Evaluation of calibration procedures for the measurement of total soil suction using filter paper

Material Information

Evaluation of calibration procedures for the measurement of total soil suction using filter paper
McFarland, Tara Catherine Schenk
Publication Date:
Physical Description:
83 leaves : illustrations ; 28 cm

Thesis/Dissertation Information

Master's ( Master of Science)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Civil Engineering, CU Denver
Degree Disciplines:
Civil Engineering
Committee Chair:
Chang, Nien-Yin
Committee Members:
Stalnaker, Judy
Brady, Brian


Subjects / Keywords:
Soil matric potential ( lcsh )
Soil capillarity ( lcsh )
Filters and filtration ( lcsh )
Filters and filtration ( fast )
Soil capillarity ( fast )
Soil matric potential ( fast )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references (leaves 81-83).
General Note:
Department of Civil Engineering
Statement of Responsibility:
by Tara Catherine Schenk McFarland.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
55535860 ( OCLC )
LD1190.E53 2003m M42 ( lcc )

Full Text
Tara Catherine Schenk McFarland
B.Sc., Colorado School of Mines, 1997
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Science
Civil Engineering

This thesis for the Master of Science
degree by
Tara Catherine Schenk McFarland
has been approved

McFarland, Tara Catherine Schenk (M.Sc. Civil Engineering)
Evaluation of Calibration Procedures for the Measurement of Total Soil Suction
Using Filter Paper
Thesis directed by Professor Nien-Yin Chang
Soil suction is the interaction between soil particles and water. The higher
the soil suction value is, the less the pore water in the soil is able to escape. There
are two types of soil suction. Matric suction is the capillary tension in the pore
water. Osmotic suction is the difference in pore water potential caused by the
difference in salt concentration. Total soil suction is the combination of both
osmotic and matric. Suction is measured many different ways ranging from insitu
field devices to special laboratory devices. A simple way to measure the property
is by using the filter paper method. This method can measure total or matric soil
suction. For this research, the calibration curves for the total soil suction test were
studied. There are several different methods researchers developed to acquire
calibration curves for the filter paper test. However, often the curves are different
for each researcher. The purpose of this research is to study and test different
procedures for calibration. Six different procedures were tested using both
sodium chloride and magnesium chloride solutions. These procedures all had

variations such as: equilibration time, if the filter paper was oven dried before
testing, or whether the filter paper was weighed immediately after being dried
back. The results showed that although the test was run six different ways with
two different solutions, the curves obtained were all statistically good. One
calibration curve stood out slightly better than the rest of the curves when it was
run with both solutions.
This abstract accurately represents the content of the candidates thesis. I
recommend its publication

I dedicate this thesis to my husband Scott who has supported my efforts. I
also dedicate this to my parents who inspired me to continue my education.

My thanks to my advisor, N.Y. Chang for his help and guidance and Judy
Schenk for helping me with some of the data analysis. I also would like to
acknowledge A.G. Wassenaar, Inc. which funded part of my education.

Figures................................................................ xi
Tables................................................................. xiii
1. Introduction.................................................... 1
1.1 Problem Statement............................................... 1
1.2 Objective....................................................... 2
1.3 Approach........................................................ 2
2. Literature Review............................................... 4
2.1 Introduction.................................................... 4
2.2 Soil Suction.................................................... 4
2.3 Effects of Soil Properties on Suction........................... 7
2.4 The Active Zone and Soil Suction................................ 8
2.5 Suction-Swelling Relationships.................................. 9
2.6 Soil Suction and Heave Prediction Research Studies.............. 10
2.7 ASTM D5298-94 Method of Soil Suction Measurement................ 14
2.8 Comparison of Filter Paper Measurement Research................. 15
2.8.1 McQueen and Miller, 1968........................................ 15
2.8.2 Al-Khafafand Hanks, 1974........................................ 15

2.8.3 Houston, Houston and Wagner, 1994,
2.8.4 Swarbrick, 1995..................................................... 18
2.8.5 Snethen and Johnson, 1980........................................... 20
2.8.6 Bulut, Lytton and Wray, 2001........................................ 21
2.8.7 Leong, He and Rahardjo, 2002........................................ 21
2.9 Filter Materials.................................................... 22
3. Measurement Techniques.............................................. 25
3.1 Introduction........................................................ 25
3.2 Pressure Plate Calibration.......................................... 25
3.3 Thermocouple Psychrometer........................................... 27
3.4 Pressure Membrane or Pressure Plate................................. 28
3.5 Electrical Resistance Sensor........................................ 28
3.6 Electronic Humidity Sensors......................................... 29
3.7 Tensiometer......................................................... 29
3.8 Heat Dissipation Sensor............................................. 30
3.9 Thermal Matric Potential............................................ 31
3.10 Filter Paper Method................................................. 31
4. Laboratory Facility for Suction Measurement......................... 34
4.1 Standard Laboratory Requirements.................................... 34
4.2 Laboratory Facilities for This Research............................. 35

4.3 Suction Measurement Program.................................... 37
4.3.1 ASTM D5298-94.................................................. 37
4.3.2 7-Day Test Base on Houston, Houston and Wagner, 1994........... 38
4.3.3 5-Day Test Based on Swarbrick, 1995............................ 38
4.3.4 Wait before Weigh Test based on Snethen and Johnson, 1980...... 38
4.3.5 Wait to Weigh with Two Papers Based on Snethen and 39
Johnson, 1980.........................................................
4.3.6 14-Day Test Based on Bulut, Lytton and Wray, 2001.............. 39
4.3.7 4-Day Test based on Swarbrick, 1995.............................. 39
4.3.8 2-Day Test Based on Swarbrick, 1995.............................. 39
4.3.9 Testing Matrix................................................... 41
5. Data Analysis and Discussion..................................... 41
5.1 Sodium Chloride Tests............................................ 41
5.1.1 ASTM D5298-94.................................................... 46
5.1.2 7-Day Test Based on Houston et al, 1994.......................... 49
5.1.3 5-Day Test Based on Swarbrick, 1995.............................. 49
5.1.4 Wait to Weigh Based on Snethen and Johnson, 1980................. 49
5.1.5 14-Day Test Based on Bulut etal, 2001............................ 53
5.1.6 4-Day Test Based on Swarbrick, 1995.............................. 57
5.1.7 2-Day Test Based on Swarbrick, 1995.............................. 57

5.2 Magnesium Chloride Tests....................................... 57
5.3 Comparison of the Equations of Linear Regression Analysis for both 59
Sodium Chloride and Magnesium Chloride Solutions...................
6. Conclusions.................................................... 67
A. Sample calculations.............................................. 71
B. All data from sodium chloride tests.............................. 72
C. All data and graphs from magnesium chloride tests.............. 74
References........................................................... 81
' x

2.1 - Calibration Curve Obtained by McQueen and Miller, 1968.......... 17
2.2 - Calibration Curves from Four Different Testing Methods (Snethen and 19
Johnson, 1980)........................................................
3.1 - Calibration Using a Pressure Plate (Lee and Wray, 1995)........... 26
3.2 - Thermocouple Psychrometer (Krahn and Fredlund, 1972)............. 27
3.3 - Quickdraw Tensiometer (Nelson and Miller, 1992).................. 30
5.1 - Graph of All Tests run with NaCl Solution......................... 42
5.2 - Linear Trend of All the Data for Tests Run with NaCl Solution..... 44
5.3 - Comparison of the Top and Bottom Filter Papers at 1.99 log kPa 45
5.4 - Average, Standard Deviation and Coefficient of Variation of All the Data 47
for NaCl Solutions....................................................
5.5 - Average, Standard Deviation, Coefficient of Variation for Each 48
5.6 - ASTM-1 Test...................................................... 50
5.7 - ASTM-2 Test...................................................... 51
5.8 - 7-Day Test....................................................... 52
5.9 - 5-Day Test....................................................... 54
5.10-5-Day Test (2)................................................... 55

5.11 - Wait to Weigh Test.............................................. 56
5.12- Wait to Weigh Test (2)........................................... 60
5.13 14-Day Test..................................................... 61
5.14- 4-Day Test...................................................... 62
5.15- 2-Day T est..................................................... 63
5.16- MgCh All T ests................................................. 64
5.17 Linear Trend for All Tests Run with MgC^ solutions.............. 65

2.1- Calculated Results Using the Four Heave Prediction Methods........ 13
2.2 - Calibration Data, Several Days (McQueen and Miller, 1968)........ 16
3.1- Conversion of Suction Units....................................... 25
3.2 Suction Error vs. Balance Accuracy (Swarbrick, 1995).............. 32
4.1 Suction Error vs. Balance Accuracy (Swarbrick, 1995).............. 35
4.2 - Tests run for This Research...................................... 40
5.1 - Average, Standard Deviation and Coefficient of Variation for Each 43
Suction Level...........................................................
5.2 - Standard Deviation, Average, and Coeffcient of Variation for Each 46
5.3 - Standard Deviation, Average and Coeffcient of Variation for Each 58
Suction Level...........................................................
5.4 - Hypothetical Suction Value Calculated from Equations Determined 66
from Each Procedure's Data..............................................

