940
2.015
27

-------
Drawdown values calculated from equation (1) differ significantly from
those based on Theis1 and Jacob's equations during the early portion of the
pumping test when a relatively high percentage of the discharge comes from
casing storage. During the later stages of the pumping test, when only a neg-
ligible quantity of water is obtained from casing storage, the equations pro-
duce equivalent results. If the effects of casing storage are not taken into
account, it is possible with many monitoring wells to misinterpret the data
and assume an erroneous T (transmissivity) value based on early drawdown
data.
In order to avoid misinterpretation of the data, it is necessary to have
some method for determining when the effect of casing storage becomes negli-
gible. To accomplish this, Papadopulos and Cooper developed equation (3) for
calculating the time at which the effects of casing storage are no longer sig-
nificant.
1440(r ^ - rn^)
tc =—^^	(3)
where
tc = time in minutes after which the effects of casing storage can be
ignored (assuming a 1 percent error in drawdown values)
rc = radius of well casing (inside dimension) in meters
rp = radius of pump column or discharge pipe (outside dimension) in
meters
T = transmissivity in nr/day
The above equation requires prior knowledge of a transmissivity value and
assumes a 100 percent efficient well. Schafer (1978) suggests using the fol-
lowing equation to estimate the tc:
= 1440^ - 0
c	Q/s
tc = time in minutes when casing storage effects become negligible
dc = inside diameter of well casing in meters
dp = outside diameter of pump column in meters
Q/s = specific capacity of the well in nr/day of drawdown at tc
Pump test data for sites 1-SDV, 2-ELG, 3-FLR, 4-TYL, and 6-DUP were ana-
lyzed using equations (1), (2), and (4) described above. At all of the sites
analyzed, the nonpumping water levels were significantly above the tops of the
aquifers tapped, suggesting artesian conditions. A storage coefficient (S) of
0.0001 was chosen and used in all analyses. The drawdown values as described
by the Papadopulos and Cooper equation are relatively insensitive to changes
in storage coefficient. The storage coefficient value selected, therefore,
should have little effect on the aquifer properties determined for each site.
where
28

-------
Site 1-SDV
A pumping test was conducted on the site 1-SDV well on January 11, 1979.
The well at site 1-SDV was pumped with a Jabsco self-priming pump at rates
varying from 1759 mL/min to 1000 mL/min for a period of 15 minutes before the
pump broke suction. Water level recovery data were collected for a period of
30 minutes after pumping stopped. Since the length of time of pumping was not
long, the recovery data were used to determine an aquifer transmissivity of
about 1.24 m2/day (100 gpd/ft). Figure 4 illustrates the general agreement
between adjusted pump test results and the theoretical drawdown curve.
The Papadopulos and Cooper method of analyses can be used to develop
theoretical drawdown curves, but more importantly for collection of monitoring
well samples, the amount of water coming from the aquifer can be calculated
for each time increment. These amounts as percentages of total pumpage,
(Qa/Qt) x 100> are presented in Figure 5 for a pumping rate of 500 mL/min
for all six sites. Based on these calculations it appears that a pumping rate
of 500 mL/min for the site 1-SDV well should produce water samples containing
approximately 90 percent aquifer water in about 10 minutes. This conclusion
is based on the assumption that the pump intake is located in the screened
portion of the well near the bottom so that mixing of aquifer and casing water
will be minimal. The net effect is to "isolate" the stagnant water in the
well casing.
Site 2-ELG
A pumping test was conducted on the site 2-ELG well on January 17, 1979.
The well at site 2-ELG was pumped with the peristaltic pump at rates varying
from 725 to 775 mL/min for a period of 2 hours. Recovery measurements were
taken for 30 minutes after pumping stopped.
With the analysis method d^eloped by Papadopulos and Cooper, an aquifer
transmissivity of about 0.56 nr/day (45 gpd/ft) was calculated. Figure 6
illustrates the general agreement between the early pump test results and the
theoretical drawdown curve.
Calculations of the percent of aquifer water pumped at 500 mL/min indi-
cate that this well also will deliver approximately 90 percent aquifer water
in about 10 minutes (Figure 5). Because of the small diameter of the well and
the low yield potential of the aquifer at the start of pumping, a smaller per-
centage of aquifer water is obtained than for the site 1-SDV well.
Site 3-FLR
A pump test was conducted on the site 3-FLR well on August 9, 1979. The
well at site 3-FLR was pumped with the peristaltic pump at rates of 425 to
400 mL/min for a period of 90 minutes. Water level recovery measurements were
made for a period of 30 minutes after pumping stopped.
The method of analysis developed by Papadopulos and Cooper was used and
an aquifer transmissivity of about 1.55 nr/day (125 gpd/ft) was calculated.
29

-------
7
SITE 1 — SDV
Qt = 1000 ml/min
THEORETICAL
CURVE
6
RAW DATA
5
4
3
2
1
0
0
5
TIME, minutes
Figure 4. Site 1-SDV: time-drawdown data and theoretical curve
100
Site 5
ir
u_
Q = 500 ml/min
TIME, minutes
Figure 5. Aquifer yield curves for all six sites
30

-------
15
'10
z
5
o
o
3
<
CE
a
T
SITE 2 - ELG
Q = 750 mL/min
THEORETICAL CURVE
RAW DATA
_L
_L
_L
10 15 20
TIME, minutes
25
1.4
1.2
0.2
30
T
SITE 3 — FLR
Q = 400 mL/min
JL
_L
_L
10 15 20
TIME, minutes
25
30
Figure 6. Site 2-ELG: time-drawdown
data and theoretical curve
Figure 7. Site 3-FLR: time-drawdown
data and theoretical curve
i r
SITE 4 - TVL
Q = 800 mL/min
SITE 6 - DUP
Q = 700 mL/min
10 15 20 25 30
TIME, minutes
10 15 20 25 30
TIME, minutes
Figure 8. Site 4-TYL: time-drawdown Figure 9. Site 6-DUP: time-drawdown
data and theoretical curve	data and theoretical curve
31

