EPA Report Number
June, 1991
PARAMETERS AFFECTING THE MEASUREMENT OF
HYDRAULIC CONDUCTIVITY FOR SOLIDIFIED/STABILIZED WASTES
by
DJ. Conrad, S.A. Shumborski, L.Z. Florence, AJ. Liem
Alberta Environmental Centre
Vegreville, Alberta, Canada TOB 4LO
CR814860-01-1
Project Officer
C. Mashni
Risk Reduction Engineering Lab.
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
This study was conducted in cooperation with
the Alberta Environmental Centre,
Vegreville, Alberta, Canada TOB 4LO
Risk Reduction Engineering Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Cincinnati, Ohio 45268
-------
-------
NOTICE
The information in this document has been funded in part by the U.S. Environmental
Protection Agency under Cooperative Agreement # CR814860-01 with the Alberta
Environmental Centre, Vegreville, Alberta, Canada. It has been subjected to the Agency's
peer and administrative review and approved for publication as an EPA document. Mention
of trade names or commercial products does not constitute endorsement or recommendation
for use.
- 11 -
-------
-------
FOREWORD
Today's rapidly developing and changing technologies and industrial products and
practices frequently carry with them the increased generation of materials that, if
improperly dealt with, can threaten both public health and the environment. The U.S.
Environmental Protection Agency is charged by Congress with protecting the Nation's
land, air, and water resources. Under a mandate of national environmental laws, the
Agency strives to formulate and implement actions leading to a compatible balance
between improving the quality of life and minimizing the risks to the environment. These
laws direct the EPA to perform research to define our environmental problems, measure
the impacts, and search for solutions.
The Risk Reduction Engineering Laboratory is responsible for planning,
implementing, and managing research, development, and demonstration programs to
provide an authoritative and defensible information that can be used by both regulators
and the regulated in their common efforts to protect the environment from the hazards
of industrial and municipal waste. This publication is one of the products of that
research and provides a vital communication between the researcher and the user
community.
This report describes the effects of varying cement content on the hydraulic
conductivity of a solidified hazardous waste. The goal of this work is to gain a better
understanding of the mobility of constituents of solidified/stabilized hazardous waste and
how to minimize that mobility. This information should be of assistance to regulators
and businesses subjected to the waste management requirements of the Resource
Conservation and Recovery Act.
E. Timothy Oppelt, Director
Risk Reduction Engineering Laboratory
- iii -
-------
-------
ABSTRACT
A series of experiments conducted at the Alberta Environmental Centre examined the
variation in hydraulic conductivity (K) within and among three matrices formed by steel mill
baghouse dust treated with 8%, 9% and 10% Normal Portland Cement at a water/cement ratio
of 1:1. Within the 8% and 9% matrices, test gradient (i) and back pressure (P) were combined
into 3x3 factorial treatments. Commercially available equipment was modified to allow
sensitive and .continuous monitoring of hydraulic conductivity. A permeant-matrix interaction
was indicated by K decreasing with time at a rate which increased with higher cement
contents. After hydraulic conductivity testing, the samples were examined by scanning
electron microscopy and energy dispersive x-ray analysis. A cement hydration product,
identified as ettringite, had formed in the solidified/stabilized waste pores. This product
reduced hydraulic conductivity by two orders of magnitude by restricting conducting pores.
Four to seven weeks of testing were required before an acceptable equilibrium had been
reached and statistical comparisons among the i x P treatments were made. Within each
matrix, gradient was statistically the most significant parameter accounting for 60% of the
variation in results. The response to gradient was different than that observed with clay and
soil-liners in the literature. The overall mean hydraulic conductivity (p<0.01) of the 8%
matrix (10 ± 5 x 10"6 crasec"1) was significantly greater than that of the 9% matrix (0.06 ±
0.03 x 10'6 cm-sec'1) (p<0.01).
Temporal effects, gradient and cement content were identified as important factors
affecting hydraulic conductivity measurements and must be considered by regulatory tests.
Bulk density was a useful quality control criterion for minimizing sample variance within each
matrix.
- iv -
-------
This work was submitted in fulfillment of Assistance ID No. 814860-01-1 under the partial
sponsorship of the U.S. Environmental Protection Agency. This report covers a period from
October 14, 1988 to December 31, 1990 and work was completed as of December 31, 1990.
- v -
-------
CONTENTS
FOREWORD iii
ABSTRACT iv
FIGURES viii
TABLES x
ACKNOWLEDGEMENTS xi
1.0 INTRODUCTION 1
2.0 CONCLUSIONS AND RECOMMENDATIONS 4
3.0 LITERATURE SEARCH 6
4.0 MATERIALS AND METHODS 10
Raw Materials 10
Sample Preparation and Acceptance 10
Instrumentation . 14
Sample Saturation 16
Sample Porosity . 16
Sample Characterization 17
Electron Microscopy 1 ..... 17
Chemical Analyses . . 17
5.0 EXPERIMENTAL PROCEDURES , 20
Sample Preparation 20
Hydraulic Conductivity Data Recording 20
Determination of Hydraulic Conductivity
Equilibrium Conditions 20
Model Testing and Test Precision 21
Experimental Design 21
Physical Characteristics . 22
6.0 RESULTS AND DISCUSSION 23
Choice of Raw Materials 23
Bulk Density Measurements 25
Temporal Effects During Equilibration 26
- vi -
-------
Model Testing and Test Precision ....... 37
Solubility Effects During Testing 38
Response Surface Regression Analysis „ 42
Effect of Matrix ; 42
Effect of Instrument Parameters 44
Effect of Sample Porosity 50
Saturation Considerations 51
Comparison of Solidified/Stabilized Waste
to Soil/Clay Liners 53
REFERENCES 55
APPENDICES 59
A. Calibration of Geotest Permeameter Interfaces 59
B. In-House Computer Program • 62
C. Experimental Data 65
vii -
-------
FIGURES
1. Modified Sample Support 12
2. Modified Tamping Bar 13
3. Schematic of Hydraulic Conductivity Apparatus 15
4. Flexible Wall Membrane Arrangement 18
5. Variation of Hydraulic Conductivity at
8%, 9% and 10% Cement 27
6. Fibrous Growth in Tested 8% Matrix (lOOOx) 30
7. Fibrous Growth in Tested 8% Matrix Showing
Morphological Similarities to Ettringite (6000x) 30
8. X-ray Analysis of Individual Fibres 31
9. Ettringite Growth in Pores of the 9% Matrix
Sample (lOOOx) 32
10. Ettringite Growth Blocking Pores in the 9%
Matrix Sample (lOOOx) 32
11. Ettringite Growth in the 10% Matrix Pores (200x) 34
12. Ettringite Growth in the 10% Matrix Pores (lOOOx) 34
13. Samples Cured in Humidity Chamber Showing No
Evidence of Ettringite .35
14. Cement Control Sample Cured Under Water 36
15. Cement Control Sample Cured in the Humidity Chamber 36
16. Deviation from the Predicted 8% Model ...'.. 39
17. Hydraulic Conductivity Replicates for 8% Matrix
at Median Levels of i and P 39
18. Leaching of Hydraulic Conductivity Samples During Testing 40
19. Log K vs Cumulative TDS of Median Levels of i and P 42
20. Predicted Values for the 8% Matrix Showing the Predicted Maxima . . 47
21. Contour Plot of Predicted Values for the 8% Matrix Showing Hydraulic
Conductivity Contour Intervals 47
- vui -
-------
22. Response Surface Plot of Predicted Values for the 9% Matrix Showing
the Linear Relationship of Gradient and Back Pressure 48
23. Contour Plot of Predicted Values for the 8% Matrix Showing
Hydraulic Conductivity Intervals 48
A.1 Apparatus for Calibration of Geotest Permeameter Interfaces 60
- rx -
-------
. TABLES
.1. Hydraulic Conductivity Sample Formulations 11
2. Sample Compaction Procedure Modifications 11
3. Physical Test Methods 19
4. Test Parameters and Levels 21
5. Baghouse Dust Metal Analyses 24
6. Sample Bulk Densities 25
7. Model Coefficients for the Change in Hydraulic Conductivity
with Time .26
8. Descriptive Statistics of Log10k for the 8% Matrix Replicates 38
9. Composite Permeant Analyses of the 9% Matrix 41
10. Response Surface Regression Analyses 46
11. Sample Porosity -.50
12. Sample Saturation 52
13. Unconfined Compressive Strength Tests (kPa) .. 53
A.1 Interface Calibrations ...'.. 61
C.I Experimental Trial Data 64
C.2 8% Matrix Replicate Data .66
- x -
-------
-------
ACKNOWLEDGEMENTS
The authors wish to thank the contributors in Alberta business, research institutions and
the Alberta Environmental Centre who assisted in the production of this study. Specifically
the contributions of the following individuals and companies are noted: Mr. A. Slessor of
Lafarge Canada Inc. and Stelco, Edmonton (Alberta) Works for contributing raw materials;
Ms. P. Soldan, Ms. C. Jackson and Ms. M. Zadkovich (Environmental Technology Division)
for their assistance and patience in manuscript preparation; Mr. J. Kirtz, Mr. P. Henry (Animal
Sciences Division), Mr. K. Klingbeil (Environmental Technology) and Mr. M. Herbut (Plant
Sciences Division) for computer graphics; Dr. M. Wilson for his valuable suggestions during
the course of this project; and Dr. D. Ivey (University of Alberta), Dr. R. Mikula (Energy,
Mines and Resources Canada), and Dr. M. Neuwirth (Chemistry Division) hi providing SEM,
STEM and X-ray analyses.
- xi -
-------
-------
SECTION 1.0
INTRODUCTION
Solidification and stabilization (SS) technologies are often used to treat -hazardous
wastes to reduce the environmental impact of then- disposal. Solidification removes free
water, usually by the hydration reactions of lime or cementitious materials, producing a
monolithic solid with reduced surface area. Stabilization with cementitious materials reduces
the solubility of wastes by the alkaline precipitation of metal hydroxides or metal
incorporation into the hydration products of cement.