1.1 Problem Statement
Soil suction is an energy quantity used to evaluate the capability of a soil
to retain the soil water. (Lee and Wray, 1995) The filter paper method is a
relatively simple and inexpensive way to measure suction. This test can be used
to determine how expansive a soil is. In the state of Colorado, expansive soils are
a problem, especially for lightly loaded structures. Typically, a Denver Swell test,
Atterberg Limits tests and Gradation tests are run to determine the expansive
properties of soil. The filter paper test for total soil suction, another tool to
determine the expansion potential is not widely used in the Denver Metro Area.
The reason why this test is not used by practicing geotechnical engineers is
because each lab has different calibration curves and suction results.
Presently the ASTM D5298-94 Standard Test Method for Measurement of
Soil Potential (Suction) Using Filter Paper is somewhat difficult to run because of
the requirements on when to weigh the filter paper, especially after drying. The
test also has to run for seven days, where a swell test usually only runs for 24
hours. Several researchers have calibration procedures varying from the ASTM
D5298-94 standard. Their procedures as well as the ASTM D5298-94 procedure
will be tested to find the easiest and most repeatable filter paper calibration to

In the literature search performed for this research, it was found that while many
researchers come up with their own procedures, they typically do not compare
their procedure with any other procedures. If the procedures are compared it is
done using old data instead of producing new data. The purpose of this research
is to physically compare these calibrations by running each of them in the same
lab. The resulting calibration curves will then be evaluated using linear
regression analysis and compared to each of the other curves produced by the
testing. The test that exhibits the best linear fit and highest coefficient of
determination (R2) will be repeated.
1.2 Objective
The objective is to develop a new calibration procedure by comparing
several researchers procedures. In addition, this lab will test the equilibration
time to see if the test can be shortened. If this lab can produce reasonably
accurate calibration curves with repeatability then perhaps the filter paper suction
test will be used more often in conjunction with swell tests to test the
expansiveness of soils.
1.3 Approach
The laboratory testing will be set up as follows. The insulated box in
which tests will equilibrate will not have more than a three degree Celsius
fluctuation in temperature. Ten solutions will be made at five different suction
levels. Five solutions will be sodium chloride and five solutions will be

magnesium chloride. Ten tests will be run at each suction level of the sodium
chloride solutions. A series of tests will be run for each of the following testing
procedures developed by these researchers:
1. ASTM D5298-94
2. 7-day based on Houston, Houston and Wagner, 1994
3. 5-day based on Swarbrick, 1995
4. Wait to Weigh based on Snethen and Johnson, 1980
5. 14-day based on Bulut, Lytton and Wray, 2001
6. 4-day test based on Swarbrick, 1995 (sodium chloride solution only)
7. 2-day test based on Swarbrick, 1995
In addition, five tests will be run at each suction level of the magnesium chloride
solutions. These series of tests will follow the above listed testing procedures.
The testing will take approximately two months.
Each calibration curve evaluated will be compared and the equation of the
line that best fits each curve determined.

Literature Review
2.1 Introduction
Soil suction theory was developed in the early 1900s for agricultural
studies of the soil-water relation to plant systems. In 1906, Livingston presented
a study on suction measurement in the desert. (Mahler and deOliveira, 1997)
Suction was not further pursued until 1937 when Gardener proposed a method to
use filter paper to determine soil suction in a given sample. The test was based on
the hypothesis that in equilibrium state the water potential of a certain quantity of
soil and the water potential in the filter paper in contact with or near the soil are
the same.
Today, soil suction is a tool used by geotechnical engineering laboratories
and there are several ways to measure the property. The filter paper and
thermocouple psychrometer methods are two of the most widely used and studied
methods. The filter paper method is an easy test to run but sensitive to
temperature fluctuations and requires an accurate balance to 0.000lg. Also, there
are some issues with calibration which make it difficult to compare suction values
from different laboratories.
2.2 Soil Suction
Soil suction is the interaction between soil particles and water. It is an
energy quantity to evaluate the capability of a soil to retain the soil water. (Lee
and Wray, 1995) Although suction is an energy term, the units used to describe

it are pressure units kilopascals (kPa), atmospheres (atm), bars, or pF (the log of a
height in centimeters of water column needed to provide the suction in question).
The effects of the water in soil can be described both mechanically and
thermodynamically. Mechanically, the negative pore water pressure can be
measured in samples using a special consolidometer and pressure membranes.
The membranes used have pore sizes small enough to prevent cavitation. Air
pressure is applied to increase the positive pore water pressure without losing air
or control of the volume of water through cavitation. The water leaves the soil if
the pore water pressure is positive and is sucked into the soil if the pore water
pressure is negative. If the water in the soil stays the same then the suction of the
soil has been found. Thermodynamically the free energy (Af) needed to move
the free pore water into soil pores containing soil water is described by Equation
Equation 2.1
where R = the ideal gas constant
T = absolute temperature
p = vapor pressure of the soil
= vapor pressure of the free water. Relative humidity is equal to
(Snethen 1981)
Total soil suction is equal to the sum of matric and osmotic suction.
Johnson (1973) described the total soil suction by Equation 2.2.

r = -RT/V,lnp/p,
Equation 2.2
where r = total suction in atmospheres
R= ideal gas constant
T = absolute temperature
Vw = volume of one mole of water
p/p0 = relative humidity in percent
The matric suction is the capillary tension in the pore water. The surface
attractive forces for water and cations and surface tension effects of water in the
soil are all components of matric suction. Matric suction is described by Equation
2.3. This type of suction is water content and surcharge dependent. (Nelson and
Miller, 1992)
where tm = matric suction in atmospheres
ua = pore air pressure
uw = pore water pressure
Osmotic suction is the difference in pore water potential caused by the
difference in salt concentration. The soluble salts in the soil water cause free
water to be attracted to the soil water causing a pressure differential as the forces
draw water into the salt solution. Osmotic suction is described by Equation 2.4.
r = (u u )
w m \a- vv /
Equation 2.3
& = PsgK =^ + [CJ
Equation 2.4
where Q = osmotic pressure
ps = solute mass density
g = gravitational acceleration
ho = osmotic pressure head

R = universal gas constant
T = absolute temperature
[Cs] = molar concentrate of the solute
This type of suction is independent of water content and surcharge
pressures. Krahn and Fredlund found that significant changes in osmotic suction
do not occur in most practical applications. (Nelson and Miller 1992)
2.3 Effects of Soil Properties on Suction
One of the soil properties that affect suction is pore size distribution. The
suction value depends on pore size and water content. The water content depends
on the type of clay minerals present. A hysteresis is observed between the
wetting and drying curves. A possible cause of the phenomenon is attributed to
changes in the contact angle of air-water interface between the drying and wetting
phases. The permeability in unsaturated soil decreases with increasing suction.
(Johnson 1973)
Another soil property that affects suction is density. In clean sand, gravel,
and limestone, which can be considered incompressible soils, large matrix
changes accompany small water content changes because the fluctuations in water
content do not cause volume changes. (Johnson 1973) In general, a low-density
material drains large pores at a low suction while a high-density material drains
small pores with a higher suction. Clays exhibit different behaviors than the
sands and gravels. Because clays are compressible and remain saturated over a

large part of the drying period they yield increased suctions compared to
incompressible soils. Changes in water content changes the volume and pore
geometry of clay so that when the soil dries the matrix suction increases and void
ratio decreases. The increased suction can introduce fissures to the soil. (Johnson
1973) Expansive clays, such as montmorillonite, with fissured open-filled joints
undergo a volume change in which the suction is driving force.
However, Krahn and Fredlund in 1972 studied the relationship between
dry density and suction. Their results showed that the matric and total suctions
are not necessarily affected by the dry density. The same soil was tested at
different densities and even loose soil was tested with a thermocouple
psychrometer. Those results showed agreement with compacted soil.
2.4 The Active Zone and Soil Suction
The active zone is the depth to which moisture will migrate after
construction of a structure potentially causing expansion and heave. The active
zone of a specific area is influenced by the thickness of the layer of soil in which
the moisture determination exists, and the transient influence by the soil type,
structure, topography, and climate. (Snethen 1973) Suction values can be used to
estimate the depth of the active zone by trying to detect where the soil suction
becomes constant on soil suction versus depth graph.

2.5 Suction-Swelling Relationships
Suction and swelling were studied together by Cocka and Birand (2000).
Commercially processed kaolinite and bentonite clay minerals were mixed in
different proportions and subjected to matric suction and free-swell testing. They
compared the oedometer method (ASTM D4546-90) of heave prediction with soil
suction heave prediction developed by Johnson and Snethen. (Equation 2.5)
AH/H = (C

where: AH/H = calculated swell
Tj = initial soil suction
Tf = final soil suction
av=vertical stress
eo=initial void ratio
C,p= suction index
The free-swell tests were run using an oedometer (ASTM D4546-90) on the
mixtures at 10 percent water content and 1.64Mg/m3 dry density. The samples
were confined in special consolidation rings due to the highly expansive bentonite
clay and set up under a small surcharge of 1.94 kPa. The test results were
compared to the log soil suction and water content. The swell percentages
predicted by Johnson and Snethens equation were compared with the direct
measurements from the free swell test. The results showed that the suction
method underestimated the swell percentage. Although the suction method
underestimated actual swell, there was a strong relationship between the two. As
the swell percent increased, the soil suction also increased.