-------
Figure 7 illustrates the general agreement between the early pump test results
and the theoretical drawdown curve.
The aquifer yield curves presented in Figure 5 suggest that when pumped
at 500 mL/min this well will produce about 90 percent aquifer water in about
5 minutes.
Site 4-TYL
A pumping test was conducted on the site 4-TYL well on November 18, 1978.
The monitoring well at site 4-TYL was pumped with the peristaltic pump at
rates from 715 to 600 mL/min for a period of 20 minutes, and water level
recovery measurements were made for a period of 30 minutes after pumping
stopped.
On the basis of the results of the recovery data, an aquifer transmissiv-
ity of about 0.05 rrr/day (4 gpd/ft) was determined. Figure 8 illustrates
the general agreement between the early pump test results and the theoretical
drawdown curve.
Figure 5 illustrates the percent aquifer water pumped with time at 500
mL/min. Because of the extremely low transmissivity value at this site, it is
unlikely that a significant percentage of aquifer water could be obtained at
even lower pumping rates. At pumping rates of 50 and 100 mL/min, calculations
show that only 42 percent aquifer water is pumped after 30 minutes. For this
reason it may be desirable and more practical to simply pump the well dry,
allow it to recover, and then collect the sample. Results of sampling using
this procedure are presented later in this report for wells at sites 2-ELG and
4-TYL.
Site 5-BRD
A pump test was attempted on the site 5-BRD well on August 22, 1979. The
well at site 5-BRD was pumped with the peristaltic pump at rates from 750 to
650 mL/min for a period of 2 hours. The total water level drawdown for the
period of pumping was only 5.18 cm (0.17 ft). The data obtained from this
test proved to be of little value in determining aquifer properties except to
point out that the transmissivities were very high relative to the other
sites.
Because of the small diameter of the well and the high water-yielding
character of the aquifer, a pump of adequate capacity that would fit into the
well and still permit measurement of water levels could not be found. Region-
al data on larger production wells tapping the same aquifer in the same area
indicate a range of aquifer transmissivities from about 3,700 to 9,900
m^/day (300,000 to 800,000 gpd/ft).
Based on the inability to pump the well at rates sufficient to create
significantly measurable drawdown, it can be assumed that essentially all
water pumped during the sampling runs to be discussed later was "aquifer
water" and not water from storage within the well casing (Figure 5).
32

-------
Site 6-DUP
A pumping test was conducted on the site 6-DUP well on July 24, 1979.
The well at site 6-DUP was also pumped with a peristaltic pump at rates from
765 to 620 mL/min for a period of 2 hours. Recovery measurements were taken
for 10 minutes after pumping stopped.
An aquifer transmissivity of 2.48 m^/day (200 gpd/ft) was determined.
Figure 9 shows the general agreement between the early pump test results and
the calculated theoretical drawdown curve. According to Figure 5, this well
should yield about 92 percent aquifer water in about 5 minutes when pumped at
500 mL/min.
On the basis of the results of the pumping tests at five of the sites,
three basic principles have been substantiated: (1) constant rate pumping
tests can be performed on monitoring wells and the results analyzed to deter-
mine aquifer properties; (2) the proper applications of the equations devel-
oped by Papadopulos and Cooper allow for straightforward determination of
realistic transmissivity values for the water strata being sampled by the
monitoring well; and (3) once the transmissi vity of the aquifer tapped by a
monitoring well has been determined, the percent of water coming from the
aquifer and that from storage can be determined as a function of pumping time.
Figure 10 illustrates the percent of water pumped that is coming from the
aquifer at a pumping rate of 500 mL/min. As expected from the results of the
pump tests presented in Figure 5, the higher percentages of aquifer water are
derived from the higher yielding wells (larger T values). Once a T value of
62.0 nr/day (5000 gpd/ft) is encountered, larger values of T make little or
no difference at a pumping rate of 500 mL/min. Thus, if the yield capability
of the well is sufficiently large (T>5000 gpd/ft), then the effects of water
from storage become negligible for 5.08 cm (2 in.) diameter wells and can be
ignored when establishing a sampling protocol at this pumping rate.
Additional calculations illustrate the effect of the diameter of the well
on the percent of aquifer water removed at a constant pumping rate and trans-
missi vity (see Figure 11). For a T of 2.48 m^/day (200 gpd/ft) and a pump-
ing rate of 500 mL/min, well diameters larger than 5.08 cm (2 in.) have a sig-
nificant effect on the percent of aquifer water pumped, particularly during
the very early stages of pumping. Results from analyses of pump test data
provide the basic hydrologic information needed to permit a better understand-
ing of the changes in chemical quality of water that are likely to occur as a
function of pumping time as the sample is pumped from a monitoring well.
During the course of conducting pumping tests on monitoring wells, vari-
ous problems can be encounterd. The following are suggestions to avoid those
potential problems:
1) During the course of a pumping test the water levels drop and the
yield capability of the peristaltic pump also declines. To compen-
sate for declining pumping rates (or to maintain a constant pumping
rate) it may be necessary to use a variable speed peristaltic pump
and a slower overall pumping rate. As the water levels drop, the
speed of the pump can be increased to maintain a relatively constant
33

-------
100
uj 80
I-
<
3
tr
uj 60
u.
O
<
H 40
z
UJ
o
cc
£ 20
0
0 5 10 15 20 25 30
TIME, minutes
Figure 10. Percent of aquifer water versus time for
different transmissivities
100
UJ 80
I—
<
5
cc
uj 60
u.
O
o
<
H 40
2
UJ
O
CC.
uj
o. 20
0
0 5 10 15 20 25 30
TIME, minutes
Q = 500 mL/min
DIAMETER = 5.08 cm
CASING
DIAMETERS
Figure 11. Percent of aquifer water versus time for
different well casing diameters
34

-------
pumping rate. The selection of a smaller pumping rate will probably
be dictated by the anticipated total final drawdown or water level
and the pump yield capacity at that particular pumping lift.
2)	For low yielding wells it may be very difficult to maintain a con-
stant pumping rate for a period of time adequate to provide data
suitable for analyses. Pumping rates less than 100 mL/min for peri-
ods up to 2 or 3 hours may be necessary.
3)	For very high yielding wells the peristaltic pump may not be capable
of pumping at rates sufficient to induce enough drawdown to permit
meaningful analyses. The use of the diaphragm type Middleburg pump
with higher pumping rates may be helpful.
If pumping tests on monitoring wells are conducted correctly, it is felt
that they provide more useful and valid data than those obtained from slug
tests.
EFFECTS OF PUMPING MECHANISMS ON CHEMICAL COMPOSITION
As noted earlier, each of the six study sites was sampled using a peri-
staltic, air-lift, nitrogen-lift, and bailing mechanism. The sites were
visited at one-month intervals to simulate monthly sampling as required by
many regulatory agencies. For the purpose of discussing the effects of pump-
ing mechanisms on sample chemistry, only data for samples filtered through
0.45 ym pore size membranes are presented.
Probably the single most important parameter affecting the chemical com-
position of groundwater is pH. Therefore, anything that alters the pH of the
groundwater samples is likely to alter the groundwater composition. On-site
pH measurements at all six sites using the four pumping mechanisms were ana-
lyzed. The data illustrate two possible effects of the type of pumping mecha-
nism on chemical composition of samples. For comparative purposes, the aver-
ages of the pH values for the sixth, eighth, and tenth well volumes collected
at each site by each of the four pump mechanisms are listed in Table 4. The
pH values for the samples collected with the peristaltic pump and the bailer
are lower by as much as 1.1 pH unit than the pH values for those samples taken
with the air- and nitrogen-lift systems.
The size of the discharge pipe used with the air- and nitrogen-lift
systems when collecting samples also affects their pH. Samples collected with
either the air- or nitrogen-lift systems using a 0.952 cm (3/8 in.) diameter
discharge pipe possess lower pH's than samples taken by either pump system
when using a 1.27 cm (1/2 in.) diameter discharge pipe.
The increased pH values of samples collected with these pump systems are
a result of bubbles of air or nitrogen rising through the water in the dis-
charge pipe and stripping dissolved CO2 from the water. The smaller changes
in pH noted in the samples collected using the 0.952 cm (3/8 in.) discharge
pipe occurred because less air or nitrogen was needed to pump the same quan-
tity of water than with the 1.27 cm (1/2 in.) discharge pipe.
35