The long term behaviour of these treated wastes is the subject of much concern.
Depending on the disposal scenario, treated wastes are eventually subject to leaching by
ground water, precipitation, or leachate. If the treated waste is relatively permeable, leachant
flow will be through the whole of the material rather than being confined to the external
surface area. Thus, a major benefit of SS treatment, the reduction of surface area available
for leaching is compromised.
The flow of liquid through a porous medium is described by Darcy's Law. The liquid
superficial velocity (Flowrate/Area) per unit gradient is defined as hydraulic conductivity,
which is a function of the properties of the medium and the liquid. Gradient is defined as the
headloss which occurs over the sample (cm of H2O) divided by the .sample length (cm).
Darcy's Law may be written as:
where K is hydraulic conductivity (crasec"1), Q is flow rate (cm3.sec"1), i is gradient
(dimensionless) and A is cross-sectional area (cm2).
-------
Permeability, a property of the medium alone, is related to hydraulic conductivity by
the following relationship:
PS
where K is permeability (cm2), u is absolute viscosity of the liquid (g.cm^.sec"1), p is density
of the liquid (g.crn"3) and g is acceleration of gravity (980 cm.sec"2).
When there is no medium-liquid interaction, permeability is an intrinsic and useful
property of a medium. The flow rates of different liquids through a medium can be readily
predicted from its permeability and the properties of the liquids. However, when there are
changes in liquid properties, due to dissolution, or in the internal structure of the medium, as
shown in this paper, the meaning of permeability becomes obscure. Since it is the flow rate
of aqueous permeant through SS waste which is of environmental interest, hydraulic
conductivity is the proper terminology and is used herein.
The literature available on hydraulic conductivity measurement with environmental
implications deals predominantly with clay and soil liners. Researchers are interested in the
effects of permeants, specifically inorganic salt solutions (1), organic fluids (2) and landfill
leachates (3). Test parameters such as saturation (4), temporal effects (5, 6) and gradient (6,
7) have been studied.
The corresponding information on solidified/stabilized waste is, however, practically
non-existent Nor is there sufficient information which' addresses the differences between clay
or soil liners and solidified/stabilized waste, such as compressive strength and permeant-matrix
interactions.
This report deals with the effects of parameters affecting the measurement of hydraulic
conductivity of solidified/stabilized waste. The study was undertaken to form bases for the
development of a regulatory test method - to improve intra and inter-laboratory precision - and
to correlate accelerated laboratory test results to those occurring under field conditions.
-2-
-------
The scope of the investigation was confined to the following:
one waste, steel mill baghouse dust, treated with three portions of Normal Portland
.Cement to produce a range of hydraulic conductivities similar to those found in
commercial solidification/stabilization processes.
the following test and instrument parameters: sample preparation, temporal effects,
gradient and back pressure.
New equipment was acquired and modified to allow for sensitive, accurate and
continuous flow measurements. Particular attention was given to minimize variance due to
sample preparation, and statistical methods and experimental designs were used to delineate
parametric effects. Electron microscopy, chemical analyses and measurements of physical
properties were conducted to assist in the interpretation of the observed phenomena.
-3-
-------
SECTION 2.0
CONCLUSIONS AND RECOMMENDATIONS
1. Hydraulic conductivity was sensitive to matrix composition. Significantly different
hydraulic conductivities were measured between samples differing only by 1% in
cement content. The ability to distinguish such samples was attributed to the
institution of a strict quality control criterion for sample preparation based on bulk
density. The corollary is that the variance of hydraulic conductivity due to sample
preparation can be minimized by using that criterion.
2. Hydraulic conductivity decreased with elapsed time during testing. A power function
in the form of y = axb describes the relationship. The decrease could be explained by
long-term cement hydration reactions forming ettringite in the permeant-conducting
pores. Although matrix dissolution occurred, no effect was observed over the testing
period of up to 80 days.
3. The effects of gradient and back pressure on hydraulic conductivity were the
following:
- Gradient was the most significant parameter and its correlation with hydraulic
conductivity was positive, the opposite to that for soil and clay liners. This was
attributed to the higher unconfined compressive strength and the corresponding lesser
degree of sample consolidation.
- Medium and high levels of gradient and back pressure were the less sensitive region
for hydraulic conductivity measurements. Falling-head permeameters, in which low
levels of these parameters are used, are thus operated hi the more sensitive region.
- Over the entire region of the chosen experimental levels, the measured hydraulic
conductivities varied by a factor of four or less. Values obtained in the laboratory are
thus reasonable estimates of those in the field conditions, provided that compaction and
curing conditions are similar.
-4-
-------
4. The precision attainable for hydraulic conductivity measurements utilizing the quality
control criteria, waste type and instrumentation of this study was improved by nearly
50% over that reported in the literature.
5. An exponential relationship between hydraulic conductivity and sample porosity was
shown to be statistically significant. How such a relationship varies with different
matrices was not investigated.
Recommendations for regulatory test development:
1. Bulk density should be used as a quality control criterion to reduce variance due to
sample preparation.
2. To improve precision, temporal effects should be taken into account and measurements
carried out at high levels of gradient and back pressure.
3. To estimate maximum hydraulic conductivity, measurements should be made as soon
as the sample is cured.
Recommendations for future work: .
1. Different matrices should be tested for morphological changes during testing and
improving the confidence in the effect of instrument parameters.
2. Saturation effects should be studied to predict field hydraulic conductivity.
-5-
-------
SECTION 3.0
LITERATURE SEARCH
The literature available on hydraulic conductivity measurement with environmental
implications deals predominantly with clay and soil-liners. Few references exist in the
literature concerning the measurement of hydraulic conductivity of solidified/stabilized wastes.
This section will briefly review the available literature in terms of the research undertaken
from a variety of sources and the direction it suggests for solidified/stabilized waste.
Clay and soil-liner researchers are interested in the effects of dissolved inorganic and
organic materials on clay-liner permeability as measured by changes in permeant velocities.
A study by Ather et al (1) investigated the effects of inorganic permeants upon the
hydraulic conductivity of bentonite. Increased ionic strength and cationic valence increased
the measured hydraulic conductivity. The Gouy-Chapman double layer model of hyclrated
ions was used to explain postulated changes in the flocculated nature of the bentonite. The
degree of flocculation would affect the permeability of the bentonite clay.
Lentz et al (8) observed changes in hydraulic conductivity due to cation exchange and
salt precipitation in a magnesium - montmorillonite clay when an alkaline permeant (NaOH)
of pH 13 was used.
Acar et al (2) found that pore size distributions of a kaolinite clay were unaffected by
a change in permeant from 0.1 N CaSO4 to an organic fluid. However, organic solvents of
low water solubility, such as benzene and nitrobenzene, were found to decrease the measured
hydraulic conductivity by three orders of magnitude. Soluble organics, such as phenol and
acetone, had minimal effects. Thus, differences in measured hydraulic conductivity were due
to the ability of solvents to displace pore water.
-6-
-------
The work of the above authors suggests that interactions between permeants and clays
have measurable effects on hydraulic conductivity. Bowders (9), discussing the work of
Pierce and Witter (10) in determining termination criteria for clay permeability testing,
stressed that chemical equilibrium must be established when chemically reactive permeants
are used. For clays this would mean monitoring influent and effluent chemical characteristics.
Bowders agrees with Pierce and Witter on their termination criteria when water is used as the
permeant; one pore volume of flow has been passed and hydraulic conductivity vs cumulative
pore volumes cannot be shown to differ significantly from zero.
The degree of sample saturation has been found to affect hydraulic conductivity.
Elzeftawy and Cartwright (4) used soil water retention curves to predict hydraulic
conductivity. They found that the hydraulic conductivity of a soil at 100% saturation was
approximately four orders of magnitude greater than at 50% saturation. Carpenter and
Stephenson (6) observed that saturation generally increased during hydraulic conductivity
testing.
Temporal effects have been noted. Parker et al (5) observed that the permeability of
flyash stabilized soils decreased over 13 days of testing possibly due to a flyash - soil
interaction. Carpenter and Stephenson (6) noted decreased hydraulic conductivity for clays
over shorter time periods (2 hours).
The effect of applied gradient has been studied. Carpenter and Stephenson (6) and Edil
and Erikson (7) both noted that hydraulic conductivity declined for clays as the gradient was
increased when flexible wall permeameter cells were used. The authors explained these
results as due to sample consolidation. Edil and Erikson (7) noted that at high gradients (290-
360) gas bubbles were evolved in the effluent lines. This suggests an alternative explanation.
High pressure drops across a sample produce unsaturated conditions, due to liquid degassing,
which result in decreased hydraulic conductivity.
-7-
-------
Bryant and Bodocsi (11) collected historical data on hydraulic conductivity
measurements for clay liners and analyzed them for the effects of sample variation,
preparation, equilibration and gradient. They noted many confounding effects, and suggested
that suitable experimental designs should be chosen to properly estimate parametric effects.
Longer test periods and statistical approaches to determine equilibrium, as indicated by stable
hydraulic conductivity, were suggested. Soil hydraulic conductivity was found to be very
sensitive to preparation technique. Some results showing decreased hydraulic conductivity
were explained by sample consolidation resulting from increased gradients. -
Pierce et al (12), who conducted ruggedness tests using both rigid wall and triaxial cell
permeameters, found that water content, lift thickness and back pressure had the greatest effect
on the measured hydraulic conductivity of a clay liner. The first two factors pertain to sample
preparation while the third is an instrument measurement parameter. Gradient was not found
to be significant at relatively high levels (i = 100, 200), typical of laboratory tests. The
hydraulic conductivity results exhibited large variability. Thus, inter-laboratory results could
exhibit large variation due to individual laboratories performing hydraulic conductivity tests
at different levels of these sensitive parameters.
Cement and concrete researchers have investigated the permeability of hardened cement
pastes, as cement permeability will affect the weathering properties and corrosion of metal
reinforcing rods. This work has some applications to solidified/stabilized wastes because low
permeability products are desired for both products and industrial cements are often used as
solidification/stabilization additives.
