ESLAMI-FELLENIUS

MODULUS NUMBER CHART

CHART
100
45 Sand to Sandy Gravel

35
Sand
Cone Stress, qE (MPa)

25
10
20 Silty Sand

Sandy Silt
15 Clayey Silt

1 10 Silty Clay

5
Clay

0.1
0 20 40 60 80 100
Sleeve Friction (kPa)

Massarsch, K.R., Fellenius, B.H., and Terceros, M.H., 2020.

Compressibility and stress history of granular soils from
CTPU and DMT. Proceedings of the 6th International
Conference on Geotechnical and Geophysical Site
Characterization, Budapest, September 7 -11. 10p.
 Compressibility of granular soils from CPTU and DMT
K.R. Massarsch
Geo Risk & Vibration Scandinavia AB, Bromma, Sweden, [email protected]
B.H. Fellenius
2475 Sidney, BC, Canada, V8L 2B9, [email protected]
M. Terceros H.
Incotec S.A., Santa Cruz de la Sierra, Bolivia, [email protected]

ABSTRACT: The compressibility of a granular soil is evaluated, based on CPTU and DMT data. A concept is described
how the cone resistance can be adjusted with respect to the mean effective stress in order to reflect soil strength and
stiffness independent of depth. This stress adjustment makes it possible to estimate the tangent modulus number and the
constrained modulus, based on cone resistance measurement. The constrained modulus as measured by DMT was con-
verted into tangent modulus number. Thus, it is possible to compare the tangent modulus number and constrained modulus
obtained from CPT and DMT measurements, respectively. The results of extensive CPTU and DMT investigations from
an ISSMGE test site, composed of sand and silty sand, were analyzed to determine the stress conditions and compressi-
bility. Empirical values of the modulus number, published in the literature, are generally representative for normally
consolidated soils. In the present study, a concept is presented how the effect of stress history (pre-loading) on the modulus
number and the tangent modulus can be taken into account. The constrained modulus derived from CPTU and DMT
results are compared and show good agreement.

Keywords: in situ tests; cone penetration test; dilatometer; tangent modulus method; preconsolidation.