-------
Table 4. Average pH Values for the Sixth, Eighth,
and Tenth Well Volumes Collected at
Each Site by Each Pump Mechanism
Site	Pump System

peristaltic
air
N2
bailer
1-SDV
4.5
5.4
5.3
5.2
2-ELG
7.0
8.02
8.12
ND3
3-FLR
7.1
8.22
7.61
•
CO
4-TYL
7.2
8.22
7.81
7.0
5-BRD
6.8
7.61
7.51
6.8
6-DUP
7.1
7.81
7.61
7.0
1	0.952 cm (3/8") id inlet line
2	1.27 cm (1/2") id inlet line
3	Not Determined
The gas-to-water ratios for the samples collected using the 0.952 cm
(3/8 in.) diameter pipe varied from about 3.4 to 9.5, with the higher ratios
generally resulting in larger pH changes. The gas-to-water ratios for the
samples collected using the 1.27 cm (1/2 in.) diameter pipe were much higher
(30 to 40) and appear to have effectively stripped the samples of most of
their excess dissolved CO2, resulting in pH values of 8.2 to 8.3. This can
be compared with pH values of 6.9 to 7.0 when the peristaltic pump was used to
collect samples.
Data from samples collected at site 5-BRD using an air-lift pumping mech-
anism dramatically illustrate the effect of discharge pipe size on pH. During
the first stage of pumping (volumes 0 through 2), a 1.27 cm (1/2 in.) dis-
charge pipe was used. To increase the pumping rate, the 1.27 cm (1/2 in.)
pipe was withdrawn and a 0.952 cm (3/8 in.) pipe inserted between well volumes
2 and 4. The pH immediately dropped from 8.2 to 7.5, confirming the important
effect of the gas-to-water ratios on the pH of water samples collected with
gas-lift devices.
With the exception of site 1-SDV, the pH of samples collected with the
peristaltic pump and bailer were similar. During the period between sampling
at site 1-SDV with the nitrogen-lift system and with the bailer, the scrubber
waste disposal pond had drained completely as it was no longer being used for
disposal. Therefore, the hydrogeology of the site and the source of low pH
water had been altered, making interpretation of the bailer data difficult.
36

-------
It was not possible to make direct interpretation of the bailer data for
site 1-SDV shown in table 4 to that collected by the other mechanisms.
Of the 20 chemical constituents and parameters measured, only pH, Fe, and
Zn were found to be affected by the type of pump mechanism used to collect the
samples. There were no obvious effects of any of the four pump mechanisms on
the concentrations of alkalinity, Ca, Cl_, F^, Mg, Mn, Na, specific conduct-
ance, or TOC in the samples. Only in the site 1-SDV samples were there suffi-
cient concentrations of A1, B, Cd, Cr, Ca, and Pb to assess the effects of the
pump mechanisms. The overall variability of the water composition at site 1-
SDV and the high concentrations of Zn (= 25,000 mg/L) and Ca (= 2,000 mg/L)
resulted in both chemical and physical interferences to the analytical proce-
dures being used. Thus an evaluation of the effects of the pump mechanisms on
A1, B, Cd, Cr, Cu, and Pb was not possible. It is suspected, however, that
some if not all of these constituents will have their concentrations affected
by the pump mechanism used for sample collection and their responses will be
similar to that displayed by Fe and Zn.
For all of the sites where detectable Fe concentrations were found
(sites 2-ELG, 5-BRD, and 6-DUP), the highest concentrations were in the
samples collected with a peristaltic pump or bailer, followed by the samples
taken with a nitrogen-1ift. Only trace dissolved Fe concentrations were de-
tected in any of the samples collected with the air-lift mechanism.
Figure 12 is a plot of the Fe concentrations in samples from site 5-BRD
and clearly illustrates the effect of the pumping mechanisms on Fe content.
Soluble Fe concentrations in the initial (zero) well volume taken by the peri-
staltic pump, bailer, and nitrogen-lift were 14.6, 17.4, and 22.8 mg/L, re-
spectively. The dissolved Fe concentration in the zero well volume taken with
the air-lift was 0.66 mg/L. The indication is that the use of an air-lift
pumping mechanism results in partial removal of the available Fe from solu-
tion, probably by oxidation and/or a rise in pH with subsequent precipitation.
The soluble Fe data plotted for the samples taken with the other mechanisms
show a decline in concentration until the fourth well volume had been removed.
The samples collected with the peristaltic pump and bailer maintained rela-
tively similar Fe concentrations throughout the series. Although the samples
taken with the nitrogen-lift exhibit an Fe-concentration well-volume curve
similar to that of the bailed and mechanically pumped samples, the values are
considerably lower. This lower concentration of Fe was felt to represent
partial oxidation of the Fe by oxygen contamination in the compressed nitrogen
gas, and/or precipitation due to the increase in pH brought about by the
stripping of COg by the nitrogen gas.
Figures 13 and 14 are plots of dissolved Fe versus volumes pumped for
sites 2-ELG and 6-DUP. Again, the bailed and peristaltic pumped samples for
both sites follow very similar trends. For site 2-ELG (Figure 13) the soluble
Fe concentrations remain relatively constant, and for site 6-DUP (Figure 14)
the values decrease during pumping of the first three to four well volumes,
after which the concentrations stabilize. The differences in the dissolved Fe
concentrations between the samples collected with the peristaltic pump and the
bailer at each site may represent seasonal variations in the groundwater qual-
ity. The samples collected using nitrogen- and air-lift pumping mechanisms
37

-------
24
SITE 5 - BRD
PERISTALTIC
AIR-LIFT
NITROGEN-LIFT
BAILER
20
16
12
8
4
0
0
2
4
6
8
10
WELL VOLUMES PUMPED
Figure 12. Effects of pumping mechanism on iron concentrations at
site 5-BRD as a function of well volumes pumped
38