Powers et al (13) noted that long term curing of cement pastes revealed a seven order
of magnitude decline in permeability compared to that of the fresh paste. This was explained
by the fact that cement gel occupies 2.1 times the volume of the unhydrated paste. Thus, long
term curing causes discontinuities in capillary porosity (14) and decreased hydraulic
conductivity. Patel et al (15) noted a similar phenomena, that large pore fractions decreased
while gel (paste) porosity increased during cement hydration. Later research has attempted
-8-
-------
to develop models for hydraulic conductivity based on pore parameters. Nyame and Illston
(16) described their hydraulic conductivity data in terms of hydraulic radius theory. Hughes
(17) considered pore isotropy and tortuosity in developing a hydraulic conductivity model.
Stegemann and Cote (18) included hydraulic conductivity in an inter-laboratory
evaluation of solidified/stabilized waste. A falling head method with a triaxial cell was used
to measure hydraulic conductivity for thirty-seven waste samples. Commercially available
solidification/stabilization treatment processes yielded hydraulic conductivities between 3 x
10"4 and 7 x 10"8 cm-sec"1. At the interlaboratory level, the variance was too large to
determine differences between individual products. A pooled estimate of intralaboratory
variance (which does not include the systematic error of each laboratory) was made. Four
replicates of the test were estimated to yield a one order of magnitude precision at the 95%
confidence level. (Based on a log-normal data distribution, the actual precision is X/-T 7.3.)
Stegemann and Cote (18) felt that "differences between specimens of the same
solidified product appeared to be the main cause of the variability observed for this method".
This indicates waste heterogeneity and sample preparation as sources of variance. Bulk
density is one measure of sample homogeneity. Stegemann and Cote (18) reported a precision
for bulk density of ± 0.14 g.cmT3 for three replicates of the 95% confidence level. This would
correspond to a relative standard deviation of 3.9% at the reported median of 1.44 g.cm'3.
In summary, a review of the pertinent literature suggests several areas of consideration.
Clay and soil-liner research indicates that the instrument parameters of gradient and back
pressure, as well as sample preparation and temporal considerations, should be elements -of
a hydraulic conductivity study. Cement research indicates that cement hydration reactions
continue over a long time period. A study of solidified/stabilized waste indicates that, on an
interlaboratory and intralaboratory level, hydraulic conductivity as practiced offers poor
precision, but that this may largely be due to sample variance.
-9-
-------
SECTION 4.0
MATERIALS AND METHODS
RAW MATERIALS
<
The following raw materials were used to produce the surrogate solidified/stabilized
waste studied:
1. Steel Mill Baghouse Dust
2. 16 - 30 mesh silica sand (Badger/Cardium Service and .Supply)
3. ASTM Type I Normal Portland Cement (Canada Cement Lafarge)
4. Tap Water (North Saskatchewan River, conventional solids removal and partial lime
softening; typical, finished water pH 8.5, hardness 100 ppm as CaCO3)
SAMPLE PREPARATION AND ACCEPTANCE
The sample formulations, shown in Table 1 were chosen to give a range of hydraulic
conductivities typical of solidified/stabilized wastes (10~6 to 10"8 cm-sec"1).
Samples were prepared by adding weighed portions of the dry ingredients to a Hobart
(Model N-50) mixer and mixing for two minutes at low speed (62 rpm, 139 rpm planetary).
The required water was added and mixing continued for two minutes with three pauses to
scrape material from the side of the mixing bowl.
- 10-
-------
Table 1. Hydraulic Conductivity Sample Preparations
Matrix
8%
9%
10%
Steelmill
Foundry
Dust (wt.%)
• 42.0
41.0
40.0
Silica
Sand (wt.%)
42.0
41.0
40.0
Typel
Portland
Cement (wt.%)
8.0
9.0
10.0
Water (wt.%)
8.0
9.0
10.0
Samples were compacted in plastic molds (M.A. Industries, Peachtree, Georgia) using
method ASTM D558-82 (19), modified for the smaller sample size according to Table 2. The
modified sample preparation support and tamping bar are detailed in Figures 1 and 2
respectively.
Table 2. Sample Compaction Procedure Modifications
Present Study ASTM D558-82
Mold: Diameter (cm)
Height (cm)
Volume (cm3)
Sample: Compacted Height (cm)
No. of lifts
Blows/layer
Rammer: Diameter (cm)
Area (cm2)
Mass (g)
Drop (cm)
Compactive Energy (J.cm"3)
7.62
15.2
693
16.5
3
25
3.805
11.37
1390
30.48
0.448
10.16
11.643
944
12.7
3
25
5.08
20.30
2490
30.48
0.591
-11-
-------
\\\\\\\\\\\N
cs
1
§•
CO
I
I
4J
CL,
I
s
i
-------
c
in
I
cu
•<*•
ru
ci
c
in
cd
e
u
-p O)
- ^ I,
<* -a ,
« "g.
u t. o-
d S|
o; —
o —<*•
cu
x
CO
(D
• o
-p
i
en
-------
Samples were leveled to the top of the mold with a metal straight edge and cured at
95% min. relative humidity at 23°C for a minimum of 28 days.
After curing, samples were removed from their, molds, trimmed to ensure parallel end
faces and their physical dimensions and mass determined prior to hydraulic conductivity
testing.
INSTRUMENTATION
Three flexible wall permeameters (Geotest Model S5425) were used in this study.
Materials of construction were compatible with hazardous waste testing. Metal parts in
contact with permeants were stainless steel (type 316). Permeant flow lines were made of
teflon and permeant interfaces were teflon-lined. The piston interface was sealed with Viton
"o-rings". The permeameters were modified by the substitution of the Transtek #0243 linear
variable displacement transducers, with Mitutoyo Digimatic Indicators Type ID-130ME
(accuracy of 0.001 mm over a 30 mm range)'to measure interface piston displacement.
Digitized signals of interface displacements were recorded by a Mitutoyo DP2-DX
miniprocessor, which is capable of storing up to 678 displacement and time records. An
additional modification was the introduction of a Brainard Kilman model S-545 bladder
interface. This modification isolated the air pressurization system from permeameter cell
water. A schematic of the permeameter/interface/datalogger arrangement is shown in Figure
3. Details of interface calibration are given in Appendix A.
Permeameter system pressures were measured by a Shape Instruments Ltd pressure
transducer model #SP1020 and displayed on a Shape Instruments Ltd transmitter model
#SD7500/C. The pressure measuring system was calibrated with a dead weight tester.
Results were downloaded to a Fujikama Model #fK 286 MT computer using in-house
produced software.. The computer program is described in Appendix B.
- 14-
-------
»n
I
I
u
I
u
u
CO
g
-------
SAMPLE SATURATION
Cured and trimmed samples were inserted into a latex membrane capped with porous
stainless steel disks (K > 1 x 10"4 cm-sec"1) and mounted on the permeameter pedestal. A
layer of aluminum foil covered with vacuum grease (for caustic corrosion protection) and a
second latex membrane were added to provide a barrier to gas diffusion into the sample from
the cell water. Details of the modified flexible wall membrane are shown in Figure 4. .
Samples were saturated by evacuating the sample for one hour from the outlet side
with a model D150 Precision vacuum pump protected by a liquid nitrogen cold trap.
Degassed water was introduced to the evacuated sample from the inlet end to begin saturation
and completed with the permeant pressurized to 7 kPa.
SAMPLE POROSITY . .
Sample porosity is defined as the ratio of interstitial volume to the total sample
volume. For a granular solid in the liquid saturated state, porosity is one of the parameters
affecting headless and thus flow rate. Porosity, (E, - dimensionless) may be determined by
the following relationship:
E = l-^-(l-W)
P
where pB is the wet sample bulk density (g.cm"3), p is the density of the solids or "true
density" (g.cm"3) and W is the proportion of water content in the sample (dimensionless).
The actual pore volume (PV) in cm3 may be determined by:
PV = Vs - Ms (1 " *>
P
where Vs is the sample volume (cm3) and Ms is the mass of the wet sample (g).
- 16-
-------
SAMPLE CHARACTERIZATION
The methods used to measure the physical characteristics of the hydraulic conductivity
samples and the exuded permeants are listed in Table 3.
ELECTRON MICROSCOPY
Electron microscopical analyses were performed on a Hitachi S510 scanning electron
microscope (SEM), a Hitachi X-650 (SEM) with energy dispersive x-ray spectrometer and a
Hitachi H-600 scanning transmission electron microscope (STEM) equipped with a Kevex Be
window X-ray detector.
CHEMICAL ANALYSES
Chemical analyses of the foundry baghouse dust and exuded permeants was performed
utilizing the methods of ion coupled spectroscopy, atomic absorption spectroscopy, colorimetry
and potentiometric titration.
- 17-
-------
I
co
Q.
-------
Table 3. Physical Test Methods
Parameter
Method
Modifications
Moisture
ASTM D2216-80 (20)
Standard Method for Laboratory
Determination of Water Content of
Soil, Rock and Soil Aggregate
Mixtures
60°C
constant weight
Bulk Density
ASTM D558-82 (19)
Standard Test Methods for
Moisture-Density Relations of Soil-
Cement Mixtures
See Table 2
True Density
ASTM C604-86 (21)
Standard Test Method for True
Specific Gravity of Refractory
Materials by Gas-Comparison
Pycnometer
Drie.d at 60°C, ground
to pass a 1.18 mm
screen.
Unconfined
Compressive Strength
ASTM D1633-84 (22)
Standard Test Method for
Compressive Strength of Molded
Soil-Cement Cylinders.
Tested at 2.5 mm/min.