2.6 Soil Suction and Heave Prediction Research Studies
Snethen and Huang, 1992 compared four different methods of heave
prediction using soil suction. The methods compared were Snethen and Johnson,
Mitchell, Nelson and Hamburg, and McKeen.
Snethen and Johnsons method is based on the concept that as soil suction
decreases the soil volume increases. The equation developed from this concept is
shown in Equation 2.6.
AH/H = [Cr /(I + e0 )][logA0 log^ + aaf)] Equation 2.6
where Cr = suction index,
ho= A-Bw0> matric soil suction without surcharge pressure, kPa
a = compressibility index, slope of the specific volume versus the
moisture content curve
hf= final matric soil suction
kPa, Of = final applied pressure
kPa and e0 = initial void ratio
A = log soil suction value at zero moisture content
B = slope of the soil suction versus moisture content curve
The procedure is to measure the initial matric suction with a
thermocouple psychrometer or filter paper, and find A, B, and a from the plotted
results of the soil suction test procedure. The initial soil suction is measured
during testing and an assumed final profile is chosen from one of four listed
1) Zero throughout the depth of the active zone
2) Linearly increase with depth through the active zone

3) Saturated water content profile
4) Constant at some equilibrium volume
Snethen and Huang chose to use the assumption numbers two and three to
evaluate all of the heave methods.
Mitchells method assumes the vertical strain of expansive soil is linearly
proportional to suction. The heave is then calculated from Equation 2.7
where Ipt = instability index determined by a core shrinkage test
Au = change in suction, and
Hi = thickness of the ith layer
Nelson and Hamburgs method operates on the principle that since suction
is dependent on moisture content and heave is related to soil suction, heave can be
predicted by measuring the changes in moisture content. Their equation is shown
by Equation 2.8.
where Hj = thickness of the ith layer
Cw = suction index ratio
Aw = moisture content change
Actual heave can be adjusted depending on the confinement situation of the soil.
The correction factor (f) which ranges from 0.33 to 1.0 is multiplied by the
calculated heave.
Equation 2.7
Equation 2.8

McKeens method is based on the principle that total heave is related to
suction change by a parameter called the suction compression index. (Equation
2.9) The suction compression index is found with COLE or CLOD tests.
Xh x log(hf /h0) Equation 2.9
where Xh = suction compression index
hf = final soil suction
h0 = initial soil suction.
Snethen and Huang sampled 1-35 in Oklahoma which had been affected
by heaving soils for over twenty years. The active zone was estimated at 7.75
feet. They ran suction tests following ASTM (1990). They also determined bulk
density, moisture content and graphed the suction versus moisture content curve.
The actual heave measured in the field was 7.1 inches. It was determined that the
assumption that the soil suction linearly increases with depth through the active
zone is more conservative than the assumption that the soil suction profile is
based on the saturated moisture content profile and that the second assumption
was more realistic. Therefore the second assumption was used to compare
methods. The results from the four methods are shown in Table 2.1

Method Heave Prediction
Snethen and Johnson 6.2 inches
Mitchell 7.1 inches
Nelson and Hamburg 13.2 inches
McKeen 2.2 inches
Table 2.1 Calculated results using the four heave prediction methods
Snethen and Johnson and Mitchells methods were the closest to the actual
measured heave. Nelson ahd Hamburgs method overestimated the heave by
approximately 85 percent and McKeen underestimated the heave by
approximately 70 percent.
McKeen discussed the results in the 7th International Conference on
Expansive Soils proceedings. He pointed out that the paper put together by
Snethen and Huang differed from Huangs doctoral dissertation. In the
dissertation, Huang stated that McKeens method of heave prediction produced a
heave within an error of six percent and was considered to be a viable method.
McKeen also stated that the total field heave used was not correct because of the
amount of remediation the pavement had undergone plus the heave was measured
at the pavement surface which could be different from the heave of the soils due
to traffic and pavement structural stiffness. Another problem with Snethen and
Huangs research was the initial and final suction profiles. The initial profile had
been determined dining the site investigation, which could be different than the

initial profile at the time of construction of the highway. Also, the final soil
profile was determined by assumptions instead of sampling and testing.
2.7 ASTM D5298-94 Method of Soil Suction Measurement
To run the test on a soil sample, or to calibrate the filter paper according to
ASTM D5298-94, filter papers are dried in an oven for 16 hours and stored in a
desiccant jar. The suggested filter paper types are Whatman #42, Schleicher and
Schuell #589 White Ribbon and Fisherbrand 9-790A. For total suction
measurement, the soil sample is placed in a metal container with a ring to separate
the sample from the filter paper. Two pieces of filter paper are placed on the ring
and the container is sealed and placed in an insulated chest where temperature
variations are less than three degrees Celsius and left for seven days. After seven
days, the filter papers are taken out of the container and each is weighed before
being placed in an oven. After the papers have been in the oven for
approximately two hours, they are weighed again. The water content of the filter
paper is calculated and converted to suction by Equation 2.7 Water content
calculations are found in Appendix A.
h = mwf + b Equation 2.10
where h = suction
m = slope of the filter paper calibration curve
b=intercept of the filter paper calibration curve
Wf=water content of the filter paper

2.8 Comparison of Filter Paper Measurement Research
2.8.1 McQueen and Miller, 1968
McQueen and Miller developed a calibration curve in 1968. They used
Schleicher and Schuell #589 White Ribbon (S&S #589 WR) filter paper treated
with pentachlorophenol. The paper is air dried. The calibration curve used was
determined by three different methods. For stresses greater than 15 bars, the filter
paper method of suspending the paper above saturated salt solutions, measuring
total soil suction. Table 2.1 shows that they determined seven days are needed for
equilibrium. The second method used for stress between one and 15 bars was the
pressure membrane. The pressure membrane measures matric suction. The third
method used was field samples obtained at known height above the water table.
This method is less precise because the soil suction in the field at a specific point
above the water table is difficult to determine. The calibration curve obtained is
illustrated in Figure 2.1.
McQueen and Miller also checked the equilibration time needed for the
filter paper test. Table 2.2 shows the data obtained from these tests. The
conclusion was that the test has to be run for seven days.
2.8.2 Al-Khafaf and Hanks, 1974
These researchers tested the filter paper method to evaluate the influence
of the type of contact, temperature and temperature variations. They used S&S

Calibration data for stress levels controlled by saturated salt solutions in a constant temperature chamber
Salt Relative vapor pressure* Log of stress in cm Stress in bars NatSOi 0.93 5.00 97.9 CaSO 0.98 4.44 27.25
Moisture Content B A B A B
Time: 0 days 22.79 3.50 21.95 3.50 22.04 3.61
7 days 10.64 8.01 17.84 16.36 23.70 22.71
14 days 10.54 7.90 17.90 16.35 24.05 23.10
28 days 10.42 7.96 17.93 16.39 24.12 23.93
76 days 10.30 8.04 17.79 16.58 25.21 26.10
* Published values do not agree. Data given are from report by O'Brien (7). Corresponding values
computed from International Critical Tables (Washburn [14]) are: 0.798, 0.94, and 0.98. The range of
disagreement is small.
f "A papers were premoistened by exposing over distilled water for two days prior to start of test.
"B" papers were started in air-dry condition.
Table 2.2 Calibration data, several days. (McQueen and Miller, 1968)__________________
#589 WR paper treated in pentachlorophenol. Tests run were no contact, contact
on top of the soil and buried in the soil. Samples were placed in an insulated box
in a constant temperature room and allowed to equilibrate. One conclusion
reached by Al-Khafaf and Hanks was that equilibration occurred in two days.
They also concluded that temperature variations of 2Celsius caused
condensation within the sample containers affecting results of the test. Total
suction was calibrated using potassium chloride solutions. The range measured
was between 10 and 15 bars. The results matched the curve created by McQueen
and Miller for that range.
2.8.3 Houston, Houston and Wagner, 1994
Three different salt solutions and one pure water solution (suction=0 pF)
were used. The salt solutions were MgN03 (pF = 5.96), NH4C1 (pF = 5.51), and
CaS04 (pF = 4.49). The solutions were put into a glass container tipped 10-20 to

Log of Stress in Centimeters
Log 3*6.24617-0.0723M
Gordner's Curvo
Vacuum Desiccator
Saturated Salt Solutions
Pressure Membrane
'Pressure Plate
Height Above Water Table
Data Points
Calibration Line
Log S 2.8946 -0.0I025M
1 1
CO 100 150
Moisture Content of Filter Paper (Percent of Ory Weight)
Figure 2.1 Calibration curve obtained by McQueen and Miller, 1968

prevent condensation, and the lids of the containers were greased with mineral oil
to further prevent condensation. The temperature variation in the box was
0.1Celsius. The paper was suspended from the lid. The rest of the test
followed ASTM D5498-94. Two papers were used, equilibration time was seven
days and the papers were weighed immediately before and after drying.
Other research in this paper focused on the comparison of calibration
curves other researchers had produced. In particular, Houston, Houston and
Wagner focused on the McQueen and Miller (1968) curve because it is often
referred. The researchers focused on the fact that the curve was developed using
three different methods, two of which are matric suction and one is total suction.
They pointed out two potential problems when using this curve. First is that the
curve was developed to be used with the matric suction filter paper technique
where the paper is simply laid on top instead of buried in the soil. This technique
may not be the best technique because placing the paper on top of the soil does
not achieve the same contact as burying it in the soil. Other problem with the
curve is that the mixture of total and matric suction in the calibration may cause
too much scatter in the data.
2.8.4 Swarbrick, 1995
Another issue in the filter paper test is that there is little data indicating the
required time for filter paper to reach equilibration, and the suggested time is two
to seven days. Swarbrick ran three tests on three different soil samples of

Figure 2.2 Calibration Curves from four different testing methods (Snethen and
Johnson, 1980)

suctions 6, 3,6, and 3,1 pF (dry, intermediate, wet) and found that most samples
reach equilibration in four days, but that wet samples can take up to seven days.
Swarbrick's procedure for measuring total suction with filter paper is to
run the test for five days.
2.8.5 Snethen and Johnson, 1980
Snethen and Johnson, 1980, studied four different procedures for calibrating the
filter paper: the McQueen and Miller (1968), Miller (1978), U.S. Army
Waterway Experiment Station I (WES I) and WES II. The difference in the
procedures was the amount of time between removing filter paper from the oven
and weighing it for a final dry weight. McQueen and Miller weighed the paper
within five seconds, WES I requires the paper be covered after removal from the
oven and weighed within fifteen minutes. The WES II method did not require the
paper be covered and only required that the paper be weighed within 15 minutes
to four hours. Miller (1978) used a combination of the 1968 procedure and data
from Al-Khafaf and Hanks (1974). These curves are shown in Figure 2.2. All the
curves are slightly different. The conclusion by Snethen and Johnson was that if a
lab uses the same standardized procedure all the time, the suction results for that
particular lab are reliable, but that results from other labs can't be compared
against each other since the calibration curves are most likely different.