-------
7
SITE 2-ELG
PERISTALTIC
AIR-LIFT
NITROGEN-LIFT
BAILER
6
5
4
3
2
1
0
0
8
4
6
10
WELL VOLUMES PUMPED
Figure 13. Effects of pumping mechanism on iron concentrations at
site 2-ELG as a function of well volumes pumped
39

-------
8
PERISTALTIC
AIR-LIFT
NITROGEN-LIFT
BAILER
x SITE 6 - DUP
7
6
5
4
3
2
1
0
8
10
6
2
4
0
WELL VOLUMES PUMPED
Figure 14. Effects of pumping mechanism on iron concentrations at
site 6-DUP as a function of well volumes pumped
40

-------
for site 2-ELG show only trace levels of dissolved Fe present and are con-
sidered to be a response of the samples to oxidation and changes in pH.
At site 6-DUP the soluble Fe concentrations in the samples taken with an
air-lift mechanism remain almost constant through the ten well volumes
(approximately 0.60 mg/L). The initial Fe concentrations in the samples col-
lected with the mechanical and nitrogen-lift mechanisms were 4.0 and 1.7 mg/L,
respectively. However, after pumping four well volumes using the three mecha-
nisms, the resultant dissolved Fe concentrations in the samples were nearly
identical. This would indicate that the groundwater in the aquifer (volumes
four through ten) possessed a lower level (approximately 0.60 mg/L) of Fe than
the storage water in the well casing and that the Fe is in a form not readily
affected by the introduction of air or nitrogen into the system.
The sampling routine, using all four pump mechanisms on one date, was
conducted to verify that the changes in pH, Fe, and Zn concentrations were due
to the pump used and not to seasonal variations over the initial four-month
sampling period. Table 5 lists the results of the sample collections at sites
5-BRD and 6-DUP, where all four pumping mechanisms were used on the same day
after ten well volumes had been pumped with the peristaltic pump. The order
of pumping mechanisms listed in the tables was the order in which they were
used. Again, the pH values of the samples taken with the peristaltic pump and
bailer are as much as 1.0 pH unit less than those samples taken with the air-
and nitrogen-lifts. Table 5 also shows that the Fe concentrations for both
sites and the Zn concentrations for site 5-BRD were affected by the choice of
pumping system used to collect the samples. The choice of pump systems did
not affect the concentrations of Ca, K, Mg, and Na in any consistent way. The
loss of Fe from solution is most probably due to the precipitation of hydrous
Fe oxides. Adsorption of Zn onto the Fe precipitates is presumed to be re-
sponsible for the loss of Zn. At both sites, the dissolved Fe concentrations
found in the bailed samples were higher than in the samples taken with the
peristaltic pump. This was believed to be a result of precipitated iron from
the air- and nitrogen-lift pumping sequences being taken back into solution by
the lower pH water entering the well from the aquifer and subsequently being
collected with the bailer.
On the basis of the results of this portion of the study, the peristaltic
pump and bailer appear to be the best types of sampling mechanisms to minimize
chemical changes in water samples. This recommendation should be considered
particularly when sampling for pH sensitive or volatile chemical parameters.
The air- and nitrogen-1ift systems are not recommended except in cases where
pH insensitive and non-volatile species are to be measured in the samples.
For wells with water levels within suction lift capabilities, a peristal-
tic pump can be used. For wells with water levels below suction lift capa-
bilities (about 5.2 m or 17 ft), the submersible diaphragm type of pump is
recommended when the air- and nitrogen-1 ift systems are precluded due to the
presence of sensitive chemical constituents. If the use of one of these types
of pumps is not practical, a bailer is the next best choice. However, when
collecting a water sample with a bailer, certain procedures need to be fol-
lowed. The following is a partial listing of procedures which will aid in
collecting representative samples with a bailer:
41

-------
Table 5. Analysis of Samples from Sites 5-BRD and 6-DUP Collected by
Four Pumping Mechanisms Immediately after Flushing Ten Well Volumes
(Concentrations in mg/L)
Site 5-BRD
Sequence
Pumping Mechanism
pH
Ca
Fe
K
Mq
Na
Zn
1
Peristaltic
6.7
111
11.6
ND*
44.6
186
0.18
2
Nitrogen-1ift
7.7
106
3.7
ND*
44.0
183
0.03
3
Ai r-1ift
7.5
105
0.7
ND*
43.6
182
0.03
4
Bailer
6.8
105
13.5
ND*
42.2
179
0.96
Site 6-DUP








Sequence
Pumping Mechanism







1
Peri staltic
6.8
32.5
5.74
189
94.2
215
ND*
2
Nitrogen-lift
7.8
36.1
1.79
202
111
239
ND*
3
Ai r-1ift
7.8
37.1
.57
199
115
247
ND*
4
Bailer
6.8
42.3
7.42
193
114
235
ND*
*ND - not detectable
42

-------
1)	The bailer should be constructed of a noncontaminatirig material.
2)	A pass-through type flapper valve should be used to minimize disturb-
ance as the bailer is lowered through the water column.
3)	The bailer should be lowered to the same depth (the top of the well
screen) every time to create the same effect as pumping with a peri-
staltic pump.
4)	Bailing should be timed to approach a constant pumping rate and
should continue until the appropriate well volumes are removed prior
to collecting the sample.
5)	The rope used to operate the bailer should be of a noncontaminating
material and should be held off the ground during the bailing pro-
cess.
6)	The bailer and rope should be thoroughly cleaned before use in each
wel 1.
EFFECTS OF WELL FLUSHING ON CHEMICAL COMPOSITION
One recommended procedure for the collection of representative ground-
water samples from monitoring wells is to flush the monitoring well to remove
the stagnant water held in storage in the well casing, or to pump until a high
percentage of aquifer water is being received. Storage water is defined as
the water that does not come into contact with the flowing groundwater. Be-
cause of the existence of site-specific variables of geology, hydrology,
transmissivity values, and chemistry between waste disposal sites, the extent
of well flushing required will be different for each well. As demonstrated in
Figures 5 and 10, a specific volume of water pumped from one monitoring well
before sample collection may be sufficient to produce a representative sample
of the aquifer, but at another site the same volume may be insufficient or
perhaps may result in overpumping. Over-pumping may introduce groundwater
from a distant source that could dilute or concentrate certain constituents
and result in erratic or misleading data.
To assess the phenomenon of well flushing, the concentrations of the
major constituents in the groundwater at each of the six sites were plotted
versus the well volumes pumped by the peristaltic pump (Figures 15 through
20). Due to the differences in chemistry and hydrology between the sites, it
was instructive to plot the data for each site individually with the exception
of sites 3-FLR and 4-TYL, which were chemically similar. Unless otherwise
noted, all the data discussed were from samples prepared by filtration through
a 0.45 urn pore size membrane.
Determinations of SO^^ ci—, ancj p— were conducted. However, only
the CI- data for a few of the sites will be discussed. Overall, these con-
stituents were not present in the samples in concentrations exceeding the de-
tection limits of the analytical procedures used.
43