Total Solids AWWA #209 A (23)
Total Residue Dried at 103 - 105°C
- 19-
-------
SECTION 5.0
EXPERIMENTAL PROCEDURES
SAMPLE PREPARATION AND ACCEPTANCE
Samples for hydraulic conductivity testing were prepared for each matrix in batches
of approximately eight samples. Samples accepted for testing met a quality assurance criteria
of ± 0.5% of the mean of batch bulk density. (See Bulk Density Measurements in Results and
Discussion)
HYDRAULIC CONDUCTIVITY DATA RECORDING
Initial measurements were conducted at median experimental levels-of gradient (116)
and back pressure (69 kPa). Data were recorded when inflow and outflow rates were
determined to be within 5%.
DETERMINATION OF HYDRAULIC CONDUCTIVITY EQUILIBRIUM CONDITIONS
Simple, linear regression was used to identify the time (x = day = independent
variable) when changes in hydraulic conductivity (y= K « 106 = dependent variable) had
attained sufficient stable equilibrium that the regression coefficient, or slope, was not different
than zero, as suggested by Pierce and Witter (10).
-20-
-------
MODEL TESTING AND TEST PRECISION
Three 8% matrix samples were tested concurrently on the three Geotest permeameters
of median levels of gradient and back pressure over a twenty-nine day period. Their measured
hydraulic conductivity was compared to that predicted from the equation derived previously
for the 8% matrix.
EXPERIMENTAL DESIGN
Gradient and back pressure were varied and treated as independent continuous variables
at three levels in a completely randomized 3x3 factorial design (see Table 4). Gradient
levels corresponded to pressures of 10 kPa to 340 kPa. Two independent experiments were
performed at 8%, and 9% matrix to provide a range of hydraulic conductivities. Second-
degree polynomial coefficients were calculated using response surface regression (SAS 1988
Release 6.03 (24)) to model the response of hydraulic conductivity to varying levels of
gradient and back pressure.
Table 4. Test Parameters and Levels"1"
Parameter . Level
-1 0 +1
Gradient 8 116 227
Back Pressure
(kPa) 14 69 124"
* Performed at 8% and 9% matrix
-21 -
-------
PHYSICAL CHARACTERISTICS
Companion samples were tested for the criteria listed in Table 3, before commencing
hydraulic conductivity testing, to determine the initial porosity, saturation and unconfined
compressive strength. Hydraulic conductivity samples were tested for the same criteria to
determine if changes in physical characteristics could account for the changes observed in
hydraulic conductivity.
-22-
-------
SECTION 6.0
RESULTS AND DISCUSSION
CHOICE OF RAW MATERIALS
The raw materials used were chosen for this study to fulfill the following criteria:
1) the waste material should be produced in high volume, be amenable to
solidification/stabilization treatment and have wide concern in terms of North
American industry;
2) the solidification/stabilization treatment should be in common use and generic
rather than proprietary;
3) the solidified/stabilized waste treatments should produce values of measured
hydraulic conductivity typical of commercially treated solidified/stabilized
wastes.
The waste material chosen was a baghouse dust collected from a steel mill paniculate
emission control system. Baghouse dusts constitute a large portion of the 9000 tonnes of
wastes produced annually by the foundry industry in Alberta (population 2.4 million). A nitric
acid digestate of the baghouse dust used in this study was analyzed with an ion coupled
plasma spectrometer. Analyses are summarized in Table 5.
-23-
-------
Table 5. Baghouse Dust Metal Analyses
Metal
Fe
Zn
Pb
Cu
Cr
No. of
Samples
3
3
3
3
3
Mean
(mg/kg)
372000
53700
9740
2510
1450
Standard
Deviation
(mg/kg)
5240
762
552
44
19
This waste, consisting of environmentally sensitive heavy metals in an inorganic
matrix, is suitable for treatment by solidification/stabilization technologies. Foundry dusts are
classified in Canada under the Transportation of Dangerous Goods Act as Waste,Type 80.
In the United States the EPA specifies emission control dusts and sludges from the electric
furnace production of steel under Hazardous Waste No. K 061.
The solidification/stabilization treatment chosen for this study utilized ASTM Type I
Normal Portland Cement Cement treatment of wastes is widely practiced.
During preliminary sample preparation studies it was noted that swelling and surface
cracking of the samples occurred during curing. Samples which remained intact during curing
demonstrated very low hydraulic conductivities (- 10"10 onusec"1).
Diluting the baghouse dust with 16-30 mesh silica sand allowed a range of cement
treatment levels to be tested without sample swelling and cracking. Furthermore, this addition
of silica sand allowed samples with a range of hydraulic conductivities to be tested.
-24-
-------
BULK DENSITY MEASUREMENTS
The samples prepared in this study had the consistency of soil-cement and thus a
standard compaction method for soil cements was used. Bulk density measurements of these
samples are summarized in Table 6.
Table 6. Sample Bulk Densities
Matrix
n+
Bulk Density
(g.crn"3) Mean
Standard Deviation
Relative Standard
Deviation (%)
8%
16
2.398
0.024
1.00
9%
18
2.480
0.043
1.73
10%
16
2.527
0.009
0.363
+n is the number of samples •
Results from Table 6 indicate that different nominal matrix content samples may have
similar bulk densities even when standard compaction procedures are used. Bulk density is
a function of gross sample porosity and has been implicated as a source of variation in
hydraulic conductivity measurements. The degree of compaction has been found to affect the
measured permeability of clayey silt (25). Stegemann and Cote (18) postulated that sample
differences were a major source of variation for hydraulic conductivity in their inter-laboratory
comparison study of solidified/stabilized wastes. In this study, a concerted effort was made
to minimize this source of variation in order to study only the effects of instrument
parameters. Thus, a quality assurance criterion was instituted which rejected all samples with
bulk densities not within 0.5%. of the sample mean. This criterion ensured that discrete
sample populations were obtained for each nominal matrix concentration.
-25-
-------
TEMPORAL EFFECTS DURING EQUILIBRATION
Steady state flow as indicated by equal inflow and outflow rates of permeant was
obtained on the first or second day of testing. This steady state flow was measured over
several weeks to study if the samples and permeants were in an equilibrium state. Equilibrium
would be indicated by no significant change in hydraulic conductivity with time. The results1
illustrated in Figure 5 show the temporal effects over 80 days of testing at constant gradient
and back pressure. All samples displayed a marked and rapid decrease with respect to lime.
(Note the different scales.) Hydraulic conductivity decreased by nearly two orders of
magnitude for the highest cement content sample.
The decrease in hydraulic conductivity was modelled according to the equation:
K x 106 = A (T + 1)B
where T is the elapsed time measured from the first day of testing (T=0) and A and B are the
intercept (initial condition) and slope of the power function, respectively. The regression
results are summarized in Table 7, showing the statistical significance of the models.
Table 7. Model Coefficients* for the Change hi Hydraulic Conductivity with Tune
Matrix Initial Value (A) Standard Error Slope (B) Standard Error
8%
9%
10%
82.400
1.571
1.356
±4.361
±0.044
±0.148
-0.413
-0.743
-1.476
±0.025
±0.032
±0.070
'regressions are significant at p<0.001; the models explain 98% (minimum) of the variation
in hydraulic conductivity.
: Hydraulic conductivity results are summarized in Appendix C. For the purpose
of easy comparison, all hydraulic conductivity measurements discussed in
figures and surface response regressions have been multiplied by 1 x 106 unless
specified otherwise.
-26-
-------
100
0.2
.§• _~
6 -o
1 *
1 *
0.0
10%
10 20 30 40 SO 60 70
. H«pjcd Time (days)
FIGURE 5. Variation of Hydraulic Conductivity With Time at 8%, 9% and 10% Cement
-27-
-------
The models are highly significant and explain a large portion of the variation in
hydraulic conductivity with time. Temporal effects could be a source of variance within and
between laboratories if not accounted for.
The results of hydraulic conductivity suggest that permeants and samples are not in
physical/chemical equilibrium, but that some change in physical parameters is occuring.
The models also predict that with time hydraulic conductivity approaches "zero", but,
in fact, this will never be achieved. However, if no future dissolution occurred, hydraulic
conductivity for the system studied would have a decreased impact on environmental loading.
Leaching from sample surfaces and gross defects (cracks) would have the largest impact on
waste transport, rather than transport from within the interior by permeant flow.
The hydraulic conductivity results show large differences for the initial conditions
between the 8% and 9% matrices (8.24 x 10~5 crn-sec"1 vs 1.57 x 10~6 cnxsec"1 respectively).
The initial .conductivity for the 10% matrix (1.36 x 10"6 cm.sec"1) is not significantly different
from the 9% matrix. The differences in initial hydraulic conductivity may be due to different
mechanisms. The connected porosity of the 8% matrix may be greater than those of the 9%
and 10% matrices such that it behaves as a granular material. The 9% and. 10% matrices
behave as pastes because of their higher cement and water contents.
The change in hydraulic conductivity with respect to time, (the "B" coefficient in Table
7) shows that hydraulic conductivity decreases more rapidly as the cement content increases.
This suggests that the aqueous permeant-matrix interactions may be some form of cement
hydration reaction which is promoted by passing aqueous permeant through the sample.
The phenomenon of reduced hydraulic conductivity with time has been observed with
hardened cement pastes. Powers et al. (13) noted that the hydraulic conductivity of cement
pastes cured underwater was reduced by six orders of magnitude due to increased hydration.
Powers et al. (14) described the mechanism of this phenomenon. The volume of hydrated
-28-
-------
paste is 2.1 times greater than unhydrated paste and hydration products fill pores and cavities,
causing discontinuities, effectively reducing the number of flow channels. The number and
radii of conducting capillary pores appear to define hydraulic conductivity in cement pastes.
Nyame and Illston (16) compared their hydraulic conductivity data to that predicted by
hydraulic radius theory and suggested a correlation. Hughes (17) considered the effects of
pore characteristics (isotropy and tortuosity) in developing a model for cement-paste hydraulic
conductivity, which considered conducting channels as Poiseuille tubes.
Visual inspections of tested samples revealed inclusions of white material in the
normally dark solidified/stabilized samples. Examination of the 8% matrix by scanning
electron microscopy (SEM) revealed profuse fibrous growth (see Figure 6). Examination of
this material at enhanced magnification revealed a material morphologically similar to
ettringite (3CaO.Al2O3.CaSO4.31H2O) (see Figure 7) rather than one of the major hydration
products of normal portland cement, such as calcium silica hydrate (CSH) or calcium.
hydroxide (CH).