1 Introduction 2 Geotechnical Setting

The results of extensive field investigations, which The geology of the area is characterized by an almost
were carried out at the Bolivian Experimental Site for 100 m deep sedimentary river basin, created by the Piray
Testing Piles (B.E.S.T.), are reported. The site is located River and its tributaries. The upper about 20 m thick soil
24 kilometers North-east of the city of Santa Cruz de la deposit is composed of fine to medium sands with inter-
Sierra, Bolivia. In connection with the third Bolivian In- mittent layers of silt, clay or clayey sand. The stress his-
ternational Conferences on Deep Foundation (C.F.P.B.), tory of the soil deposit is complex, as it is affected by a
Technical Committee 212, “Deep Foundations” of the sedimentation-erosion-sedimentation process. The first
International Society for Soil Mechanics and Geotech- about 5 to 6 m are composed of loose to medium dense
nical Engineering (ISSMGE), a series of full-scale load- silt and sand, overlying a 6 to 7 m thick layer of silt and
ing tests was carried out on piles and a pile group. A par- sand. At about 11 m depth follow layers of silty clay on
ticular aspect of the field tests was to investigate the top of an about 1 m thick layer of compact sand. Below
potential effect of an enlarged pile base (Expander about 12 m depth, the soil deposit consists of layers of
Body), installed at the toe of different bored pile types. A compact to dense silty sand and loose sand. The ground-
series of pile installation and pile loading tests were car- water table varies seasonally and is located about 0.5 m
ried out. The results of these investigations are docu- below the ground surface.
mented in [1]. The scope of the pile testing program has Geotechnical foundation concerns in granular soils
been reported in [2]. (silt, sand, and gravel) are usually governed not by stabil-
In order to establish the geotechnical site conditions, a ity considerations, but by total and differential settlement
comprehensive field campaign was carried out compris- restrictions. The most important aspect of a settlement
ing a variety of in-situ tests. The results of the static and analysis is the selection of realistic input parameters. It is
dynamic penetration tests were compiled and analyzed in generally accepted that estimating settlement of granular
[3]. Also, different types of seismic test were carried out. soil is an approximated process. The geotechnical litera-
The results were reported in [4]. ture provides only limited guidance on how to estimate
The results of extensive cone penetration tests with the compressibility and stress history of granular soils
pore water pressure measurement (CPTU) and the flat from in-situ tests. A major reason for this is the difficulty
dilatometer (DMT) are presented in the present study. to obtain undisturbed samples that can be tested in the
Emphasis is on the determination of compressibility (soil laboratory. Thus, the compressibility of granular soils
modulus) and stress history of granular soils (sand and must either be estimated based on empirical data, or the
silt), which is required for settlement analyses. A reliable interpretation of in-situ tests. In the absence of reliable
method of settlement analysis is the tangent modulus data, over-simplified concepts are frequently used to as-
method, which will be described in detail. A critical step sess both soil compressibility and stress history. When
in settlement analyses is the selection of realistic input deformation properties of granular soils are determined,
parameters, such as preconsolidation stress and modulus. based on in-situ tests, it is nevertheless important to use
a consistent approach based on stringent concepts.
 The accuracy of settlement calculations is usually not In normally consolidated sand, K0, be estimated from
determined by the employed analytical method used, but the relationship proposed in [8]
by the selection of relevant input parameters. The focus
𝐾0 ~ 1 − 𝑠𝑖𝑛 (𝜙′) (5)
of the present paper is to outline a method for estimating
in-put parameters for settlement analysis of granular soils where ’= effective friction angle.
based on CPT and DMT. A typical value of K0 for uncompacted sand (‘ 33°)
would be 0.43. The effective friction angle of overcon-
3 Interpretation of CPT solidated (preloaded) sands can be estimated from the ex-