-------
700
600
500
400
300
200
100
0
>.
Magnesium
Cadmium
Manganese
4	6	8	10
WELL VOLUMES PUMPED
ite l-SDV: magnesium, cadmium, and manganese concentrations
versus volumes pumped (peristaltic pump)
44

-------
4
Selenium
3
Arsenic
2
Boron
1
Copper
0
0
2
4
8
10
WELL VOLUMES PUMPED
Figure 16. Site 1-SDV: selenium, arsenic, boron, and copper
concentrations versus volumes pumped (peristaltic pump)
45

-------
40
Magnesium
CT>
E
O
I-
<
GC
I-
Z
LU
o
Potassium
Sodium
Iron
10
4
6
8
0
2
WELL VOLUMES PUMPED
Figure 17. Site 2-ELG: magnesium, potassium, sodium, and iron concen-
trations versus volumes pumped (peristaltic pump)
46

-------
200
160
120
80
/
40
30
20
10
0
/\^ _ 			 		
Sodium
Calcium
/
	SITE 3- FLR
	SITE 4-TYL
A
\ Magnesium.
\ ^ *
Calcium
Sodium
j Magnesium
4	6
WELL VOLUMES PUMPED
10
8. Sites 3-FLR and 4-TYL: sodium, calcium, magnesiurr
se concentrations versus volumes pumped (peristaltic p
47

-------
32
28
24
20
16
12
8
4
0
Potassium
Magnesium
Iron
Manganese
¦ Zinc
4	6
WELL VOLUMES PUMPED
10
). Site 5-BRD: potassium, magnesium, manganese, zinc, a
:oncentrations versus volumes pumped (peristaltic pump)
48

-------
32
28
24
20
16
12
8
4
0
Iron (February)
Iron (August)
Zinc (February)
Zinc (August)
4	6
WELL VOLUMES PUMPED
10
3. Site 5-BRD: iron and zinc concentrations versus volumes
pumped for two sampling periods (peristaltic pump}
49

-------
Sites 1-SDV and 6-DUP possessed the most unusual hydrogeologic settings
of the six sites. In both cases, the monitoring wells were either located in
or immediately adjacent to a source of leachate. The close proximity of the
monitoring wells to the leachate source presents some additional complications
in the interpretation of results. The difficulties encountered at site 6-DUP
will be discussed later.
Figures 15 and 16 are plots of some of the major constituents found in
the groundwater at site 1-SDV. These figures indicate the changes in these
constituent concentrations with the number of well volumes pumped. The sixth,
eighth, and tenth well volumes represent relatively consistent concentrations
for B, Cu, Cd, and Mn. Boron, Cd, and Mn display similar trends of increasing
concentrations in the first four well volumes. The plot of the Cu values
shows a steady decrease in the first four well volumes until a stable concen-
tration was reached. Variability in constituent concentration during the
first two to four well volumes pumped was presumed to represent mixing of
aquifer and storage water. The stable concentrations obtained during the
sixth, eighth, and tenth well volumes could be interpreted as "representative"
aquifer water. However, the data for As, Se, and Mg, along with Ca and Zn,
which were not plotted, indicate an overall increase in concentration of these
constituents during the pumping of all ten well volumes. It is believed that
this is a case of overpumping due to the placement of the monitorng well next
to the waste settling pond, which permits the waste leachate to be drawn
directly into the well during extensive pumping. This makes it difficult to
determine when a "representative" sample of the aquifer has been obtained for
these latter constituents.
Figures 17 and 18 are plots of the major constituents found in the
groundwater at sites 2-ELG, 3-FLR, and 4-TYL. The data for these wells indi-
cate that the constituent concentrations do not change appreciably during
pumping. This indicates that the water in storage is similar to the water in
the aquifer. The lack of any change in the Fe concentration at site 2-ELG was
of particular interest. Of all the constituents measured, soluble Fe was
perhaps the most likely to exhibit concentration changes between water being
held in storage and that present in the groundwater.
Data collected at site 5-BRD dramatically demonstrate the effects of well
flushing on sample collection. Figure 19 illustrates the effect of pumping on
the concentrations of the major constituents in samples from site 5-BRD. As
was the case for the previous sites, K, Mg, and Mn show only slight changes in
concentration with the number of well volumes pumped. However, Figure 19
shows that it was necessary to remove as many as eight to ten well volumes
before samples with stable concentrations of Fe and Zn were received and a
"representative" sample of the aquifer could be collected.
This is in apparent conflict with information derived from the pump test
analysis portion of this study, which suggests that nearly 100 percent aquifer
water should be expected from the very start of pumping. The decrease of dis-
solved Fe and Zn concentrations with pumping would indicate that the water in
storage was higher in soluble Fe than the aquifer water. It is quite possible
since the well at site 5-BRD was the only one studied where the casing mate-
rial was galvanized steel. The presence of Fe and Zn in the storage water may
50

-------
be a result of corrosion of or leaching from the well casing. The Fe and Zn
may also migrate into the aquifer for a short distance around the well. The
removal of this anomalously high Fe and Zn water from the aquifer would be
necessary before representative aquifer water samples could be obtained.
For another phase of the project an additional series of samples were
collected at sites 5-BRD and 6-DUP using a peristaltic pump. These samples
were collected in early February, approximately six months after the first
series of samples were taken from the sites using a peristaltic pump. Com-
parison of the data from the samples collected in February and those collected
in August illustrates two additional problems encountered in sampling from
shallow monitoring wells.
Figure 20 shows the Fe and Zn values for samples collected from site 5-
BRD. The trends of decreasing concentration with volumes pumped were similar
for both contituents during both sampling intervals. Although the trends were
nearly identical, the concentrations in the samples collected in February were
somewhat higher than those in the samples collected in August. This is proba-
bly due to seasonal fluctuations in groundwater chemistry induced by rainfall
events and temperature changes.
As mentioned previously, the monitoring well at site 6-DUP is similar to
the well at site 1-SDV with respect to its proximity to the disposal site.
The well at site 6-DUP was placed directly beneath the refuse at a municipal
landfill. The landfill also possesses a relatively thin soil cover that
allows for rapid recharge by infiltrating rainwater. These factors increase
the possibility of seasonal variations in groundwater composition at the
site.
Groundwater within the landfill also has an extremely heterogeneous
chemical nature, dependent upon its immediate environment within the landfill
(Hughes et al., 1968). Thus, sampling from a well placed directly into or
beneath the refuse could result in samples exhibiting large fluctuations in
chemical composition.
Figure 21 depicts the change in Fe concentration during the pumping of
ten well volumes when the well was sampled in both August and February using a
peristaltic pump. The decrease in Fe during pumping in August was believed to
reflect an aquifer under oxidizing conditions. The trend of increasing Fe in
the February samples would indicate that the aquifer (or refuse) was under a
reduced environment with more Fe in solution. The presence of a frozen soil
cover during winter, creating a closed system and not permitting the introduc-
tion of air (oxidants), could explain the reducing environment. The samples
collected during August reflect the effects of well flushing (volumes 0
through 4), the oxidizing environment of the summer months (the lower overall
Fe concentrations), and the heterogeneous nature of water contained in the
refuse (volumes 6 through 10). It appears that it would be very difficult to
obtain reproducible "representative" samples of water from this monitoring
well. There are too many factors affecting the chemical characteristics of
water contained in the water-bearing strata.
51