The morphology of the observed fibres was similar to that of the Aft fibres described
by Dalgliesh and Pratt (26). Analysis of individual fibres by SEM with an energy dispersive
x-ray spectrometer, revealed the presence of Ca, Al, S and traces of Fe (see Figure 8). This
result als.o suggests the presence of minor cement phases similar to ettringite, specifically the
aluminoferrite monosulphate (Afm) and aluminoferrite trisulphate (Aft) phases.
Individual fibres were subsequently analyzed by a scanning transmission electron
microscope coupled with an X-ray analyzer. Ca:S ratios were determined to be 2.62 ± 6.52.
The trisulphate form and monosulphate form of the aluminoferrites have Ca:S ratios of 2.1 and
4.1 respectively. Thus, the phase found in this solidified/stabilized waste matrix is composed
largely of the Aft phase.
-29-
-------
FIGURE 6. Fibrous Growth in Tested 8% Matrix (1000X)
FIGURE 7. Fibrous Growth in Tested 8% Matrix
Showing Morphological Similarities to Ettringite (6000X)
-'30-
-------
Series II- Energy Mines & Resources
Cursors 0.000keV = 0
MON 20-RUG-90 12:09
Q
...
M
0
/
;
;
0
u. ._....
;
t*
f
n||fiiii
: : :
kill 1
r
twv....
\
:
s
j
II
!
i ! ilfoiM
: ! :
: : :
___
fl
%
l|
:
* - & ••>
.
»|
Mrte
: : : : :::::;:::
M
N"
"Vilify i ,
rVIVWW
j
^h-!tf*,,.p
•A-r-y^,^ $..„
rT^'i'M^r-
^-IT.T-V'V.O..
.000 VFS = 256 10.240
120 flEC CEMENT, 15KEV, SINGLE STRRND, 300820
FIGURE 8. X-ray Analysis of Individual Fibres
The ettringite phase was found in all matrices. Figures 9 and 10 show the presence
of ettringite in the 9% matrix which formed larger structures and in some cases blocked off
pores. Patel et al. (15) noted a similar phenomenon that large pore fractions decreased while
gel (paste) porosity increased during cement paste hydration. Figures 11 and 12 show the
presence of ettringite in the 10% matrix. No evidence of ettringite was found in companion
samples of 8%, 9% and 10% matrices which were cured over a similar time period in a .
humidity chamber at 95% minimum Relative Humidity, as shown by Figures 13a to 13c.
Control samples containing only 23% Normal Portland Cement, 68% sand and 9%
were cured under water or in an environmental chamber for 28 days. Ettringite was found in
the cement samples cured under water (Figure 14) but not in the samples cured in the
-31-
-------
FIGURE 9. Ettringite Growth in Pores of the 9% Matrix'Sample (1000X)
FIGURE 10. Ettringite Growth Blocking Pores in the 9% Matrix Sample (X1000)
-32-
-------
4.
•sspiadoid onptjouotii sit SuiXonssp sjtratngo
jo spBWB ajBqdjns aqj Suunp pguuoj st 'gjdurexg ioj 'gjtSuuwg -ppgausq ApiBtnpp 3Aoid
sjonpoid uopBipA'q IJB jojst issj psjBispoDB ire ui Supssj Siiunp ssidccres sqj qSnojip
SuissBd Xq pajoinoid jqnop ou SJB SAoqB passnosip suopoBsi uop^jpXq aqj, -sjssi uuaj
lioqs uo pgssq SSJSBAV jo znotA^qaq uiiaj Suoj aqj Supoipaid ugi{A\ jgpio tit st
9SBO A"q SSBO B uo psipms aq ppioqs
ptre SQnbtuqosj lusuqBari gs ussAvjsq XIBA Xp3[T[ HJAV suotjaBisjuT SSQUJ, -suopoBisjut xujBtn
-jtreauusd amsBsra oj 'acnp JSAO uopBondsi jtigptjpis SutAjioads Xq 'sjoajjs [Bjtodoisj japtsuoo
ppoqs stiopBpiSQj: XjTAponpuoo onnBipXq jBqj sjssSSns stqx 'snip araos loj anupuoo suopossj:
jusuiao jBqj ptre sjtresuuad snosnbB qjpvv XpjBtnput JOBSJ np^ SSJSBAV snoppuaraso
A\oqs sasXretiB Xdoosoiotoi uouosp ptre /CjtAponpuoo onnBJpXq jo
jo uopBtujoj sqj sjotnojd XBUI suopoBSJ
;-XIJJBUI jBqj jss§§ns ssop it iuopBJpXq snoioSti SSQ\ oj anp XpjBd aq XBUI srqj qSnoqjp 'ptre
spojuoo juatngo aqj ut ssnjojd sssj siBsddB gjtSuujjg sqx '(SI 9-rn2tij) jgqureqo
-------
FIGURE 11. Ettringite Growth in the 10% Matrix Pores (X200)
FIGURE 12. Ettringite Growth in the 10% Matrix Pores (X1000)
-34-
-------
S3
a
'Eh
'I
s
1 8
W I
I §
I £
03
CO
I
W)
ffi
fe ^
O 00
co "3*
I
en
-------
FIGURE 14. Cement Control Sample Cured Under Water
FIGURE 15. Cement Control Sample Cured in the Humidity Chamber
-36-
-------
MODEL TESTING AND TEST PRECISION
.- _ - JL' .4 *
The results from testing three 8% matrix samples on three hydraulic conductivity
apparatus termed #925, #929 and #930, against the model derived previously for the 8%
matrix are shown in Figure 16. The y axis represents the difference between observed and
predicted hydraulic conductivity. Thus "zero" on this axis would represent perfect agreement
between predicted and observed values. Machine #925 demonstrated a nearly constant
difference (-6 x 10~6). This machine was used to generate the original 8% matrix model and
so represents variance between sample preparations. The deviations from the model of
machines 928 and 930 reflect variances due to between sample preparation and between
machine. Results from Machines 928 and 930 tend to converge to the predicted model values
with time. Thus, the temporal changes observed in this' solidified waste cause differences in
observed hydraulic conductivity between the same matrix samples to become smaller with
time.
Figure 17 illustrates the experimental results of hydraulic conductivity for the replicate
8% samples (Raw Data is shown in Appendix C). The graph shows that the range of
observed values decreases with time. Table 8 summarizes the means and standard deviations
of the log transformed data. A log transformation has been applied by Bryant and Bodocsi
(11) for clay and soil-liners and Stegemann and Cote (18) for solidified/stabilized
wastes. Bryant and Bodocsi (11) demonstrated that the log of hydraulic conductivity
measurements should stabilize variance, are more meaningful because of the large ranges of
data, cause sources of variance to become additive rather than multiplicative and simplify
relationships between hydraulic conductivity and related quantities such as void ratios.
Appendix C shows that dfter a log transformation of the raw hydraulic conductivity data the
null hypothesis of a normal distribution cannot be rejected at <0.10.
Precision is defined in Table 8 as the 95% confidence limits for the estimate of the
mean with three samples. This allows direct comparison of precision estimates for hydraulic
conductivity measurements with those obtained by Stegemann and Cote (18) who report a
-37-
-------
Table 8. Descriptive Statistics of Logi0 K for the 8% Matrix Replicates
Test Day Mean log K Standard Deviation Precision*
r
8
15
22
29
-3.924
-4.308
-4.475
-4.613
-4.782
0.184
0.248
0.233
0.256
0.244
x/-r 3.80
x/4- 4.40
x/+ 4.25
x/-r 4.48
x/+ 4.36
* 95% confidence limit of mean K for. 3 samples
precision of x/-r 7.3 among four replicates. In this study a precision of x/-f 4.36 (median
estimate) was obtained with only three samples. The estimate in this study contains both
machine and sample variance. Recall that a strict quality control criterion was instituted for
sample preparation and this is reflected by the small deviation from the model predictions for
machine # 925. The hydraulic conductivity measurements were obtained at median levels of
gradient and back pressure. It will be shown later that hydraulic conductivity is least sensitive
to gradient and back pressure at these levels.
SOLUBILITY EFFECTS DURING TESTING
During hydraulic conductivity testing solidified/stabilized wastes are subject to
dissolution by permeants. In the absence of waste/permeant hydration products, as discussed
previously, the processes of leaching will enlarge conducting pores, which theoretically should
increase sample permeability and measured hydraulic conductivity. The extent of leaching
was studied as the surrogate waste chosen did not display this behaviour.
Total dissolved solids (TDS) analysis of effluent from hydraulic conductivity samples
were routinely performed for each matrix. Results are illustrated in Figure 18.
-38-
-------
so
70
6O
GO
••s
8 2 20
t p! C
O ^ 10
'i
•> o
«• -
--B-
'« 0
1
o
I
1O
Test Day
FIGURE 16. Deviation from the Predicted 8% Model
I
20
T
ao
I.80&04-
I1
I
HM
1.60&04-
1.4000&04-
1.0000&04-
8.0000&05-
6.00(X)E-05-
4.0000&05-
2.0000&OS-
O.OOOOE+00
Machine #
0925
CI928
x930
10
20
30
FIGURE 17. Hydraulic Conductivity Replicates for 8% Matrix at Median Levels of i and P
-39-
-------
I
1
24-
20-
16-
12-
8-
4-
0 8% Matrix
D9% Matrix
x 10% Matrix
200.
400
600
800
1000
1200
Cumulative Volume (mL)
FIGURE 18. Leaching of Hydraulic Conductivity Samples During Testing
A measurable portion of the sample was leached during testing. More than 20 grams of solids
were leached from each sample which corresponds to slightly more than 1% of the sample
weight. The leaching rate appears to be proportional to the matrix which implicates cement
and water content as the controlling parameters for leaching.