2.8.6 Bulut, Lytton and Wray 2001
Bulut, Lytton and Wray evaluated the difference between the wetting and
drying calibration curve in both the total and matric suction techniques. The
wetting curve was obtained using sodium chloride solutions and S&S #589 WR
filter paper. The drying curve was obtained using pressure plates. The filter
paper calibration used 150mL of solution and an equilibration period of 14 days.
The containers were put into a water bath to control the temperature within
0.1 Celsius. Papers were immediately weighed after equilibration and then dried
for 24 hours at which time they were immediately weighed after being taken out
of the oven.
2.8.7 Leong, He and Rahardjo, 2002
This researched focused on why calibrations of filter paper are different in
different labs. They studied data previously obtained from different labs and
checked the quality of paper, suction source, hysteresis and equilibrium time.
The quality of the Whatman #42 and S&S #589 filter paper was studied.
Hamblin in 1981 studied curves with the Whatman #42 paper produced two years
apart and concluded that they were almost identical. The data occupies a narrow
band. Some researchers treat paper with HgCL, pentachlorophenol or ethanol.
Others dont, but there have not been any reports of growth on the papers.
If the suction source is soil with a known suction the calibration data of
known matric suction shows a large scatter due to the uncertainty of the actual

soil suction. This data was taken from Ridley (1995) who wanted calibration to
be performed under conditions close to expected experimental conditions.
2.9 Filter Materials
In 1990, Sibley and Williams studied different filter materials for use in
the filter paper test. Typically the filter paper is used as a matter of convention
but other materials may work better in certain situations. Other materials
researchers have used include sugared blotting paper, salted blotting paper and
Visking membrane. Sibley and Williams chose to test Millipore MF filtration
membrane; pore sizes 0.025 and 0.05 pm, cellulose seamless tubing with and
without sugar coating and Whatman #42 filter paper. These materials were
chosen for their uniformity, stability, sensitivity and durability. The researchers
set up suction plate, pressure membrane and vacuum desiccators for the testing.
Suction plate and pressure membrane were used for matric suction and the
vacuum desiccators were used for total suction. Each filter material was tested in
these apparatus. Temperature fluctuations were 0.1Celsius. The matric suction
testing was allowed to equilibrate for three days while the total suction testing
required 10 days for equilibration. After equilibration the gravimetric water
content was measured. The sensitivity of the filter material was calculated by
using Equation 2.11
Equation 2.11

where w = water content
pF = suction
Results of this testing found that the sensitivity of the Millipore 0.025 pm
was higher than Whatman #42 filter paper between pF 2.5 and 3.5. The cellulose
tubing had different results depending on if it was sugared or not. Ideally, to
obtain the best accuracy the researchers concluded that from pF 5 to 6 sugared
tubing is the best filter material. From pF 4 to 5, unsugared tubing is the best
material. From pF 2.5 to 4, Millipore membrane is the best material, and from pF
2.5 and lower the Whatman #42 filter paper is the best material. However, Sibley
and Williams also concluded that if one filter type is desired then the Whatman
#42 filter paper is the best over the range of suction from pF 2.5 to 6.
Because the test is so sensitive, other filter questions arise. One in
particular that would affect the testing procedure is whether the filter paper has to
be calibrated from every single box used. In 1967, Fawcett and Collis-George
found that filter paper needs to be calibrated if different outlets or batches are
used in the testing. (Sibley and Williams, 1990) If paper from the different outlets
or batches is not calibrated then there are inaccuracies in using a previous
calibration curve for suction calculation. The next investigation by Sibley, Smyth
and Williams in 1990 was to test different boxes of filter paper from the same
batch. The purpose was to find any reason to have to do another calibration each
time you use a new box of paper. Whatman #42 paper was used and four

specimens were cut from each piece of paper. Several suction values from 0 to
6.35 were tested with several papers from each box. For suction from pF 1.7 to
1.95 a ceramic suction plate was used. Pressure membranes were used for suction
2.49,3.01,3.49, and 4.0. Vacuum desiccators for pF 5.15, 5.49, 5.88 and 6.35
were used. After the papers were tested at the various suction levels, Sibley,
Smyth and Williams used a statistical technique called random effects analysis of
variants-using a generalized linear interactive model to estimate with reasonable
certainty whether the variability of the moisture content in the paper was random
or systemic. The results found that the moisture content of the filter paper was
not significantly different. Each box does not have to be calibrated if they are
bought at the same time from the same manufacturer.

Measurement Techniques
3.1 Introduction
Soil suction is measured in the lab using several techniques. Most tests
have no contact between the sample and the measuring device. Most of the
techniques require calibration in an environment of known suction. (Lee and
Wray, 1995) In addition, all rely on the law of physics, that at equilibrium, the
total suction of porous media is related to the temperature, humidity and vapor
pressure of the surrounding air. Suction is then inferred by measuring or defining
the relative humidity and temperature of an air space in equilibrium with the
sample. Several units are used to measure suction. Table 3.1 shows the
conversions for these units.
Atmospheric Pressure (cm) pF kPa Bars psi
1 0 0.0981 0.00098 0.1422
10 1 0.981 0.0098 1.422
10" 2 9.81 0.098 14.22
10J 3 98.1 0.98 142.2
104 4 981 9.81 1422
10" 5 9810 98.1 14220
lo5 6 98100 981 142200
10' 7 981000 9810 1422000
Table 3.1 Conversion of suction units
3.2 Pressure Plate Calibration
The point of the calibration is to provide external energy by applying air
pressure to overcome the retention forces on the soil water. Figure 3.1 illustrates

the test. The applied air pressure drives water out until equilibrium is reached
signaled by water outflow stopping. At that point, applied air pressure equals the
water retention energy of soil at equilibrium. The procedure is repeated at
increments of air pressure change until a curve is produced. This calibration
curve is used for heat dissipation sensors, pressure plate, and fiberglass or gypsum
block techniques of measuring soil suction.
Extension steal ring
Figure 3.1 Calibration using a pressure plate (Lee and Wray, 1995)

3.3 Thermocouple Psychrometer
Thermocouple psychrometers measure relative vapor pressure. This
method can measure suction over a range of 50 to 150,000 psi. The psychrometer
method of measuring suction in the lab uses a thermocouple psychrometer and
psychrometric micro voltmeter to take readings. The thermocouple psychrometer
(Figure 3.2) is placed into a Teflon container with calibration solution or soil
sample, sealed and placed into a larger box lined with polystyrene to help control
temperature fluctuations. A psychrometric micro voltmeter such as a MJ55
Wescor is used to take readings. A cooling circuit is applied to each
psychrometer to cause condensation on the junction when the temperature reaches
dew point and a reading is taken. The voltage difference between the
thermocouple and the reference junctions is compared to a calibration curve to
5 copper vine
LUcrrE tusuw 4 ELASTIC TUIE-nTTHTC-4 | rJl j 1 -SCAUP VITH CASTMO fltSIN £ fwsscR o.mN0 A
Figure 3.2 Thermocouple Psychrometer (Krahn and Fredlund, 1972)

determine suction. This test can also be used in the field for in situ suction
3.4 Pressure Membrane or Pressure Plate
The pressure plate was introduced by Richards and Fireman in 1943. (Lee
and Wray, 1995) High air entry pressure membranes or porous stone provide a
separation between the air and water phases. The membrane is usually a very fine
ceramic stone or other material having a very small pore space that allows the
movement of water but not air. The principle of operation for this test is to
provide pore spaces so small that the displacement pressure is higher than the
suction. The soil sample is placed over a membrane and a known suction is then
applied to the sample. This method measures suction from 0.0 to 6.2 pF. The
extracted air pressure elevates the negative pore water pressure to induce a
positive pore water pressure in the soil causing the water to flow through the
membrane into a reservoir and raise the water level in the burette. Therefore, a
stationary level in the burette indicates the pressure is equal to the soil suction.
This test is sensitive to temperature and a good contact between the sample and
membrane is essential.
3.5 Electrical Resistance Sensor
The sensor was introduced by Boyoucos and Mick in 1940. It consists of
electrodes within a standard porous material to detect the electrical resistance of
the soil. A pair of electrodes is encased within a standard matrix of porous

material which is allowed to equilibrate with the soil. Under a matric suction
gradient between the sensor and surrounding soil an exchange of water occurs
resulting in a change between the electrodes within the sensor. The results are
inferred by electrical resistance. To calibrate the sensor, a material of known
matric suction is tested with the sensor.
3.6 Electronic Humidity Sensors
The relative humidity (RH) is related to suction by Equation 3.1
S = log[1.284657£6 + 4.7030](ln%)0) Equation 3.1
S = suction in pF
t = C
H = relative humidity (%)
Most of these devices are accurate to five percent relative humidity. To run the
tests, sensors are placed in a jar of soil to estimate the total suction. Temperature
fluctuations must be kept to 2 Celsius.
3.7 Tensiometer
Tensiometers are saturated fine-grained ceramic cups that are buried in soil and
attached to a device that measures negative pressure. Tensiometers cannot
measure suction values above seven to eight psi. However, these instruments can
be used easily in the field. An example of a tensiometer is shown in Figure 3.2

3.8 Heat Dissipation Sensor
Heat dissipation sensors measure the rate of heat dissipation so that the
water content can be obtained. A standardized porous ceramic probe is inserted
into the soil and allowed to come to equilibrium. After it reaches equilibrium, a
heat flux is applied to the center of the porous probe and the increase in
temperature is measured over a fixed period. Suction is determined from the
temperature readings. The temperature rise of the sensor can be calibrated against
the matric suction by using a pressure plate apparatus.