-------
8
February
7
6
5
4
3
2
August
1
0
0
8
10
2
4
6
WELL VOLUMES PUMPED
Figure 21. Site 6-DUP: iron concentrations versus volumes pumped
for two sampling periods (peristaltic pump)
52

-------
As previously stated, SO^, F~, and Cl~~ determinations were made on
samples collected by peristaltic pump at each of the sites. However, only Cl—
was found at levels above the detection limits of the analytical procedures
used. The results of the Cl~* analysis versus well volumes pumped at sites
4-TYL,	5-BRD, and 6-DUP are shown in Figure 22. There were no appreciable
changes in CI- concentration in the samples taken from sites 4-TYL and 5-BRD.
Data from site 6-DUP show some initial fluctuation in the Cl~ values before
stable concentrations were obtained. Overall, the data suggest that Cl~ con-
centrations found in the storage waters were similar to those in the aquifers.
The effect of pumping on Cl_ concentrations is of concern since CI-, along
with pH, temperature, and specific conductance, has often been used to indi-
cate when representative groundwater samples have been obtained. For sites
5-BRD	and 6-DUP the Cl~ values vary only slightly with volumes pumped, as do
many of the other principal constituents. This suggests that CI- would not be
a good indicator of when representative aquifer water has been obtained.
In addition to collecting water samples while pumping to determine the
effects of well flushing, two sites (2-ELG and 4-TYL) were pumped dry and
allowed to recover, and samples were collected. Because of the extremely low
permeabilities of the materials tapped by some monitoring wells, it is not
always practical to pump all monitoring wells for long periods of time.
Tables 6 and 7 present data from samples collected during pumping and immedi-
ately after the site 2-ELG and 4-TYL wells were pumped dry with the peristal-
tic pump.
On the basis of the results of these two brief experiments, it appears
that pumping low yielding monitoring wells dry, allowing them to recover, and
collecting a water sample during recovery would yield representative chemical
results. However, because of the numerous geologic, hydrologic, and chemical
factors that can affect the chemical composition of collected samples, it is
recommended that a brief experiment similar to those at sites 2-ELG and 4-TYL
be conducted prior to accepting this sampling procedure for a particular
site.
The chemical data from this portion of the study have verified the theo-
retical ratios of aquifer to stored water predicted during the pump tests.
One group of elements--Na, K, Mg, Ca, and Cl--has been shown to be relatively
insensitive to time of pumping or well flushing. Another group of elements—
Fe, Mn, Mg, Zn, Cd, Cu, As, Se, and B—is sensitive to the effects of well
flushing. These constituents apparently undergo chemical changes when stored
in the well casings. These constituents, along with pH, could be used as
indicators of when representative samples have been obtained from an aquifer
when using a pump system that will not affect their concentrations. The
selection of the most suitable indicator should be based on the hydrologic and
chemical characteristics of the groundwater at each individual site.
EFFECTS OF SAMPLE PREPARATION, PRESERVATION, AND STORAGE
ON CHEMICAL COMPOSITION
The effort to collect a representative groundwater sample from an aquifer
via a monitoring well would be futile if the chemical composition of the
53

-------
800
SITE 6 - DUP
700
600
500
- 400
300
200
SITE 5-BRD
100
SITE 4 -TYL
0
2
8
4
10
6
WELL VOLUMES PUMPED
Figure 22. Sites 4-TYL, 5-BRD, and 6-DUP: chloride concentrations
versus volumes pumped (peristaltic pump)
54

-------
Table 6. Analysis of Samples from Site 2-ELG Where the Well Was
Placed in Geologic Materials of Low Hydraulic Conductivity
(Concentrations in mg/L)
Well Volume	Ca Fe K	Mg	Mn	Na	Zn
0	13.6
2	61.6
4	60.9
6	60.6
after recharge	61.4
<.01
5.47
4.45
5.40
2.51
35.2
4.86
2.41
34.5
4.53
2.00
34.5
4.66
2.03
34.9
011
4.35
.188
111
14.0
.359
098
13.4
.795
097
13.4
.222
094
13.8
.203
Note: The well was sampled as it was being pumped dry with the peristaltic
pump and then resampled after recharge.
Table 7. Analysis of Samples from Site 4-TYL Where the Well Was
Placed in Geologic Materials of Low Hydraulic Conductivity
(Concentrations in mg/L)
Well Volume	Ca Fe K	Mg	Mn	Na	Zn
0
104
.018
<1.0
40.0
.118
14.9
.037
1
102
.182
<1.0
39.5
.727
17.1
.025
1 1/2
107
.503
<1.0
39.2
2.06
15.1
.019
after recharge
102
.757
<1.0
39.2
2.03
16.2
.003
Note: The well was sampled as it was being pumped dry with the peristaltic
pump and then resampled after recharge.
55