Chemical analysis of a composite permeant sample from a 9% matrix scoping sample
is shown in Table 9. Analyses show that the permeant is highly basic, and that the freely
soluble sulphate, chloride and hydroxide salts of Na and K are the largest components of
dissolved solids. Calcium is relatively insoluble at high pH which explains its low value. Zn
and Cr analyses are included to demonstrate that amphoteric metals can be quite soluble at
high pH.
-40-
-------
Table 9. Composite Permeant Analyses of the 9% Matrix
Analyte
pH
Na
K
OH-
Cl-
S042-
Ca
Zn
Cr
Concentration (mg/L)*
14 (standard units)
20800
17000
12300
10100
14900
59.2
69.3
44.0
* Unless specified
Figure 19 shows the relationship between the log of hydraulic conductivity and
cumulative total dissolved solids. The graph, demonstrates that hydraulic conductivity
continues to decline even under conditions which leach alkali which has been attributed to
ettringite formation.
A comprehensive explanation for the behaviour of the solidified/stabilized waste is now
possible. In low porosity products, produced by low water-contents and no aggregate,
ettringite can cause swelling cracks. In the surrogate waste matrix chosen for this study, the
added sand increases sample porosity to such an extent that ettringite can form without
forming swelling cracks. During hydraulic conductivity testing, soluble alkali, including
sulphates, dissolve and attack the tricalcium aluminate phases forming ettringite. This material
swells and fills sample pores causing reduced hydraulic conductivity as shown in this study.
The beneficial formation of ettringite may be more specific to this waste matrix than
general in application. Waste treatment engineers normally try to minimize sample porosity
-41-
-------
•3.0-r
-35-
-4.0-
-45-
-5.0-
-55-
-6.0-
-65-
-7.0-
-75-
-8.0-
5 10 15
Cumulative IDS (g)
20
25
FIGURE 19. Log K vs Cumulative IDS at Median Levels of i and P
and water content. Swelling hydration products, such as ettringite, would cause cracking,
resulting in an increase in surface area and hydraulic conductivity.
RESPONSE SURFACE REGRESSION ANALYSES
Effect of Matrix
Hydraulic conductivity data were collected for the 8%, 9% and 10% matrix over 80
days of testing at the various trial levels of gradient and back pressure outlined in the
experimental design. The overall mean hydraulic conductivity was significantly greater
(p<0.01) for the 8% matrix (10 ± 5 x W6 cm.sec'1) compared to the 9% matrix (0.06 ±0.03
-42-
-------
x 10~6 cm-sec"1). For the waste system and measurement method chosen, hydraulic
conductivity could differentiate between the 8% and 9% matrix treatments even when the
confounding factors of time and instrument measurement parameters are retained.
The overall mean of the 10% matrix was not significantly different than the 9% matrix.
As discussed previously, the 9% and 10% matrices behave as pastes in sample preparation;
but the 8% matrix behaves as a soil-cement (granular solid).
The matrix-permeant interactions caused a decrease.in hydraulic conductivity with
time. If the differences between matrices are analyzed with non-parametric statistics (sign
test) or treated as dependant samples paired according to date and instrument conditions,
differences between all matrices would likely become significant.
Hydraulic conductivity has been shown to be very sensitive to matrix for the waste
treatment system chosen, if temporal and machine conditions are considered.
-43-
-------
Effect of Instrument Parameters
After equilibrium had been reached as indicated by the criteria outlined previously, the
variation in hydraulic conductivity with gradient and back pressure was modelled by a second
order polynomial:
K x 106 = b0 + b^j + b2x2 + bnx2! + b^i22 + b^x^
where xt and x2 are the coded (-1 to +1) design coefficients of gradient and back pressure,
respectively.
Regressions were performed for the 8% and 9% matrices. The 10% matrix reached
such a low value of K that single measurements often required more than 1 week of testing.
Eight and nine percent matrices could be tested concurrently over much shorter periods.
Because our goal was to compare matrices under similar test conditions, the 10% matrix tests
were not deemed to be comparable and thus were not included.
The results of the response surface regression analyses are summarized in Table 10
using coded levels of the test parameters.
For the 8% matrix, only gradient has a significant effect on measured hydraulic
conductivity. Both linear (bt positive) and quadratic (bu negative) terms of gradient are
significant. Back pressure at the levels chosen in this experiment did not have a significant
effect on measured hydraulic conductivity. Figure 20 shows the values of hydraulic
conductivity predicted by the response surface regression as a function of gradient and back
pressure using decoded (actual) levels. Hydraulic conductivity is maximized at the median
levels of gradient (154) then declines due to the quadratic effect and is minimized at the low
levels of gradient. Figure 21 displays a contour plot of hydraulic conductivity from the same
model. In addition to predicting maximum values for measured hydraulic conductivity at the
median levels of gradient (and back pressure), the hydraulic conductivity intervals (conditions
-44-
-------
yielding equal values of K) show that the sensitivity of K to gradient and back pressure is
minimized at median levels.
For the 9% matrix, Table 10 shows that only linear terms of gradient (positive) and
back pressure (positive) are significant. The effect of gradient is estimated to be double that
of back pressure. Figure 22 displays the values of hydraulic conductivity predicted by the
response surface regression for the 9% matrix. Maximum hydraulic conductivity is predicted
to be at the highest levels of gradient and back pressure which is different from that observed
with the 8% matrix.
Figure 23 is the 2 dimensional contour plot of the response surface regression for the
9% matrix. The K intervals suggest that K is least sensitive to gradient at high levels. This
observation must be with reservation, because although the quadratic term of gradient appears
as large as the effect of back pressure, it is statistically less significant.
-45-
-------
o
CO
§
o
J—(
^
*l
J ^
en 'O
es o.
P g
en <-5
»— i *s-
oo ^
CQ
O
o
0
•
CO
d
oo
0
»
T-I en"
S5 g.
X-S
0 g
C5 >^->
xli'
O ^
^^ C5
1 s-~'
IB
. ^-v
11
,0 0
0 Q,
ON ^
CO
13
1
^
1
CO
-8
O
Q CO
O hJ
n II
•1— «09
l-l
JH - *" p
0) O
*o ^1
t"1 0
*o W
1 I
CQ »i-j
"8 -2?
PM »5
II II
• »
3
13
i
o
'c
W)
ei—4
CO
1
II
CO
*«
vo
I
-------
7.55 •
4.67
227
154
GRADIENT
81
124
FIGURE 20. Response Surface Plot for the 8% Matrix Showing the Predicted Maxima
227
42
69
BACK PRESSURE
(kPa)
97
124
FIGURE 21. Contour Plot of Predicted Values for the 8% Matrix Showing
Hydraulic Conductivity is Maximized -at Median Levels of i and P
-47-
-------
0.042
124
0.022
227
154
GRADIENT
81
FIGURE 22. Response Surface Plot of Predicted Values for the 9% Matrix Showing the
Linear Relationship of Gradient and Back Pressure
t-
z
UJ
DC
CD
227
172
117
14
42
69
BACK PRESSURE
(kPa)
124
FIGURE 23. Contour Plot of Predicted Values for the 8% Matrix
-48-
-------
The variation in hydraulic conductivity was less than four-fold for the 9% matrix and
less than three-fold for the 8% matrix as a result of increasing gradient and back pressure from
low-levels close to field conditions to the high levels used in accelerated testing. Therefore
laboratory measurements conducted to accelerate testing are a reasonable approximation of
field conditions. The gradient was varied by a factor of twenty-five while the back pressure
was varied by a factor of eight. This relative insensitivity to pressure terms suggests that the
samples are highly liquid saturated.
Hydraulic conductivity measurement appears to be most sensitive to gradient at low
levels. This has implications for falling-head permeameters which measure hydraulic
conductivity at very low gradients where hydraulic conductivity is most sensitive. Regulatory
tests could minimize variance by specifying relatively high levels of gradient (i = 150, 220
kPa across a 15 cm sample).
The models developed for hydraulic conductivity based on the parameters of gradient
and back pressure describe 58% - 68% of .the variance in experimental results. A portion of
the variance not explained is likely due to the temporal effects noted previously. Linear
regression was used to identify equilibrium, when temporal changes in hydraulic conductivity
at constant gradient and back pressure (median levels) were not different than zero.
Equilibria were reached by 59, 34 and 27 days for 8, 9 and 10 percent matrices
respectively; the null hypothesis, that the estimated slope equalled zero, could not be rejected
for any of the three matrices with a probability of Type I error less than 0.13.
This suggests that considering only Type I error may be inappropriate for determining
termination criteria for hydraulic conductivity tests. Type II error should be considered in
determining hydraulic conductivity equilibria by considering the power of the test. Proper
sample size must be determined for an experiment to have sufficient power to detect a given
effect size from the null condition with a fixed Type I error rate.
-49-
-------
EFFECT OF SAMPLE POROSITY
Sample porosities and the equilibrium value of K (mean of trial 5 data after attaining
"equilibrium") are shown in Table 11. '
Table 11. Sample Porosity
Hydraulic Conductivity
Matrix Porosity* Pore Volume (cm3) (cm.sec"1)
8%
9%
10%
0.321
0.301
0.272
236
223
201
(1.36 ± 0.14) x lO'5'
(6.62 ± 1.44) x 10'8
(3.95 ± 0.48) x 10'9
"* determined after hydraulic conductivity testing
The samples tested demonstrate a negative correlation between matrix and porosity.
Higher cement content is associated with lower sample porosity. Lower sample porosity is
also associated with lower hydraulic conductivity.
The relationship of porosity (E) and hydraulic conductivity from Table C.I (all trial
5 data after attaining "equilibrium" rather than the means) may be modelled according to:
K = c 10bE
where c is a constant and b is the slope of the regression. The coefficients are c = 2.834 x
10'29, and b = 72.495 (r2 = 0.891) (p<0.0001).
Porosity appears to have some merit as a predictor of hydraulic conductivity.
However, bulk porosity does not define pore parameters such as hydraulic radii or whether
pores are discrete or capillary.