3.9 Thermal Matric Potential
Thermal matric potential uses sensors that operate on the principle that the
thermal conductivity of a porous material is directly proportional to the volume
water content of that material. It measures the rate of heat dissipation from
porous ceramic. The water content of the ceramic can be determined which is a
function of the matrix suction. Instruments have been developed with calibrated
porous ceramic sensors that can be inserted into the soil. Once the sensor reaches
equilibrium a controlled heat pulse applied in the center of the probe and the
increase in temperature is measured over a period. Applying the factory
calibration allows the user to calculate the matrix suction. This method works up
to 30 psi suction.
3.10 Filter Paper Method
Filter paper is another widely used method and is simple to run. It
measures a range of suction from 0-150,000 psi. Total and matric suction can be
measure with the filter paper test, although usually only total suction is measured.
For total suction measurement, the filter paper must be separated from the soil
sample, and for matric suction the filter paper is placed directly on the sample.
The test requires a sensitive scale that measures to 0.001. If a scale is not
sensitive enough the results of the test can be miscalculated as shown in Table
3.2. Hysteresis is another concern of filter paper measurement. The hysteresis is

the difference observed in the moisture retention characteristics depending on
whether the sample is wetting or drying, To account for this phenomenon, the
filter paper should always be used in the same cycle (wetting or drying) during the
test as was used in the calibration.
Balance Accuracy, g Water Cont., Mw/Ms wet Water Cont., Mw/Ms Dry Suction (h), kPa wet Suction (h), kPa dry Suction (hXpF wet Suction (h), pF dry
0.01 1.333 200 7.997 46.46 3.062 3.196
0.001 0.133 20 0.799 4.61 0.306 0.320
0.0001 0.013 2 0.080 0.461 0.031 0.032
Wet: w=150%, h=1.334 kPa or 1.134 pF, Dry: w=l%, h=177800kPa or 6.259 pF
Table 3.2 Suction error vs. balance accuracy (Swarbrick, 1995)
The filter papers moisture retention characteristic is used to infer the
suction; therefore, calibration curves must be determined before running the test
with a soil sample. Calibration curves using salt solutions of known
concentration are determined for the type of filter paper used before running the
test with a sample. The most commonly used filter paper is Schleicher and
Shuell, No. 589 white ribbon type or equivalent.
This test can also be adapted to the field. The paper is sealed underground
for a period of equilibration then weighed. Problems can arise when the paper is
taken out of the ground to be weighed because the paper rapidly loses moisture to
the atmosphere. An alternative is to use an electric balance in a soil container or
down a borehole. The balance is placed in the same container as the soil and the

filter paper is hung from the balance. The balance counts the electrical impulses
needed to support the data takes readings every minute and then stores the data.

Laboratory Facility for Suction Measurement
4.1 Standard Laboratory Requirements
The Standard Test Method laid out by ASTM D5298-94 has the following
requirements for suction measurement.
1. Filter paper must be ash-free quantitative Type II filter paper such as
Whatman No. 42, Fisherbrand 9-790A or Schleicher and Schuell No. 589
White Ribbon.
2. The sample container should be 120 to 240 mL capacity glass or metal
(rust free).
3. Moisture containers should be aluminum or stainless steel
4. An insulated chest capable of maintaining one degree Celsius when
external temperature fluctuation is three degrees Celsius.
5. An analytical balance with 0.000lg readability is required. Shown in
Table 4.1 is the difference in suction calculation when a balance does not
have the proper accuracy
6. A thermostatically controlled oven capable of maintaining a uniform
temperature of 1105 degrees Celsius
7. Metal block with greater than 500 grams mass to hasten the cooling of the
metal moisture containers

Balance Accuracy, g Water Cont., Mw/Ms wet Water Cont., Mw/Ms dry Suction (h), kPa wet Suction (h), kPa dry Suction (h), pF wet Suction (h),pF dry
0.01 1.333 200 7.997 46.46 3.062 3.196
0.001 0.133 20 0.799 4.61 0.306 0.320
0.0001 0.013 2 0.080 0.461 0.031 0.032
Wet: w=150 %,h=1.33^ kPa, Dry: w=l%, h=177800kPa
Table 4.1: Suction error vs. balance accuracy (Swarbrick, 1995)
8. Spacer to measure total suction such as screen wire, o-rings or brass discs
9. Dessicator jar containing silica gel or anhydrous calcium sulfate to store
filter papers
10. Miscellaneous equipment such as gloves, tweezers, flexible electrical tape.
11. Equations to calculate the water content for each filter paper are located in
Appendix A.
4.2 Laboratory Facilities for This Research
The laboratory facilities used for this research had the following
conditions. Temperature fluctuations were 3 Celsius for most of the tests. The
oven was located in the same room as the equilibration so that the weighing could
take place within seconds of removing the samples from equilibration or from the
oven. Although this was the most convenient way to run the tests, the
temperature fluctuations became too great near the end of testing due to changes
in the building's heating and air conditioning system. The samples had to be
moved. The last few tests were then moved to a separate room for equilibration

where the temperature fluctuation was 1-2 Celsius. This room however could
not accommodate the oven and the analytical balance had to be moved between
the rooms to measure the weights of the filter paper.
The analytical balance used is a Mettler Toledo AG204 balance accurate
to O.OOOlg and included dynamic weighing. The dynamic weighing option was
useful, especially when weighing the filter paper right after it was removed from
the oven. Hot items cause air circulation that disturbs a sensitive scale making it
difficult to determine the mass. The dynamic weighing option determines the
mass of the object and then returns one number to the display.
Containers used for equilibration were stainless steel. The spacer used to
separate the paper from the solution was a crumb cup, a type of strainer used in
sinks that was turned upside down and paper balanced on top. Filter paper used
was Whatman #42. The lids of the containers were greased with mineral oil to
prevent condensation from forming and dropping onto the filter paper. (Houston
et al, 1994) The containers were sealed with flexible electrical tape.
Aluminum moisture tins were used to weigh the filter paper after equilibration
and after drying.
Two salt solutions were used in testing: sodium chloride (NaCl) and
magnesium chloride (MgCb). Five solutions each of NaCl and MgC^were mixed
using ultrapure water obtained from the University of Colorado at Denver
Chemistry department. The NaCl suction levels used were 1.99 log kPa, 2.49 log

kPa, 2.99 log kPa, 3.49 log kPa, and 3.99 log kPa. The MgC^ suction levels used
were 2.12 log kPa, 2.51 log kPa, 3.11 log kPa, 3.43 log kPa and 3.91 log kPa.
4.3 Suction Measurement Program
The suction measurements were determined by taking parts of procedures
from Swarbrick, 1995, Houston et al, 1994, Bulut et al, 2001, Snethen and
Johnson, 1980, and ASTM D5298-94. The laboratory equipment conforms to the
ASTM guidelines as far as containers, temperature fluctuations and salt solutions
is concerned. Each of the other researchers modified the ASTM procedure to
some extent. In addition, each researcher had different containers, ways of
suspending the filter paper, different filter papers, and different salt solutions. For
the purposes of this testing, it was decided to tests parts of procedures instead of
all of them which would require several different types of containers, filter paper
and salt solutions. The parts of procedures which were used for testing are oven-
drying filter paper before testing, how long the equilibration lasts, how fast the
papers are weighed after oven-drying, and how much solution is used.
4.3.1 ASTM D5298-94
1. Filter papers are oven dried for at least 16 hours before running test
2. 50 mL of solution is used
3. Two filter papers are placed in each equilibration container
4. Equilibration time is seven days
5. After drying filter weigh immediately upon removal from oven

4.3.2 7-Day Test based on Houston, Houston and Wagner, 1994
1. Filter papers are not dried before the test
2. 50 mL of solution is used
3. Two filter papers are placed in each equilibration container
4. Equilibration time is seven days
5. After drying filter weigh immediately upon removal from oven
4.3.3 5-Day Test based on Swarbrick, 1995
1. Filter papers are not dried before the test
2. 50 mL of solution is used
3. Two filter papers are placed in each equilibration container
4. Equilibration time is five days
5. After drying filter weigh immediately upon removal from oven
4.3.4 Wait before Weigh Test based on Snethen and Johnson, 1980
1. Filter papers are not dried before the test
2. 50 mL of solution is used
3. One filter paper is placed in each equilibration container
4 Equilibration time is seven days
5. After drying filter paper weigh within fifteen minutes upon
removing from oven