-------
sample changed between the time of collection and analysis. Proper sample
preparation, preservation, and storage can help prevent such changes from
occurring. Adequate procedures have been developed for sample preservation,
and many investigations into the proper vessels for sample storage have also
been completed. This phase of the project was primarily concerned with wheth-
er these prescribed procedures for preparation and preservation of samples
should be applied on-site immediately after sample collection or in the labo-
ratory some time later.
The effects of filter pore size on sample chemical composition were also
studied. Subsamples, filtered through either a 3.0, 0.45, or 0.22 um pore
size filter, were taken from each well volume or partial well volume collected
during the monthly sampling routines. The analyses of these subsamples were
then compared to determine the effect of membrane pore size on constituent
concentration in the filtered samples. Tables 8 and 9 list representative
results from the filtration study. These data show that the membrane pore
size used for sample preparation does not affect the concentrations of Ca, Mg,
and Na in solution. The same was found to be true for K and Mn.
However, the Fe and Zn concentrations showed a definite effect from fil-
ter pore sizes. The highest Fe and Zn concentrations were found in the sam-
ples filtered through the 3.0 u pore size membrane. The next highest concen-
trations were found in the samples filtered through the 0.45 um filters. The
lowest concentrations of Fe and Zn were found in samples filtered through the
0.22 ym filters. The disparity in concentration values was considerably
greater for those samples filtered through a 3.0 um pore size membrane than
for those filtered through the 0.45 and 0.22 ym pore size membranes. This was
especially true for the site 6-DUP samples.
The site 5-BRD samples did not show as great a difference in Fe concen-
tration between those samples filtered with the 3.0 um pore size membranes and
those filtered through either 0.45 or 0.22 um pore size membranes. The site
6-DUP samples were very turbid, and colloidal material was observed to pass
through the 3.0 um membrane but not the 0.45 or 0.22 um pore size membranes.
Subsequent acidification of the samples for preservation resulted in leaching
or dissolution of Fe and Zn from these colloidal particles.
The 4-L samples taken at the tenth well volumes from sites 5-BRD and
6-DUP during the February samplings were used for the storage analysis. The
results of this phase of the project are presented in Table 10.
Calcium, K, Mg, Mn, and Na show little change in chemical composition in
response to storage time. During the 72-hour period of the study, these con-
stituents displayed only a slight fluctuation in concentration. This fluctua-
tion may have been due to difficulty in subsampling the 4-L bottles.
The changes in pH, Fe, and Zn, however, were immediate. During the 7
hours between collection of the samples and return to the laboratory, where a
subsample of the larger volumes was taken, these parameters underwent signifi-
cant change. The initial pH for samples from both sites increased 0.3 and
0.4 pH units for sites 5-BRD and 6-DUP, respectively, during the first 7
hours. After that time, the values remained constant. Almost all of the
56

-------
Table 8. Sites 2-ELG, 3-FLR, 5-BRD, and 6-DUP: Calcium,
Magnesium and Sodium Concentrations of Samples Filtered Through
Different Pore Size Membranes
(Concentrations in mg/L)
Wei 1 Pore 	Ca	 	Mtj	 	Na	
Volume Size Site 3 Site 5 Site 3 Site 6 Site 2 Site 6
0	3.0
.45
.22
1/2 3.0
.45
.22
1	3.0
.45
.22
1-1/2 3.0
.45
.22
2	3.0
.45
.22
4 3.0
.45
.22
6 3.0
.45
.22
8 3.0
.45
.22
10 3.0
.45
.22
39.5
38.8
39.5
76.6
75.9
75.1
34.5
34.4
34.9
72.6
72.0
72.6
34.0
33.4
34.6
70.9
70.5
71.1
32.9
33.3
33.5
70.1
69.5
69.7
32.8
32.5
33.1
73.2
73.0
73.3
35.5
35.6
35.6
71.9
72.0
71.4
35.6
36.3
36.6
72.4
71.7
72.6
35.5
35.3
35.1
73.0
72.2
73.3
35.7
36.1
36.4
71.7
72.5
74.6
15.7
118
15.2
114
15.6
113
14.6
122
14.7
127
14.8
124
14.7
121
14.4
118
14.9
121
14.1
119
14.3
124
14.3
121
14.2
121
14.0
120
14.2
122
15.8
137
15.9
135
15.9
136
16.0
135
16.3
134
16.7
132
16.1
134
15.9
136
15.9
131
16.4
130
16.5
131
16.6
132
17.8
192
17.9
222
17.9
228
14.6
262
14.9
260
15.2
252
14.7
268
14.6
238
14.5
250
14.2
224
13.9
246
14.0
248
17.3
246
14.2
200
14.0
250
13.4
250
13.6
246
13.4
248
13.8
258
13.9
260
13.8
248
16.2
262
16.1
258
15.9
258
19.5
266
19.4
266
19.7
268
57

-------
Table 9. Sites 5-BRD and 6-DUP: Iron and Zinc Concentrations of Samples
Filtered Through Different Pore Size Membranes
(Concentrations in mg/L)



Fe


Zn

Well
Pore
Site 5
Site 6
Site 6
Site 5
Site 6
Site i
Volume
Si ze
(mech)*
(ai r)**
(mech)
(mech)
(mech)
(air)
0
3.0
14.7
3.44
4.89
4.69
.428
.137

.45
14.6
.52
4.04
4.54
.066
.078

.22
14.1
.49
2.69
4.45
.077
.075
1/2
3.0
18.1
3.40
7.65
2.22
.039
.088

.45
17.7
.62
2.92
2.09
.037
.057

.22
17.3
.51
2.30
2.00
.032
.053
1
3.0
15.4
3.67
7.35
1.54
.039
.064

.45
14.9
.61
2.61
1.43
.031
.043

.22
15.1
.52
2.51
1.43
.024
.044
1-1/2
3.0
13.1
3.12
8.37
1.19
.039
.079

.45
12.6
.63
6.96
1.11
.036
.049

. .22
12.4
.49
3.02
1.08
.022
.044
2
3.0
11.7
4.67
4.52
1.09
.037
.060

.45
11.1
.45
1.46
.89
.023
.035

.22
11.1
.42
1.39
.88
.022
.037
4
3.0
9.23
5.46
1.41
.66
.034
.084

.45
9.00
.62
.47
.51
.022
.056

.22
8.82
.50
.50
.51
.022
.053
6
3.0
9.01
4.10
.93
.55
.032
.063

.45
8.64
.49
.52
.37
.019
.037

.22
8.66
.44
.74
.36
.019
.037
8
3.0
8.09
4.91
1.67
.42
.026
.044

.45
7.75
.46
.78
.28
.017
.064

.22
7.76
.46
.75
.26
.021
.030
10
3.0
7.56
4.84
2.66
.39
.021
.040

.45
7.36
.46
1.10
.24
.016
.024

.22
7.33
.44
1.31
.22
.018
.032
* mech = peristaltic pump
** air = air lift pump
58