pression derived in [9]
The CPT - and variations thereof, such as the CPTU
(with pore water pressure measurement) or the SCPT 𝑞𝑡
⁄𝜎𝑟
(with seismic downhole test) is a widely used method for 𝜙′ = 17.6° + 11.0 ∙ 𝑙𝑜𝑔 ( ′
) (6)
the assessment of strength and deformation properties of √𝜎𝑣⁄𝜎
𝑟
granular soils. An important advantage of the CPTU is
that it measures three independent parameters: cone re- where ’= effective friction angle
sistance, qc, sleeve resistance, fs, and pore water pressure, qt = stress adjusted cone resistance
u. The measured cone resistance, qc, is usually corrected 𝜎𝑟 = reference stress (100 kPa).
for pore water pressure, u, In overconsolidated soils, it is important to estimate
the preconsolidation stress, 𝜎𝑝′ . Mayne et al. [6] have pro-
𝑞𝑡 = 𝑞𝑐 + 𝑢 (1 − 𝑎𝑐 ) (1)
posed the following general equation for all types of nat-
where: qt = cone resistance corrected for pore-pressure ural soils, including sands, silts, clays, and mixed soil
on the cone shoulder and ac = net area ratio (the sub- types
script “c” was added to prevent confusion with the mod- ′
ulus factor, a, see below). 𝜎𝑝′ = 0.33 (𝑞𝑡 − 𝜎𝑣0 )𝑚 (7)
where qt = stress corrected cone resistance
3.1 Stress adjustment 𝜎𝑣0 = vertical total stress
m' = grain size parameter, which increases with
The cone resistance is influenced by depth and, thus, fines content and decreases with mean grain
by the effective confining stress. Massarsch [5] proposed size (clean quartz sands: m' ≈ 0.72, silty sands:
a stress adjustment factor, CM, to take into account the m' ≈ 0.8, clays: m' ≈ 1.0).
effect of mean effective stress, σ’m, on the cone resistance It should be pointed out that although Eq. (7) gives
measured in sandy soils, Eq. (2) only an approximate estimate of the preconsolidation
𝜎𝑟 0.5 stress, it is recommended to use an approximate value ra-
𝐶𝑀 = ( ′
𝜎𝑚
) (2) ther than neglecting the preconsolidation effect. The
overconsolidation ratio, OCR, can now be calculated
where CM = stress adjustment factor ≤ 2.5 from the preconsolidation stress, 𝜎𝑝′ , Eq. (7) and the ver-
σr = reference stress = 100 kPa
tical effective stress, σ'v
σ’m = mean effective stress.
The stress-adjusted cone resistance, qcM can now be 𝜎𝑝′
𝑂𝐶𝑅 =
𝜎𝑣′
(8)
calculated, Eq. (3)
𝜎 0.5 Massarsch and Fellenius [10] have suggested the fol-
𝑞𝑐𝑀 = 𝑞𝑐 𝐶𝑀 = 𝑞𝑐 ( ′𝑟 ) (3) lowing relationship between the horizontal effective
𝜎 𝑚
stress ratio (K1/K0) and the overconsolidation ratio, OCR,
It should be noted that the stress adjustment using the
𝐾1
mean effective stress introduces the stress history of the = 𝑂𝐶𝑅𝛽 (9)
𝐾0
soil deposit. This is in contrast to the frequently used ver-
tical effective stress [6, 7]. In the opinion of the authors where K0 = horizontal stress coefficient of normally
it is preferable to use an approximate procedure to esti- consolidated soils
mate the mean effective stress by an approximate proce- K1 = horizontal stress coefficient of overconsol-
dure rather than neglecting the effect of stress history. idated (compacted) soils
 = empirical coefficient.
3.2 Stress history The horizontal stress coefficient is necessary for esti-
mating the mean effective stress, σ’m. Based on calibra-
The stress-adjusted cone resistance, qcM, reflects soil tion chamber (CC) tests, [11] recommended β = 0.42 and
behavior independent of depth and thus reflects funda- [12] β = 0.45. In [13] a range from 0.38 to 0.44 for me-
mental soil behavior. The method relies on knowledge dium dense sand is suggested. In [14] a conservative
(or assumption) of the mean effective stress (i.e. horizon- value,  ≈ 0.48 is proposed. Thus, the horizontal stress
tal effective stress), which is expressed in Eq. (4) coefficient, K1, can be estimated from the following rela-
′ 1+2𝐾0 tionship, accounting for the stress history
𝜎𝑚 = 𝜎𝑣′ ( ) (4)
3
𝐾1 = 𝐾0 𝑂𝐶𝑅 0.48 (10)
where σ'm = mean effective stress
σ'v = vertical effective stress
K0 = at-rest earth stress coefficient.
 Now, the mean effective stress, σ'm, can be calculated j = stress exponent.
from Eq. (4), using K0 for normally consolidated soils or Integrating Eq. (14) yields the following general rela-
K1 for overconsolidated soils. tionship for determining the strain, , of a soil layer re-
sulting from an increase of stress
4 Interpretation of DMT ′ 𝑗 𝑗
1 𝜎𝑣1 𝜎′
𝜀=
𝑚𝑗
[(
𝜎𝑟
) − ( 𝜎𝑣0 ) ] (15)
A relatively recent in-situ method is the flat dilatome- 𝑟