4.3.5 Wait to Weigh with Two Papers based on Snethen and Johnson, 1980
This test was run because with one filter paper there were half as many data
points than other tests. To better compare this test with the others, it was run
again with two filter papers in each equilibration container.
4.3.6 14-Day Test based on Bulut et al, 2001
1. Filter papers are dried before test
2. 150mLofsolutionisused
3. Two filter papers are placed in equilibration container
4. Equilibration time is 14 days
5. After drying filter weigh immediately upon removal from oven
4.3.7 4-Day Test based on Swarbrick, 1995
1. Filter papers not dried before test
2. 50 mL of solution is used
3. Two filter papers are placed in equilibration container
4. Equilibration time is two days
5. After drying filter paper weigh immediately upon removing from
the oven
4.3.8 2-Day Test based on Swarbrick, 1995
1. Filter papers not dried before test
2. 50 mL of solution is used
3. Two filter papers are placed in equilibration container

4. Equilibration time is two days
5. After drying filter paper weigh immediately upon removing from
the oven
4.3.9 Testing Matrix
Testing of these methods took place with both sodium chloride solution
and magnesium chloride solution. The purpose of running tests with both
solutions was to get an idea of how different the results are when using different
solutions as well as checking the results of each of the methods. Due to time and
equipment constraints, half as many MgC12 solutions were run as NaCl solutions.
Table 4.1 lists the tests.
Procedure NaCI Number of Tests MgCh Number of Tests
ASTM D5298-94 20 5
7-Day (Houston et al, 1994) 10 5
5-Day (Swarbrick, 1995) 20 5
Wait to Weigh (Snethen and Johnson, 1980) 10 5
4-Day Test 10 0
2-Day Test 10 5
14-Day (Bulut et al, 2001) 10 5
Wait to Weigh 2 papers (Snethen and Johnson, 1980 10 0
Total Number of Tests 100 30
Table 4.2 Tests run for this research

Data Analysis and Discussion
5.1 Sodium Chloride Tests
All of the data obtained from these tests can be found in Appendix B.
Some of the water contents calculated after the test are negative numbers or
numbers greater than 100%. Those numbers will be ignored in the rest of the
analysis and are highlighted on the spreadsheet. There werent many results
determined for the 14-Day test. This is because the test required 150 ml of
solution and the containers are only 240 ml containers plus the crumb cups had to
be stacked on one another to try to suspend the filter paper above the solution.
Often, the filter papers fell into the solution and had to be discarded. Other tests
had a few filter papers drop into the solution during removal from the
equilibration container and there are no results for those filter papers.
Statistical analysis was applied to the data as a whole and to the individual
tests. A graph showing the data for all the tests minus the negative numbers and
numbers with unusually high water contents is shown in Figure 5.1. This graph
shows that all of the data falls within a relatively narrow band. Figure 5.2 shows
the equation of the linear trend through all of the data. The trend line is a least-
squares linear regression analysis. The coefficient of determination (R ) is
included on the graph with the equation of the line.

Suction (log kPa)
Waittoweigh 14-day A7-day X5-day XASTM-2 ASTM-1 + 2-day A Wait to Weigh (2) 04-day O 5-day (2)

O *i isrvWGKS. ctK GS>ft
1 t & 1 U/ KA/9P L\ Aft
+ T LU£: WK\ /\ V
/k Ov O vJCmjPnK KJ 3Kt XX
LW£SL> v ^1 X.T H tJi L-> /K

0 5 10 15 20 25 30 35 40 45 50
Water Content (%)
Figure 5.1- Graph of all tests run with NaCl solution

A question raised was whether the top or bottom filter papers absorbed
significantly different amounts of water vapor. To check the difference between
the filter papers, the data for suction equaling 1.99 log kPa was evaluated. Figure
5.3 shows the graph of the difference in moisture content between the top and
bottom filter papers. The difference is between 0.5 and 5.4 percent water content.
The quality of the data was evaluated using standard deviation and
coefficient of variability. For each suction level, the average (AVG), standard
deviation (STDEV) and coefficient of variability (CV) percent water content were
calculated and graphed. The calculations are shown in Table 5.1 and the graph is
shown in Figure 5.4. The standard deviations and coefficient of variations for
each suction level are low suggesting that the data is reliable.
Suction (log kPa) AVG STDEV CV
1.99 33.2135 4.0450 0.1218
2.49 27.3912 3.9538 0.1443
2.99 22.2171 3.5364 0.1592
3.49 18.8937 3.5363 0.1872
3.99 13.8268 2.1373 0.1546
Table 5.1 Average, Standard Deviation and Coefficient of Variation for
each suction level______________________________________________________

Suction (log kPa)
Figure 5.2 Linear trend of all the data for tests run with NaCl solution

AW.C. Top Filter Paper a W.C. Bott. Filter Paper
Figure 5.3 Comparison of the top and bottom filter papers at 1.99 log kPa suction

The quality of the data for each procedure was also evaluated. The calculations
are shown in Table 5.2 and the graph is Figure 5.5. This data is more variable
because the water contents change with each suction level. However, the
variation is still low for each procedure. The 14-Day test has the least amount of
variation because there are not many data points for that test. The 2-Day test has
the most amount of variation. It appears that at room temperature, the test does
not fully equilibrate in two days and the data is more scattered.
Procedure STDEV AVG CV
ASTM-1 7.8415 23.4614 0.3342
ASTM-2 9.1068 24.1787 0.3766
7-Day 7.4669 27.9871 0.2668
5-Day 8.0130 26.5810 0.3015
5-Day (2) 6.1540 22.1950 0.2773
Wait to weigh 8.1965 23.6248 0.3469
Wait to weigh (2) 6.5623 21.5573 0.3044
14-Day 5.0859 20.9574 0.2427
4-Day 5.1081 22.5125 0.2269
2-Day 9.3937 21.4122 0.4387
Table 5.2 Standard Deviation, Average, Coefficient of Variation for
each procedure
5.1.1 ASTM D5298-94 Test
The ASTM procedure was followed for two sets of tests. Appendix B
contains the raw data for these two tests. The second time there was more scatter
in the data. The graphs of the data and equation of the best fit line are


Suction (log kPa)
Figure 5.4 Average, Standard Deviation and Coefficient of Variation of all the data for NaCl solutions

Figure 5.5 Average, Standard Deviation, Coefficient of Variation for each procedure

shown in Figures 5.6 and 5.7. In this case, the first time the test was run the
equation of the best fit line has a higher R2 value.
5.1.2 7-Day Test Based on Houston, Houston and Wagner, 1994
The 7-Day test was run one time. The graph and equation are shown in
Figure 5.8. This test was slightly more scattered than the first ASTM test.
However, the R2 value is still relatively high.
5.1.3 5-Day Test based on Swarbrick, 1995
The 5-Day test results shown in Figure 5.9 are much closer than any of the
other tests. This test was repeated in the temperature controlled room where
temperature varied 2 Celsius. The second test is shown in Figure 5.10. It also
suggests as some other researchers have noted that a full seven days are not
needed for equilibration. None of the calculated water contents had to be thrown
out of the data set when the test was run under more stringent temperature
variation. However the second test has slightly more scatter in the data despite
the reduced temperature variation.
5.1.4 Wait to Weigh Based on Snethen and Johnson, 1980
The wait to weigh test was run twice. The first time, only one filter paper
was used in the test as Snethen and Johnson (1980) did in that research. After
drying, the moisture tins were allowed to cool covered for five to ten minutes
before weighing. The first graph in Figure 5.11 only has nine points because only
one paper was used in each test. The data are not scattered as much as some of

Suction (log kPa)
Figure 5.6 ASTM-1 Test

Suction (log kPa)

y = -0.062 Rz = lx + 4.3387 1.6206

0 5 10 15 20 25 30 35 40 45
Water Content (%)
Figure 5.7 ASTM-2 Test

Suction (log kPa)
Figure 5.8 7-Day Test

the other tests and in the Colorado climate, it appears that if the paper is kept
covered, the results are not affected by waiting to weigh the paper after drying.
This procedure also obtains better data because when the tins are hot the air
currents are disturbed in the chamber of the scale and weighing the tins can be
challenging. The second time, the test was repeated with two filter papers and in
the temperature controlled room where temperature variation was 1.5 Celsius.
That graph is shown in Figure 5.12. None of the calculated water contents had to
be thrown out of the data set. This data has a slightly higher R2 value than the
first 5-Day test.
5.1.5 14-Day Test Based on Bulut, Lytton and Wray, 2001
Running the suction test for fourteen days is usually not feasible because
of the length of time required before getting results. However, this test was run
because Bulut et al (2001) made some significant changes to the procedure. The
first is running the test a full fourteen days. The second significant change was
that they used 150mL of salt solution instead of 50 mL. Since three times as
much solution is used in the test and the tins used in this research are only 240mL
tins, there were logistical problems running the test. Most of the filter papers fell
into the solution because they were not suspended high enough above the
solution. However, six points were obtained, shown in Figure 5.13. These points
are not significantly different than the other tests.