-------
dissolved Fe and Zn present in the samples was lost during the first 7 hours
of storage. This loss could be due to adsorption onto the walls of the con-
tainer. The latter is unlikely due to the magnitude of the concentration de-
crease. The most probable cause of reduction of the Fe concentrations is pre-
cipitation.
Parameters such as pH, Eh, and specific conductance should be measured in
the field at the time of sample collection. A closed cell that permits these
measurements to be made on samples before they are exposed to atmospheric con-
ditions is recommended. All samples should be filtered through a 0.45 pm pore
size membrane while on-site, immediately after sample collection. The samples
should then be preserved and stored in linear polyethylene bottles and return-
ed to the laboratory according to recommended methods (U.S. EPA, 1979).
Table 10. Sites 5-BRD and 6-DUP: Analysis of the Tenth Well Volume
Sample Used for Storage Study
(Concentrations in mg/L)
Site 5-BRD
Hours after collection-
preservation
pH
Ca
Fe
K
Mq
Mn
Na
Zn
0
6.7
111.
11.6
32
44.6
3.81
186
0.18
7
7.0
110.
CO
CO
•
31
41.9
3.15
172
0.02
24
7.0
104.
<.03
35
41.4
3.10
181
0.02
48
7.0
99.
.03
33
39.5
2.98
171
0.02
Site 6-DUP
Hours after collection-
before preservation
6.8
32.5
5.74
189
94.2
ND*
215
ND*
7.2
35.2
<.08
203
110.
ND*
242
ND*
7.2
32.0
<.08
204
96.6
ND*
223
ND*
7.2
30.1
.34
184
89.1
ND*
204
ND*
*ND - not detectable
59

-------
REFERENCES
Brown, E., M. W. Skougstad, arid M. J. Fishman. 1970. Methods for Collection
and Analysis of Water Samples for Dissolved Minerals and Gases. Tech.
Water-Resources Inv., Bk. 5, Chap. Al, U.S. Geological Survey.
Chian, Edward S., and Foppe B. DeWalle. 1975. Compilation of Methodology
Used for Measuring Pollution Parameters of Sanitary Landfill Leachates.
U.S. Environmental Protection Agency, Cincinnati, OH. EPA-600/3-75-011.
Fenn, D., E. Cocozza, J. Isbister, 0. Braids, B. Yare, and P. Roux. 1977.
Procedures Manual for Ground Water Monitoring at Solid Waste Disposal
Facilities. U.S. Environmental Protection Agency, Cincinnati, OH. EPA/
530/SW-611.
Gibb, J. P., K. Cartwright, D. Lindorff, and A. Hartley. 1978. Field Verifi-
cation of Hazardous Waste Migration from Land Disposal Sites, EPA publi-
cation in press.
Gilkeson, R. H., K. Cartwright, L. R. Follmer, and T. M. Johnson. 1977. Con-
tribution of Surficial Deposits, Bedrock and Industrial Wastes to Certain
Trace Elements in Ground Water. Proceedings of the 15th Annual Symposium
on Engineering Geology, Pocatello, Idaho.
Hughes, G. M., R. A. Landon, and R. N. Farvolden. 1968. Hydrogeology of
Solid Waste Disposal Sites in Northeastern Illinois. Progress Report to
the Department of Health, Education, and Welfare on Demonstration Grant
No. 5-D01-00006-02.
Hughes, G. M., R. A. Landon, and R. N. Farvolden. 1971. Hydrogeology of
Solid Waste Disposal Sites in Northeastern Illinois. U.S. Environmental
Protection Agency, Washington, D.C. SW-12d.
Jackson, R. E., and K. J. Inch. 1980. Hydrogeochemical Processes Affecting
the Migration of Radionuclides in a Fluvial Sand Aquifer at the Chalk
River Nuclear Laboratories. Environment Canada, NHRI Paper No. 7, Scien-
tific Series No. 104.
Jacob, D. E. 1950. Flow of Ground Water. Engineering Hydraulics, edited by
H. Rouse, John Wiley and Sons, New York.
Jenne, E. A. 1968. Controls of Mn, Fe, Co, Ni, Cu, and Zn Concentrations in
Soils and Water: The Significant Roles of Hydrous Mn and Fe Oxides, j_n
Trace Organics in Water. Amer. Chem. Soc. Adv. Chem., Ser. 73.
Kennedy, V. C., G. W. Zellweger, and B. F. Jones. 1974. Filter Pore-Size
Effects on the Analysis of Al, Fe, Mn, and Ti in Water. Water Resources
Research, vol. 10, p. 785-790.
Manahan, S. E. 1972. Environmental Chemistry. Willard Grant Press, Boston,
Massachusetts.
60

-------
Mooij, H., and F. Rovers. 1975. Recommended Groundwater and Soil Sampling
Procedures. Environment Canada, Seminar Proceedings, Sept. 18 and 19.
Papadopulos, I. S., and H. Cooper. 1967. Drawdown in a Well of Large Diame-
ter. Water Resources Research, 3(1):241-244.
Rattonetti, A. 1976. Stability of Metal Ions in Aqueous Environmental Sam-
ples. National Bureau of Standards Special Publication 422, p. 633-648.
Schafer, D. C. 1978. Casing Storage Can Affect Pumping Test Data. The
Johnson Drillers Journal.
Schmidt, K. D. 1977. Water Quality Variations for Pumping Wells. Ground-
water, vol. 15, no. 2, p. 130-137.
Shendrikar, A. D., V. Dharmarajan, H. Walker-Merrick, and P. W. West. 1975.
Adsorption Characteristics of Traces of Barium, Beryllium, Cadmium, Man-
ganese, Lead, and Zinc on Selected Surfaces. Anal. Chem. Acta, 84, p.
409-417.
Sommerfeldt, T. G., and D. E. Campbell. 1975. A Pneumatic System to Pump
Water from Piezometers. Ground Water, 13(3), p. 293.
Struempler, A. W. 1973. Adsorption Characteristics of Silver, Lead, Cadmium,
Zinc, and Nickel on Borosilicate Glass, Polyethylene, and Polypropylene
Container Surfaces. Anal. Chem., vol. 45, no. 13.
Stumm, W., and J. J. Morgan. 1970. Aquatic Chemistry. New York: Wiley-
Interscience.
Subramanian, K. S., C. L. Chakrabarti, J. E. Sueiras, and I. S. Maines. 1978.
Preservation of Some Trace Metals in Samples of Natural Waters. Anal.
Chem., vol. 50, no. 3.
Summers, W. K., and C. A. Brandvold. 1967. Physical and Chemical Variations
in the Discharge of a Flowing Well. Ground Water, vol. 5-6, no. 1.
Theis, C. V. 1935. The Relation Between the Lowering of the Piezometric Sur-
face and the Rate and Duration of Discharge of a Well Using Ground-Water
Storage, Trans. Am. Geophys. Union, 16, 518-524.
Trescott, P. C., and G. F. Pinder. 1970. Air Pump for Small-Diameter Piezom-
eters. Ground Water, vol. 8, no. 3.
United States Environmental Protection Agency. 1979. Methods for Chemical
Analysis of Water and Wastes. EPA-600/4-79-020. Cincinnati, Ohio.
Wagemann, R., and C. J. Brunskill. 1975. The Effect of Filter Pore-Size on
Analytical Concentrations of Some Trace Elements in Filtrates of Natural
Water. Intern. J. Environ. Anal. Chem., vol. 4, pp. 75-84.
61

-------

-------