-50-
-------
Cement researchers have studied the relationship of porosity to hydraulic conductivity.
Nyame and Hlston (16) showed a relationship between hydraulic conductivity and pore
structure in cement pastes using hydraulic radius theory. Hydraulic conductivity is related to:
r\f(E) where rh is the hydraulic radius. This relationship suggests that porosity is not a unique
function of hydraulic conductivity. Hughes (17) developed the Poiseuille formula which
shows the importance of pore size distribution rather than total porosity. The flow rate across
a unit cross section is given by:
32 N2 ji h
where F is the flow rate (nr'.sec"1), r is pore radius (m), Ap is the pressure difference (N.rrf2),
N is the tortuosity factor, u is the viscosity (N.sec.m"2) and h is the specimen thickness (m).
This model was successful in predicting hydraulic conductivity within half an order of
magnitude when simplifying assumptions on pore connectivity were made. The dependence
of flow on r2 shows that large pores account for a disproportionate amount of flow. Hydration
reactions serve to connect particles as .shown above and as suggested by Nyame and Hlston
(16) subdivide the interstitial space resulting in smaller flow channels. This reduction in flow
channel radii may not result in a detectable change in bulk porosity.
In conclusion, porosity has some value as a predictor of hydraulic conductivity within
a specific waste/treatment as shown in Table 1 1. Within a waste/treatment type sample
preparation such as additives, compaction and water content etc. may be controlled and the
hydraulic conductivity/porosity relationship defined. For between treatment comparisons,
independent pore parameters would be required to predict hydraulic conductivity.
SATURATION CONSIDERATIONS
Sample saturation was determined by the ratio of pore water (or free water) to pore
volume. Free water was determined as the sample weight loss at 60°C.
-51-
-------
Moisture content is a destructive test and thus could only be conducted on test samples
after hydraulic conductivity testing. Companion samples prepared under standard conditions
were tested to estimate the degree of saturation before subjecting the test samples to standard
vacuum saturation. Results from these companion samples and test samples after hydraulic
conductivity testing are shown hi Table 12.
Table 12. Sample Saturation
H2O
Matrix (%)
8+
9+
10+
8*
9'
10*
4.92
5.66
- 6.14
12.94
11.71
10.55
(cm3)
79.5
94.2
. ' 105.4
226.5
206.4
189.6
rore voiumc
(cm3)
227.3
223.3
204.4
235.5
222.6
201.4
OM.LUJ.a.UUU
35.0
42.2
51.6
96.2
92.7
94.1
t Companion samples after curing
* Samples after hydraulic conductivity testing
Results show that samples are highly liquid saturated, which may explain the relative
insensitivity of hydraulic conductivity to the instrument pressure parameters discussed
previously.
In the field, solidified stabilized wastes will be in the unsaturated state. Flow in
solidified/stabilized wastes in the unsaturated state will be different from that measured for
saturated wastes in the laboratory. The nature of this unsaturated flow was beyond the scope
of this study and should be considered as an area for further research.
-52-
-------
COMPARISON OF SOLIDIFIED/STABILIZED WASTE TO SOIL/CLAY LINERS
Results of the regression analysis showed that the hydraulic conductivity of the
solidified/stabilized waste studied responded in a positive linear fashion for both 8% and 9%
matrices. A negative quadratic term was significant for the 8% matrix.
Clay and soil-liner researchers suggested that high gradients could lead.to sample
consolidation and thus lower measured hydraulic conductivity. Bryant and Bodocsi (11) noted
that the experimental results of various researchers displayed a negative relationship between
hydraulic conductivity and gradient. -
The applied gradients for clay and soil liners reported were as high as 300. This would
correspond to approximately a 450 kPa pressure drop across a 15 cm sample. A compacted
wet clay which had an unconfined compressive strength of 100 kPa would thus be subjected
to pressures which would tend to consolidate, these samples.
\
This was not the case with the solidified/stabilized waste studied. The samples studied
were molded as monolithic solids with dimensions of 7.62 cm diameter and 15.2 cm height.
Companion samples were sacrificed in unconfined compressive strength tests after curing a
minimum of forty-one days. Results are summarized in Table 13.
Table 13. Unconfined Compressive Strength Tests (kPA)
Trial
' 1
2
3
Mean
Standard Deviation
8 %
3380
3080
3830
3430
380
Matrix
9 %
3900
3000
-
3450
640
10%
5320
6340
6460
6040
630
-53-
-------
The sample matrices studied displayed an unconfined compressive strength at least ten
times greater than the highest pressure drop tested (340 kPa). This may explain the different
behaviour noted for solidified/stabilized wastes. The positive correlation of hydraulic
conductivity'with gradient for solidified/stabilized wastes may represent a response in liquid
parameters rather than a response to pore parameters, as is likely the case for clay and soil
liners.
-54-
-------
REFERENCES
1. Ather, G., Evans, J.C., Fang, H.-Y., and Witmer, K. Influence of Inorganic Permeants
upon the Permeability of Bentonite. In: Hydraulic Barriers in Soil and Rock. ASTM
STP 874. American Society for Testing and Materials, Philadelphia, 1985. pp 64-73
2. Acar, Y.B., Olivieri, L, and Field, S.D. The Effects of Organic Fluids on Compacted
Kaolinite. In: Hydraulic Barriers in Soil and Rock. ASTM STP 874. American
Society for Testing and Materials, Philadelphia, 1985. pp 203-212.
3. Fang, H.-Y. and Evans, J.C. Long-Term Permeability Tests Using Leachate on a
Compacted Clayey Liner Material. In: Ground-Water Contamination: Field Methods,.
ASTM STP963., American Society for Testing and Materials, Philadelphia, 1988. pp
397-404.
4. Elzeftawy, A. and Cartwright, K. Evaluating the Saturated and Unsaturated Hydraulic
Conductivity of Soils. In: Permeability and Groundwater Contaminant Transport.
ASTM STP 746. American Society for Testing and Materials, 1981. pp 168-181.
5. Parker, D.G., Thornton, S.I., and Cheng, C.W. Permeability of Fly-Ash Stabilized
Soils. Paper presented at 1977 Specialty Conference of the Geotechnical Engineering
Division. American Society of .Civil Engineers. Ann Arbour, Michigan. June 13-15,
1977.
6. Carpenter, G.W. and Stephenson, R.W. Permeability Testing in the Triaxial Cell.
Geotechnical Testing Journal. 9:1, March 1986. pp 3-9
-55-
-------
7. Edil, T.B. and Erickson, A.E. Procedure and Equipment Factors Affecting
Permeability Testing of a Bentonite-Sand Liner Material. In: Hydraulic Barriers in
Soil and Rock. ASTM STP 874. American Society for Testing and Materials,
Philadelphia, 1985. pp 155-170.
8. Lentz, R.W., Horst, W.D., and Uppot, J.O. The Permeability of Clay to Acidic and
Caustic Permeants. In: Hydraulic Barriers in Soil and Rock. ASTM STP 874.
American Society for Testing and Materials, Philadelphia, 1985. pp 127-139.
9. Bowders, JJ. Termination Criteria for Clay Permeability Testing Discussion. Journal
of Geotechnical Engineering. 114:8,1988. pp 947-950.
10. Pierce, JJ. and Witter, K.A. Termination Criteria for Clay Permeability Testing.
Journal of Geotechnical Engineering. 112:1,1986. pp 841-854.
11. Bryant, J. and Bodocsi, A. Precision and Reliability of Laboratory Permeability
Measurements. EPA Contract No. 68-03-3210-03 U.S. Environmental Protection
Agency, Cincinnati Ohio, 1985. 177 pp.
12. Pierce, J.J., Salfors, G. and Peterson, E. Parameter Sensitivity of Hydraulic
Conductivity Testing Procedure. Geotechnical Testing Journal. 10:4, December 1987.
pp 223-228.
13. Powers, T.C., Copeland, L.E., Hayes, J.C. and Mann, H.M. Permeability of Portland
Cement Paste. Journal of the American Concrete Institute. 26:3, November 1954. pp
285-298.
14. Powers, T.C., Copeland, L.E. and Mann, H.M. Capillary Continuity or Discontinuity
in Cement Pastes. Journal of the Portland Cement Association Research and
Development Laboratories. 1:2, May, 1959. pp 38-48.
-56-
-------
15. Patel, R.G., Killoh, D.C., Parrott, LJ. and Gutteridge, W.A. Influence of Curing and
Different Relative Humidities upon Compound Reactions and Porosity in Portland
Cement Paste. Materials and Structures. 21:123, 1988. pp 192-197.
16. Nyame, B.K. and fllston, J.M. Relationships Between Permeability and Pore Structure
of Hardened Cement Paste. Magazine of Concrete Research. 33:116,1981. pp 139-
146.
17. Hughes, D.C. Pore Structure and Permeability of Hardened Cement Paste. Magazine
of Concrete Research. 37:133, 1985. pp 227-233.
18. Stegemann, J.A. and Cote, P.L, Summary of an Investigation of Test Methods for
Solidified Waste Evaluation. Waste Management, 10:1990. pp 41-52.
19. ASTM D-558-82. Standard Test Methods for Moisture-Density Relations of Soil
Cement Mixtures. American Society for Testing and Materials. Philadelphia, 1988.
V 04.08, pp 108-111.
. 20. ASTM D2216-80. Standard Test Method for Laboratory Determination of Water
Content of Soil, Rock and Soil Aggregate Mixtures American Society for Testing and
, Materials. Philadelphia, 1988. V 04.08. pp 262-264.
21. ASTM C604-86. Standard Test Method for True Specific Gravity of Refractory
Materials by Gas Comparison Pycnometer. American Society for Testing and
Materials. Philadelphia, 1988. V 15.01. pp 159-161.
22. ASTM D1633-84. Standard Test Method for Compressive Strength of Molded Soil-
Cement Cylinders. American Society for Testing and Materials. Philadelphia, 1988.
V 04.08. pp 229-231.