Suction (log kPa)
Figure 5.9 5-Day Test

Suction (log kPa)
0 r 1 i i-----------------1 i---------------1----------------1---------------
0 5 10 15 20 25 30 35 40
Water Content (%)
Figure 5.10 5-day test (second run)

Suction (log kPa)
Figure 5.11 Wait to Weigh Test

5.1.6 4-Day Test Based on Swarbrick, 1995
The 4-day test was run to check the needed equilibration time for the filter
paper. It was set up like the 5-day test because the first 5-day test data showed the
least amoimt of variability. The graph is shown in Figure 5.14. Like many of the
tests the data has some scatter but the R2 value is still high enough to be
considered a good test. Less than five days equilibration may be sufficient to
obtain a good calibration curve.
5.1.7 2-Day Test Based on Swarbrick, 1995
The 2-Day test was run to check the needed equilibration time for the filter
paper. The test was set up like the 5-Day test. The graph is shown in Figure 5.15.
The data showed more scatter than most of the tests, but is still statistically good
and this could be used with the knowledge that the error in calculating suction
value may be greater than other procedures. However, Swarbrick (1995) did find
that the wetter a soil sample is the longer equilibration takes. Two days may only
be appropriate for dry samples. If the temperature of the room were higher,
equilibration may occur sooner but that was outside the scope of this research.
5.2 Magnesium Chloride Tests
Tests were run using five suction levels of magnesium chloride solutions.
Only five tests were run for each procedure. Many researchers use different
solutions for different levels of suction (Houston, Houston and Wagner, 1994)
during testing. The purpose of running some tests with a different salt solution

was to compare the differences in the calibration curves. The raw data and graphs
of each of the curves are in Appendix C. The 14-Day test was run, but most of
the filter papers fell into the solution or were obviously saturated so no data was
obtained for that test.
All of the data was evaluated similarly to the sodium chloride data. The
graph of all the data is shown in Figure 5.15. The standard deviation, average and
coefficient of variation for each of the suction levels were calculated as shown in
Table 5.3. The coefficient of variation is lower than the sodium chloride tests.
Suction (log kPa) STDEV AVG CV
2.12 2.2898 32.5306 0.0704
2.51 3.3258 31.7124 0.1049
3.11 3.2585 25.3541 0.1285
3.43 3.1611 23.0902 0.1369
3.92 2.5601 18.3805 0.1393
Table 5.3 Standard Deviation, Average and Coefficient of
Variation for each suction level
A linear regression analysis was applied to all of the data as shown in Figure 5.16,
The R2 value of all of the data is 0.7631. The sodium chloride data has a slightly
higher R2 value than the magnesium chloride data. Of each individual test, the 5-
Day test had the best linear fit in the data. The ASTM and Wait to Weigh
procedures were respectively the second and third best.

5.3 Comparison of the Equations of the Linear Regression Analyses for
both Sodium Chloride and Magnesium Chloride Solutions
Since none of the tests repeated the same exact curve as the other tests
Table 5.4 shows calculated suctions from the different procedures and salt
solutions. There is a range of calculated values from 3.71 to 5.05 log kPa suction.
The true suction value of ten percent water content is not known. This table only
gives a representative range of values for each of the different curves. An
interesting result from these calculations is that the equations obtained from the
magnesium chloride solutions result in higher suction values than the sodium
chloride solutions.

Suction (log kPa)
Figure 5.12 Wait to Weigh (2)

Suction (log kPa)
0 5 10 15 20 25 30 35
Water Content (%)
Figure 5.13 14-Day Test

0 5 10 15 20 25 30 35
Water Content (%)
Figure 5.14 4-day Test

Figure 5.15- 2-Day Test

Suction (log kPa)
7-Day A 5-Day 3KASTM 0 Wait to Weigh X 2-Day
0 5 10 15 20 25 30 35 40
Water Content (%)
Figure 5.16- MgCl2 All Tests

X CA X4 xx x : < & £

4 iu WJ W AV ^ Si A X+


Suction (log kPa)
Figure 5.17 Linear trend for all tests run with MgCl2 solutions

NaCI Tests Equation R2 Hypothetical Water Content (%) Calculated Suction (log kPa)
Wait to Weigh (2) (Snethen and Johnson, 1980) y = -0.1084x +5.3262 0.9609 10 4.2422
5-Day (Swarbrick, 1995) y = -0.0879X + 5.2092 0.9607 10 4.3302
Wait to Weigh (Snethen and Johnson, 1980) y = -0.0914x + 5.2044 0.9504 10 4.2904
5-Day (2) (Swarbrick, 1995) y = -0.1081 x+ 5.3366 0.8903 10 4.2556
4-Day y = -0.1313x +5.8214 0.8709 10 4.5084
ASTM -1 y = -0.0994X + 5.4285 0.8684 10 4.4345
7-Day (Houston et al, 1994) y = -0.082x +5.1373 0.8121 10 4.3173
2-Day y = -0.0709X + 4.5643 0.7991 10 3.8553
ASTM 2 y =-0.0621 x + 4.3387 0.6206 10 3.7181
MgCb Tests
5-Day (Swarbrick, 1995) y = -0.1316x +6.2886 0.9646 10 4.9726
ASTM y = -0.0852x +5.1682 0.8851 10 4.3162
Wait to Weigh (Snethen and Johnson, 1980) y = -0.0979X + 5.7545 0.8838 10 4.7755
7-Day (Houston et al, 1994) y = -0.1091x +6.1495 0.7269 10 5.0585
2-Day y = -0.0845X + 5.2286 0.6947 10 4.3836
Table 5.4 Hypothetical suction value calculated from equations determined from each procedures data

The purpose of this research was to compare the different procedures
researchers use to run soil suction calibration curves and tests. Typically different
laboratories can not compare the results of filter paper suction tests because their
calibration curves are different. By running tests of different procedures and
different salt solutions it was hoped that a specific procedure would stand out as
repeatable and statistically the best procedure.
All of the procedures tested produced statistically good calibration curves.
All of the R values of the linear regressions were above 0.5. All of the equations
returned suction values that were similar when calculated with a hypothetical
water content of 10 percent.
One procedure did stand out slightly because it had the highest R value
whether the test was run with a sodium chloride solution or magnesium chloride
solution. This test was also run at temperature fluctuations of 3 Celsius and 2
Celsius using the sodium chloride solution. The results from both temperature
ranges were similar. The second time the test was run the data was more scattered
but still better than several of the other procedures. Because of these results, the
5-day procedure based on Swarbrick, 1995 is recommended. The advantages
over the ASTM D5298-94 procedure is that this test is only run for five days and
the filter papers do not have to be oven dried before the test is run.

Another conclusion reached by this research was that the measurement of
the weight of the filter paper after it has been dried does not necessarily have to
happen within seconds as the ASTM D5298-94 standard procedure requires. The
tests run using the method based on Snethen and Johnson, 1980 resulted in high
R2 values as well. After drying, the moisture tins were covered and removed from
the oven allowing them to cool for a few minutes before weighing. The
advantage of waiting is that it is easier to weigh the tins when they are cool rather
than hot. The air current disturbance from a hot tin makes weighing with
accuracy to 0.0001 difficult. However, this was not tested in a high humidity
environment, as Colorado is an arid climate, and it was not tested as part of the 5-
Day test. Further research is suggested.
Reliable calibration curves for every procedure were generated by this
research. These procedures all had different aspects whether it was equilibration
time, if the filter paper was oven dried or not before testing, or whether the filter
paper was weighed immediately after being dried back. The constant in all of the
testing was that it was all performed by one person in the same laboratory setting.
Further research is suggested to determine whether the laboratory environment or
the personnel running the tests affect the results.
Temperature fluctuations have some effect on the quality of the data. It is
difficult to determine if the temperature fluctuations 3 Celsius caused the

negative or greater than 100 percent water content values or if those erroneous
values were caused by human error. By the time the samples were moved to a
room with 2 Celsius temperature variation, the test had been run many times by
the same person. None of those tests produced negative or greater than 100
percent water contents. Experience with the test could have prevented mix-ups in
the data. However, several researchers have reported that the temperatures have
to be tightly controlled and that 3 Celsius is too great a range. Based on their
findings, it is recommended that the testing room or box be controlled to at least
2 Celsius.
Using different salt solutions slightly affected the data. The hypothetical
calculated suction values from the magnesium chloride calibration curves were
slightly higher than the values calculated from sodium chloride calibration curves.
Many types of salt solutions are used by researchers including sodium chloride,
magnesium chloride, potassium chloride and calcium chloride. More research on
the different salt solutions is recommended to determine if the type of salt affects
the total suction calibration curve.