ter, DMT, introduced by Marchetti [15]. Guidelines for

where ’vo = vertical effective stress prior to loading
the DMT equipment and interpretation of test results
have been issued by ISSMGE Technical Committee 16 ’v1 = vertical effective stress after loading.
[15]. For a detailed description of the DMT, recent devel-
opments in data interpretation, and practical application 5.1 Modulus Number from Empirical Data
of results, refer to the geotechnical literature, e.g., 3rd The most important aspect of the tangent modulus
DMT Conference Proceedings [16]. method is the selection of realistic input parameters, that
The test procedure is to advance the dilatometer blade is, the stress exponent, j, and the virgin modulus number,
into the ground. Readings are taken at depth intervals of m and re-loading modulus number, mr.
200 mm by inflating a membrane by 1.1 mm and taking Based on data by [18], values for m and j according to
pressure readings. These "raw" pressure readings are cor- soil type of coarse-grained soil were published in [19].
rected and subsequently converted into two pressure val- Table 1 shows the typical range and average value of j
ues, p0, and p1. A key characteristic, which distinguishes and m, respectively. Note that two cases of the stress ex-
the DMT from other in-situ methods, is its ability to ponent (0.5 and 1.0) are given for granular soils.
measure parameters that reflect the stress conditions in Table 1. Typical stress exponent and modulus numbers for granular
the horizontal direction. This fact has important conse- soils [19].
quences for the evaluation of soil stiffness (modulus), Soil Type
Stress exponent Range Average
which is affected by stress history. While the CPT strains j m m
the soil to failure, the DMT strains the soil to an interme- Till, very
1 1,000 – 300 650
dense to dense
diate level, thus more realistically reflecting soil defor- Gravel 1 400 - 40 220
mation properties. Sand
From the derived p0- and p1-values, the following dense 1 400 – 250 325
DMT index parameters are calculated compact 250 – 150 200
loose 0.5 150 – 100 125
𝑝1 −𝑝0 Silt
𝐼𝐷𝑀 =
𝑝0 −𝑢0
(11) dense 1 200 – 80 140
𝑝0 −𝑢0 compact 80 – 60 70
𝐾𝐷 = ′
𝜎𝑣0
(12) loose 0.5 60 – 40 50

𝐸𝐷 = 34.7(𝑝1 − 𝑝0 ) (13) In a more recent paper, Janbu [20] updated typical val-
where IDM = material index (nomenclature modified in ues of the modulus number, m, for normally consolidated
order to avoid confusion with the density index, silt and sand (j = 0.5). In [20] the modulus number was
I D) presented as a function of porosity, n, herein converted to
KD = horizontal stress index the more widely used void ratio, e. Figure 1 shows the
ED = dilatometer modulus modulus number, m, derived as a function of void ratio,
u0 = hydrostatic pore water pressure e, for silt and sand and different degrees of density. Also
indicated is the approximate range (and lower/upper
’v0 = vertical effective stress
boundaries) of modulus numbers for the respective soil
p1 = stress applied at start of expansion
category (sand and silt), according to the classification
p2 = stress applied at end of expansion.
used in Table 1.
5 Tangent modulus method 600
V. DENSE DENSE COMPACT LOOSE V. LOOSE

The tangent modulus method for settlement analyses 500

was first proposed by Ohde [17] and Janbu [18] and is
MODULUS NUMBER, m

described in detail in [19]. The tangent modulus (con- 400 SAND

strained modulus) is the ratio between a change of stress
and the change of strain induced by that stress change 300

(1−𝑗) SILT
𝑑𝜎 𝜎′ 200
𝑀𝑡 = = 𝑚 𝜎𝑟 ( 𝑣) (14)
𝑑𝜀 𝜎𝑟
100
where Mt = tangent modulus
d = change of stress 0
d = change of strain 0.35 0.45 0.55
VOID RATIO, e
0.65 0.75 0.85

m = modulus number (dimensionless)

r = reference stress (equal to 100 kPa) Figure 1. Typical values of modulus numbers for normally consoli-
'v = vertical effective stress dated sand and silt [20].
5.2 Effect of pre-loading on modulus The modulus factor, a, reflects soil type and varies within
a relatively narrow range for each soil category, as indi-
number cated in Table 2.
The empirical values of the modulus number proposed Table 2. Modulus factor, a, for different soil types [10].
in Table 1 serve as a guidance for different soil types. Soil Type Modulus Factor, a
Figure 1 shows typical values of the modulus number for Silt, organic soft 7
normally consolidated sand. However, it is well known Silt, loose 12
Silt, compact 15
that stress history (pre-loading) affects the modulus num- Silt, dense 20
ber. As a result, the modulus number can increase signif- Sand, silty loose 20
icantly. Unfortunately, only limited factual information Sand, loose 22
is available in the literature, quantifying the effect of pre- Sand, compact 28
Sand, dense 35
loading. Massarsch and Fellenius [14] re-analyzed results Gravel, loose 35
of laboratory compression tests on sand reported in [17]. Gravel, compact 40
The modulus number, m was measured during virgin Gravel, dense 45
loading as well as during unloading, mu. The modulus
number ratio, mu/m, as a function of the virgin modulus An important benefit of determining the modulus
number, m is shown in Fig. 2 number from CPT data is the fact that the cone resistance
reflects the variation of soil type and stiffness with depth
continuously as opposed to determining soil type and soil
layer boundaries from intermittent sampling.