-57-
-------
23. AWWA #209A. Total Residue Dried at 103 - 105°C In: Standard Methods for the
Examination of Water and Wastewater. 15th Ed. American Water Works Association.
Washington D'.C.,. 1981. pp 93-94.
24. SAS Institute Inc. SAS/STAT User's Guide. Release 6.03 Edition., Gary, NC:SAS
Institute Inc., 1988. 1028 pp.
25. Garcia-Bengochea, I. and Lovel, C.W. Correlative Measurements of Pore Site
Distribution and Permeability in Soils. In: Permeability and Ground Water Transport.
ASTM STP 746. American Society for Testing and Materials. Philadelphia, 1981.
pp 137-150.
26. Dalgleish, BJ. and Pratt, P.L. Fractographic Studies of Microstructural Development
in Hydrated Portland Cement. Journal of Materials Science. 17:8, 1982. pp 2199-
2207.
-58-
-------
APPENDIX A
CALIBRATION OF GEOTEST PERMEAMETER INTERFACES
Calibration of the Geotest interface was conducted by measuring the mass of water
expelled from the interface as a function of interface piston displacement Linear regression
was used to determine the linear displacement/mass response of the interface, used in
subsequent hydraulic conductivity determinations.
Equipment required included a: Geotest Interface (Model S5425), Mitutoyo Digimatic
Indicator (Type ID-130ME), Mitutoyo miniprocessor (DP2-DX), Mettler analytical balance
(HK 160) with RS232 Adapter (Mettler CL249) and computer (Fujikama Model #fk286MT).
The equipment layout is shown in Figure A.I.
Open containers of water with paper towel wicks were placed in the balance weighing
chamber twenty minutes before commencing the calibration to saturate the local environment
and minimize evaporation loss. Twenty data points were collected at each of 6, 12, 18 and
24 mm interface piston displacements. Eighty data points were generated by discharging ~
0.012 g of water into a 50 ml volumetric flask with a corresponding piston displacement of
~ 0.006 mm.
Data collected were merged into a single file by matching time readings from the Data
Logger and Balance files. Data in an ASCII format was imported into SAS (1988 Release
6.03) for statistical analysis. A least squares estimate for a linear model of interface piston
displacement versus mass of water was produced.
Calibrations were performed three times over a fifteen month period; an initial
calibration, prior to, and at the end of hydraulic conductivity experiments. The results of the
three calibrations for the six interfaces are shown in Table A.I.
-59-
-------
o a
UJ
a.
a
o
ODD
ODD
ODD
ODD
UJ
ID
13
d
UJ
o
u.
u
•z.
2
Ld
toa
K
LJ
I— O
Ul U.
u
o
-------
The points fit the linear model sufficiently well to generate a coefficient of
determination of 1.0000. The standard error in the slope was less than 0.011% in all cases.
The deviation from the linear model was determined for each interface calibration. The
maximum residual deviation was +0.0079 mm and -0.0077 mm. The predicted residual from
the sum of squares was a maximum of 0.0033 mm, which represents the machine error.
Table A.I. Interface Calibrations
Interface #
Slope # (mm.g )
Std. Error of Slope
761
925
926
928
929
930
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
0.500373
0.500949
0.502044
0.502238
0.502361
0.501614
0.499960
0.500320
0.501395
0.501553
0.502538
0.503015
0.501781
0.501484
0.501940
0.502952
0.504165
0.502900
0.000032
0.000015
0.000026
0.000017
0.000055
0.000029
0.000051
0.000034
0.000011
0.000008
0.000027.
0.000014
0.000022
0.000020
0.000020
0.000032
0.000013
0.000035
-61 -
-------
APPENDIX B
IN-HOUSE COMPUTER PROGRAM
The computer program accepted data from the Miniprocessor and stored it in a file on
the hard drive. The program included several options for calibration, hydraulic conductivity
calculation and file management The file name was coded to the Geotest interface serial #
and the date of data acquisition. An example is shown below:
761_1D90.019
This file indicates the: Mitutoyo Indicator serial number, file status (original or edited), file
number of day, file type (balance file, data file, merged balance and data calibration file, or
hydraulic conductivity calibration file), year and Julian day.
A header on the data file recorded the experimental conditions- (trial #, gradient, back
pressure and matrix) and the sample/test series numbers.
A test pressure file was coded with the sample/test series number and contains the test
values for the head, back and cell pressures. ;
The sample file was coded to the sample number and contained the sample dimensions,
weight, moisture content and sample preparation compositions.
The hydraulic conductivity calculation incorporated data from the sample, pressure and
data files. The hydraulic conductivity could be calculated for the entire data file or at a
specified time interval between any two data points. The hydraulic conductivity calculation
is shown below:
-62-
-------
K= ''
S • p (HP - BP) • PH • TC (OD • 2.54)2
D = Interface displacement (mm)
S = Slope of calibration curve from regression analysis (mm/g)
p = Water density at 24°C (0.9973 g/cm3)
L = Mean sample length (cm)
HP = Head pressure (psi)
BP = Back pressure (psi)
PH = Pressure to head conversion (70.47 cm water/psi)
OD = Mean outside diameter (in.)
t = Time of displacement reading (sec.)
-63-
-------
O
1
O CO VO t-;
LlM!]]«aa"«»«3'«3*8««»""
oo *o o| oo o\ I"* v*i O *~* l^ ^ ^** t*1" c*l ""fl* ^O ^S ^f t^ *^ ^ vi c*i ^^ t"* -•
g
LMM«,.MM«M»...M.«...B.M«I
>< °
^ 3
3
. cs
-------
w
2 Ss o 5!
^" ^t" ^O C^
CO CS CO OO VJ
:g
5? 8
-! o o o' o o
S 5
s i
s
o'
o o o
OO CS CO
58 fi?? £ X
s °
O\ ON
S
w
ss
w
CD
oq oo oj in p"; ^oOT^Tfi-noocvcS-* — -rr-Hp^oo vo in p- p-
Scsov^ — rfincovo^odi-Ivop^ — coo' — ov — ino'ocr3^Ttr":vP,t"tGNic^'"!e''31O,c"^oo'covoo'
p** ON —"' ^J* ^i1 p^ ^3 co vo ov co co co in \o P* oo oo cs co in in *~' <~^ in vo p*» p^ p* oo cs ^ vo co o *~< i—» in
1-1-" — cscscscScocococo^TC^Ti-^Tj-^TrinininintinininiriinSp-ooooooovi-ii-irti-!
ON O -H OO
cs' cs p~ -j
i = S = 5
=> =>
Ov -^J-
in co
O
3
ON oo>-«vovi»ocscxlenin r-
cNTr»-«>n»ocneSM3c^tnTrcNON»nt-i
T-H ^^ ff} \r\ ON oo O vi ^j* Q oo CN ^* ^* ^**
'^•ONONOt^'CNCM<
- -^ - ON tr ^
»-; *-; TJ; r^ •-* O en
*i *i ^T *i ^T* *7" *"^
a,O*tvO*^v^cn»-*OCNCNOOt*XOQCNONON»-t
^OcovOt^^^^vi^oooNvNcNooencnTj-,
v^^^cot^o^npcNTf^cnTftnONt^e^encn^q
t^f^r^C^v^r^c^opodooopopcoodopfj-ooooodcd
9999fcC^QOOOOOoococooooooococooo
oooooooooooooooo
pr] pg fr^ [f] [T] fr] fr1) fr] fr] [j1] fr'j [T^ pg fllj fT] fT^
m co — ,-• oor-
SSSSSS888S 88 8 8 § S 8 S 8
r^i TT'I r1"! Hi p"! rT'i r^i r^i Hi p~i Hi r^i rT'i r^i P*T rv'i Hi r*"1! Hi
P*P* — — odc^ovincoco^coc1^ — *-*^l:^l:Ti:r4
«n
I
0.1 -H T o -< o
—i ooo—" — T'ooooooT'T — oooT — —< — T'oooooooo-nooT'o — — —
t"
Q
OvC7\OVOvC7VC7vOvOvC7vO\C\Ovcy*C\cyvOvC\OVOvO\ Cv Ov Ov
oooooooooooooooo
-29
•?
>•
g
H
><
Q
•§1
-------
Table C.2. 8% Matrix Replicate Data
Test Day K930 (cm-sec'1) K928 (cm-sec'1) K925 (cm.sec'1) LOG K930
LOG K928
LOG K92S
1
5
8
12
15
19
22
25
26
27
28
29
Mean
Std.
Deviation
Normality
Test Value*
1.37E-04
1.02E-04
8.22E-05
6.63E-05
5.64E-05
4.75E-05
4.15E-05
3.68E-05
3.49E-05
3.24E-05
3.08E-05
2.96E-05
5.81 x 10'5
3.36 x 10"s .
2.03
1.67E-04
8.20E-05
5.46E-05 '
4.19E-05
3.45E-05
2.78E-05
2.38E-05
2.02E-05
1.92E-05
1.74E-05
1.66E-05
1.59E-05
4.34 x 10'5
4.36 x 10'5
9.02
7.40E-05
3.20E-05
2.66E-Q5
2.29E-05
1.93E-05
1.67E-05
1.47E-05
1.28E-05
1.19E-05
1.07E-05
9.96E-06
9.61E-06
- 2.18 x 10'5
1.79 x 10'5
6.40
-3.863
-3.991
-4.085
-4.178
-4.249
-4.323
-4.382
-4.434
-4.457
-4.489
-4.511
-4.529
-4.29
0.22
1.08
-3.777
-4.086
-4.263
-4.378
-4.462
-4.556
-4.623
-4.695
-4.717
•-4.759
-4.780
-4.799 .
• -4.49
0.32 .
1.30
-4.131
-4.495
-4.575
-4.640
-4.714
-4.777
-4.833
-4.893
-4.924
-4.971
-5.002
-5.017
-4.75
0.26
1.28
' Critical Values 1.36 (a = 0.10), 1.72 (a = 0.05)
-66-
-------
DC/ps
ET-CW
2440-CW
91.06.21
-67-
-------
------- |