Suction (log kPa) 1.99 1.99
Equilibrium Tin # 1 1
Moisture Tin # A2 A6
Top or Bottom Filter Paper T B
Cold Mass (Tc) 34.2025 34.0583
Wet Paper + Tc (Mi) 34.4768 34.3510
Dry Paper + Hot Mass (M2) 34.4077 34.2751
Hot Mass (Th) 34.1993 34.0516
M2-Th = M, 0.2084 0.2235
M1-M2-Tc+Th = Mw 0.0659 0.0692
Water Content (%) = 100*(MW/M,) 31.6219 30.9620
Sample Calculations to Determine Water Content of the Filter Paper


Suction (log kPa) 1.99 1.99 1.99 1.99 2.49 2.49 2.49 2.49 2.99 2.99
ASTM-1 (% W.C.) 34.6048 35.1313 32.7384 27.3312 27.0270 . >70.6987- 1-96.3991 DFP -96.3991 29.8395
ASTM-2 (% W.C.) 31.6590 30.6008 43.7704 ,~84.-2806i 27.9580 30.6216 13.7484 26.8561 21.9259 26.1860
7-day (% W.C.) 38.8634 33.9678 40.6741 37.0741 33.7800 30.2441 33.5853 27.5953 23.5614 22.3194
5-day (% W.C.) 33.8876 38.0249 33.5620 35.4260 32.1927 32.1927 29.0783 32.7788 25.6304 27.8584
5-day-2 (% W.C.) 29.8670 28.2243 33.5089 28.0275 23.5566 25.3358 26.8109 26.7531 22.7099 21.6682
Wait to weigh (%W.C.) 35.7045 35.0282 1FP 1FP 121.0816 26.3594 1FP 1FP 25.7117 22.3031
Wait to weigh-2 (% W.C.) 31.6219 30.9620 30.5898 32.9038 25.0576 25.2575 25.9191 25.1319 20.7195 20.4301
14-day (% W.C.) DFP DFP .;-12;2357-.X 285.947.0 : 131:8592 -7816893;;; 29.1382 25.4054 18.3784 18.8292
4-Day (% W.C.) 26.833713 29.791183 27.734554 28.558352 24.755245 22.747953 26.332574 26.4637 DFP DFP
2-Day (% W.C.) 36.1491 30.9100 38.1892 34.1252 i 91:87731 -3910678 27.7831 31.2698 16.1388 20.2673

Suction (log kPa) 2.99 2.99 3.49 3.49 3.49 3.49 3.99 3.99 3.99 3.99
ASTM-1 1^3812927 X 857600681; 23.4093 18.5047 21.9601 21.5412 15.5880 13.6302 13.6385 13.5159
ASTM-2 DFP DFP 13.0398 18.1130 DFP DFP 14.1917 15.6522 DFP DFP
7-day 23.5663 23.3094 21.3503 24.2161 -248.0271 28.7500 218.6036: -15.9427 18.2927 14.6320
Wait to weigh 1FP 1FP 20.1901 21.6192 1FP 1FP 13.0415 12.6658 1FP 1FP
Wait to weigh (2) 19.6034 21.4970 16.4778 15.1095 17.5495 17.5884 14.9294 14.9515 13.1186 11.7267
5-day 24.2424 26.4534 19.4070 20.6044 DFP DFP 13.7483 14.0000 12.7897 196:2608.
5-day (2) 21.1574 23.1850 18.4294 16.8267 19.5872 17.0183 8.1707 14.1376 16.7295 DFP
14-day DFP DFP 17.1806 16.8126 DFP DFP DFP DFP DFP DFP
4-Day 21.0099188 23.947615 17.479377 13.545455 20.429616 19.306467 DFP DFP 14.83516 16.42857
2-Day 12.2058 19.6412 15.9701 24.9877 14.2987 14.4014 8.9959 14.7770 13.4006 11.9080
DFP = dropped filter paper into solution
1FP = only one filter paper used for this test
All data obtained from the sodium chloride tests


Suction (log kPa) 2.1239 2.1239 2.5105 2.5105 3.1149 3.1149 3.4376 3.4376 3.9164 3.9164
ASTM 32.1055 33.2202 33.7882 34.7379 20.7059 23.8579 21.0967 18.7003 15.8473 17.8556
7-day 33.9286 29.8337 32.6975 37.6193 27.4519 30.0490 24.8646 27.4360 20.3076 22.6837
Wait to Weigh 36.4901 1FP 31.4001 1FP 29.9728 1FP 20.8539 1FP 20.9009 1FP
5-day 30.1684 31.9676 27.5576 30.2860 25.1729 23.8498 20.9472 21.8173 17.8063 18.7415
2-Day DFP DFP 29.4255 27.8994 25.0819 22.0447 27.7726 24.3231 14.7231 16.5584
DFP = dropped filter paper into solution
1FP = only one filter paper used for this test
All data from the magnesium chloride tests

Suction (log kPa)

Suction (log kPa)


y =-0.1091 r2 = o k + 6.1495 .7269 4

Water Content (%)
7-Day Test

5-Day Test

Suction (log kPa)
Wait to Weigh Test

Suction (log kPa)

y = -0.0845X + R2 = 0.69 5.2286 17

15 20 25
30 35
Water Content (%)
2-Day Test

Al-Khafaf, S. and RJ. Hanks. 1972. Evaluation of the Filter Paper Method for
Estimating Soil Water Potential. Soil Science, vol. 117. no. 4. 194-199.
ASTM D5298-94. 1994. Standard Test Method for Measurement of Soil
Potential (Suction) Using Filter Paper. Annual Book of ASTM Standards.
Bulut Lytton, and Wray. 2001. Expansive Clay Soils and Vegetative Influence
on Shallow Foundations, eds. C. Vipulandandan, M.B. Addison, andM.
Hasen. Houston. ASCE. 243-261
Cocka and Birand. 2000. Suction-Swelling Relations. Proceedings of Geo-
Denver 2000. eds. C.D. Shakleford, S.L. Houston, N.Y. Chang. Denver.
ASCE. 379-392
Garbulewski, K. and S. Zakowicz. 1995. Suction as an Indicator of Soil
Expansive Potential. Unsaturated Soils, eds. E.E. Alonso and P. Delage.
Rotterdam. A.A. Balkema. 593-599.
Houston, S.L., W.N., Houston and A. Wagner. 1994. Laboratory Filter Paper
Suction Measurements. Geotechnical Testing Journal, vol. 17 no. 2. 185-
Johnson, L.D. 1973. Influence of Suction on Heave of Expansive Soils. U.S.
Army Engineer Waterways Experiment Station. Miscellaneous paper S-
73-17. Vicksburg.
Krahn, J. and D. G. Fredlund. 1972. On Total, Matric and Osmotic Suction. Soil
Science, vol. 114 no. 5. 339-348.
Lapin, L.L. 1990. Probability and Statistics for Modem Engineering. 2nd ed.
Belmont. Duxbury Press.
Lee, H.C. and W.K. Wray. 1995. Techniques to Evaluate Soil Suction A Vital
Unsaturated Soil Water Variable. Unsaturated Soils, eds. E.E. Alonso and P.
Delage. Rotterdam. A.A. Balkema. 615-622.

Leong, E.C., L. He, and H. Rahardjo. 2002. Factors Affecting the Filter Paper
Method for Total and Matric Suction Measurements. Journal of Geotechnical
Testing, vol. 25 no. 3. 322-332.
Mahler, C.F. and L.C.D. de Oliveira. 1997. Measurement of Total Suction in situ
of Porous Soils of Sao Paulo Using the Filter Paper Method. Recent
Developments in Soil and Pavement Mechanics, ed. M. Almeida. Rotterdam.
A.A. Balkema. 243-247.
McKeen. 1980. Field Studies of Airport Pavements on Expansive Clays.
Proceedings of the 4th International Conference on Expansive Soils. Vol. 1.
--.1992. A Model for Predicting Expansive Soil Behavior. Proceedings of the 7th
International Conferences on Expansive Soils. Vol. 1. Dallas. ASCE. 1-6.
. 2001. Investigating Field Behavior of Expansive Clay Soils. Expansive Clay
Soils and Vegetative Influence on Shallow Foundations, eds. C.
Vipulanandan, M.B. Addison and M. Hasen. Houston. ASCE. 82-94.
McQueen, I.S. and R.F. Miller. Calibration and Evaluation of a Wide-Range
Gravimetric Method for Measuring Moisture Stress. Soil Science, vol. 106
no. 3. 225-231.
. 1974. Approximating the Soil Moisture Characteristics From Limited Data:
Empirical Evidence and Tentative Model. Water Resources Research, vol. 10
no. 3. American Geophysical Union. 521-527.
Nelson, J.D. and D.J. Miller. 1992. Expansive Soils: Problems and Practice in
Foundation and Pavement Engineering. New York. John Wiley and Sons.
Sibley, J.W., G.K. Smyth and D.J. Wiliams. 1990. Suction-Moisture Content
Calibration of Filter Papers from Different Boxes. Geotechnical Testing
Journal, vol. 13 no. 3. 257-262.
Sibley, J.W. and D.J. Williams. 1990. A New Filter Material for Measuring Soil
Suction. Geotechnical Testing Journal, vol. 13 no. 4. 381-384.

Snethen, D.R. Characterization of Expansive Soils Using Soil Suction Data.
Proceedings of the 4th International Conference on Exnansive Soils. Vol. 1.
Snethen, D.R., and G. Huang. 1992. Evaluation of Soil Suction-Heave
Prediction Methods. Proceedings of the 7th International Conference on
Expansive Soils. Vol. 1. Dallas. ASCE.
Snethen, D.R. and L.D. Johnson. 1980. Evaluation of Soil Suction from Filter
Paper. U.S. Army Engineer Waterways Experiment Station. Miscellaneous
paper GL-80-4. Vicksburg.
Swarbrick, G.E. 1995. Measurement of Soil Suction Using the Filter Paper
Method. Proceedings of the 1st International Conference on Unsaturated
Soils, ed. E.E. Alonso and P. Delage. Vol. 2. Paris.
Woodbum, J.A. and B. Lucas. New Approaches to the Laboratory and Field
Measurement of Soil Suction. Unsaturated Soils, ed. E.E. Alonso and P.
Delage. Rotterdam. A.A. Balkema. 667-671.