5.4 Modulus number from DMT

The DMT has become standardized. Marchetti et al.
[16] suggested that a vertical, drained, constrained mod-
ulus, M, can be estimated from the dilatometer modulus,
ED, as follows.
𝑀 = 𝑅𝑀 𝐸𝐷 (19)
with IDM < 0.6: RM = 0.14 + 2.36 log KD
Figure 2. Modulus number ratio as a function of the modulus number IDM > 3: RM = 0.5 + 2 log KD
during virgin loading, based on [14]. 0.6 < IDM < 3: RM = RM,0 + (2.5 - RM,0) log KD
During unloading, the modulus number, mu, is signif- with
icantly higher than at virgin loading, m. The modulus RM,0 = 0.14 + 0.15 (IDM - 0.6)
number ratio increases with decreasing initial modulus if KD > 10: RM = 0.32 + 2.18 log KD
number. The correlation between the unloading ratio, if RM < 0.85: assume RM = 0.85.
mu/m and the virgin loading modulus number, m, shown where M = vertical, drained, constrained modulus;
in Fig. 2 can be expressed by the following equation RM is a correction factor based on empirical data
𝑚𝑢 [21]
= 225 𝑚−0.76 (17) IDM = material index
𝑚
KD = horizontal stress index
In very loose sand (m = 100), the unloading modulus ED = dilatometer modulus
ratio, mu/m  7. In case of an initially denser sand (m = u0 = hydrostatic pore water pressure.
300) the unloading modulus ratio is lower, mu/m  3. The modulus number, m, can then be estimated ac-
Thus, the effect of pre-loading should be considered, es- cording to Eq. (14). By rearranging terms, the following
pecially in loose, normally consolidated sand. relationship is obtained.
There is a difference between the unloading modulus
𝑗−1
and the reloading modulus. However, for most practical 𝑀𝑡 𝜎𝑣′
𝑚=
𝜎𝑟
(𝜎 ) (20)
purposes, this effect can be neglected. 𝑟

In the case of normally consolidated sand, assuming

5.3 Modulus number from CPT that j = 0.5, the modulus number, m, is obtained from
1 0.5
Massarsch [5] proposed a correlation between the 𝑚 = 𝑀𝑡 ( ) (21)
modulus number, m, for granular soils to the stress-ad- 𝜎𝑟 𝜎𝑣′
justed cone resistance, qcM, according to Eq. (3) For the case of compacted sand, assuming that j = 1,
𝑞𝑐𝑀 0.5 the following simple relationship is obtained
𝑚 = 𝑎(
𝜎𝑟
) (18)
𝑀𝑡
𝑚= (22)
𝜎𝑟
where m = modulus number
a = empirical modulus factor Thus, the modulus number, m, can be readily obtained
qcM = stress-adjusted cone resistance by dividing M (in units of kPa) by 100, i.e., the reference
r = reference stress (100 kPa). stress, r, (in units of kPa).
5.5 Stress Exponent was calculated according to Eq. (6) and is shown in Fig.
4b. The friction angle was used to estimate the horizontal
The stress exponent, j, in Eq. (14) defines the curva- stress coefficient, K0.
ture of the load-compression relation and is based on soil An important step in the assessment of compression
type and stress conditions, which are relatively easy to properties of soil deposits is the determination of the
estimate. Typical values of the stress exponent were rec- preconsolidation stress, 𝜎𝑝′ . Figure 5a shows the vertical
ommended in [19], as summarized in Table 1. For dense effective stress and the preconsolidation stress as
sand and gravel or glacial tills (overconsolidated soils), determined according to Eq. (7). It is interesting to note
the stress exponent is usually 1.0, which represents a lin- that the soil deposit is overconsolidated by a stress
ear response (elastic) to load. For loose silt and sand, j is margin of about 200 – 300 kPa, with the exception of an
typically 0.5, but decreases with decreasing grain size. intermediate layer between 3 and 6 m depth. This effect
Although j goes toward 0.25 in silty soils, in practice, it can also be detected from the inspection of the stress-
is usually satisfactory to assume j = 0.5 also here. adjusted cone resistance, cf. Fig. 4a.
The overconsolidation ratio, OCR, is shown in Fig. 5b.
6 Geotechnical investigations High OCR values are obtained close to the ground sur-
face but decrease with depth. Below about 3.5 m, OCR
In the following, a novel concept of interpreting the varies in the range of 2 – 4. However, at shallow depth,
results of CPTU and DMT is applied to the extensive data very high OCR values are obtained, due to the low verti-
from the B.E.S.T. investigations. The aim of this study cal effective stress. At shallow depth, OCR is very sensi-
was to determine geotechnical parameters important for tive to calculation errors. Figure 5 confirms the ad-
settlement analyses, such as soil compressibility (con- vantage of expressing preconsolidation in terms of the
strained modulus) and stress history (preconsolidation stress margin (difference between the preconsolidation
stress and overconsolidation ratio). stress and the vertical effective stress) rather than OCR.
The geotechnical testing program included both in- In the overconsolidated layers below about 3 m depth, the
situ and laboratory tests. The in-situ testing program stress margin is about 150 – 200 kPa and almost constant
comprised different types of tests. However, only the re- with depth.
sults of 15 CPTUs and 6 DMTs are included in the pre- In order to determine the constrained modulus from
sent study. Supplementary laboratory tests included: CPTU data, at first the modulus factor, a, according to
grain size distribution, water content, and plastic and liq- Table 2 needs to be estimated. This is usually done based
uid limits tests. For details, reference is made to [3, 4]. on inspection of the cone resistance and the friction ratio
The in-situ tests were carried out along an about 80 m according to Fig. 3a and c. The assumed values of the
long and 6 m wide area. Although the area was essentially modulus factor, a, and the modulus number determined
level, some height variations did occur, which affected to according to Eq. (16) are shown in Fig. 6a and b.
some extent the evaluation depth readings of the CPTU Based on the information shown in Fig. 5 (preconsol-
and DMT data. idation stress) and Fig. 6b (modulus number) it is possi-
The distance between the test points was 5 m. The lat- ble to carry out a conservative settlement analysis. How-
eral distance between the CPT and DMT in each test area ever, it should be pointed out that the modulus numbers
was 1.1 m. A limited number of tests had to be eliminated shown in Fig. 6b are for normally consolidated condi-
from the analysis due to malfunctioning of the measuring tions and are thus conservative. As has been pointed out
system. Although the tests were carried out to greater above, pre-loading (overconsolidation) has a significant
depth, this study is restricted to 12 m depth, where gran- effect on the modulus number and thus also on the con-
ular soils dominated. All raw data of the field investiga- strained modulus. The following procedure was used to
tions can be downloaded from the web site of the 3rd determine the modulus number taking into account the
C.F.P.B conference: http://www.cfpbolivia.com. pre-loading effect: a) select all data with a friction ratio <
1.5 (granular soils); b) for these data, choose values with
7 Results of CPTU investigations an estimated OCR > 4; c) apply Eq. (14) to calculate the
modulus number for the case of pre-loaded layers. Figure
The results of nine CPTUs were included in this study. 7a shows the modulus number values considering the
The CPTU data were evaluated according to the concepts pre-loading effect, cf. Fig. 6b.
outlined above. The cone resistance corrected for pore Finally, the constrained modulus has been calculated
water pressure effects, qt, the sleeve resistance, fs, the based on Eq. (14). In order to eliminate the effect of
friction ratio, FR, and the pore water pressure, u, are fluctuation in calculated values, and to facilitate
shown in Fig. 3. comparison with the constrained modulus determined
The results of the cone resistance and of the sleeve re- from DMT, an average of m-values was determined over
sistance show a relatively homogeneous soil profile, with a depth interval of +/- 0.2 m. The so determined con-
a deposit of loose silty sand (qt < 5 MPa) down to about strained modulus is shown in Fig. 7b. It is important to
6 m depth, followed by a stiffer layer with cone resistance note the significant effect of pre-loading on the estimated
values varying between 5 and 15 MPa. The large friction constrained modulus. A similar effect will probably also
ratio in the silty soil and the large excess pore water pres- occur in the fine-grained soil layers (silt and clay).
sure indicate the presence of fine-grained layers. However, for such soils it is recommended to determine
From the measured cone resistance, qc, the stress-ad- the pre-loading effect based on laboratory oedometer
justed cone resistance, qCM, was determined according to tests.
Eq. (3), as shown in Fig. 4a. The effective friction angle
 CONE RESISTANCE, qt (MPa) SLEEVE RESISTANCE, fs (kPa)
0 5 10 15 20 0 50 100 150 200
0 0

1 1

2 2

3 3

4 4

5 5
DEPTH (m)

DEPTH (m)
6 6

7 7

8 8

9 9

10 10

11 11

12 12
a) b)

FRICTION RATIO, FR (%) PORE WATER PRESSURE, u (kPa)

0 1 2 3 4 5 6 0 100 200 300 400 500
0 0

1 1

2 2

3 3

4 4

5 5
DEPTH (m)
DEPTH (m)

6 6

7 7

8 8

9 9

10 10

11 11

12 12
c) d)
Figure 3. Results 9 of cone penetration tests.
 CONE RESISTANCE, qCM (MPa) FRICTION ANGLE, f (o)
0 5 10 15 20 20 25 30 35 40 45 50
0 0

1 1

2 2

3 3

4 4

5 5

DEPTH (m)
DEPTH (m)

6 6

7 7

8 8

9 9

10 10

11 11

12 12
a) b)
Figure 4. Stress-adjusted cone resistance and derived friction angle.

EFFECTIVE STRESS (kPa) OVERCONSOLIDATION RATIO, OCR

0 100 200 300 400 0 5 10 15 20
0 0

1 1

2 2

3 3

4 4

5 5
DEPTH (m)
DEPTH (m)

6 6

7 7

8 8

9 9

10 10

11 11

12 12
a) Preconsolidation stress, vertical effective stress (blackline) b) Overconsolidation ratio
Figure 5. Determination of stress history.
 MODULUS FACTOR, a MODULUS NUMBER, m
0 10 20 30 40 0 200 400 600 800 1000
0 0

1 1

2 2

3 3

4 4

5 5

DEPTH (m)
DEPTH (m)

6 6

7 7

8 8

9 9

10 10

11 11

12 12
a) b)
Figure 6. Modulus factor and modulus number for normally consolidated soil, determined from CPTU data.

MODULUS NUMBER, m CONSTRAINED MODULUS, MCPT (MPa)

0 200 400 600 800 1000 0 20 40 60 80 100
0 0

1 1

2 2

3 3

4 4

5 5
DEPTH (m)
DEPTH (m)

6 6

7 7

8 8

9 9

10 10

11 11

12 12
a) Modulus number adjusted for pre-loading effect b) Constrained modulus including pre-loading effect
Figure 7. Modulus number and constrained modulus, determined from CPTU data, with adjustment to pre-loading effect according to Eq. (17).
 Extensive field investigations were performed on a
8 Results of DMT relatively homogeneous tests site, consisting of sand and
The results of all six performed DMTs were included silt. The complex geology of the river deposit resulted in
in this study. The analysis procedure started with deter- significant variations of geotechnical properties.
mining the two pressure values, p0, and p1. Due to the An important conclusion is that in spite of the rela-
variability of the soil deposit, relatively large variations tively uniform site conditions, significant variations in
of pressure values were obtained at some depths, Fig. 8. soil properties could be observed by CPTU and DMT
From the pressure readings, the material index, ID, and measurements. For important projects, it is recommended
the horizontal stress index, KD, were then obtained. The to compare results from different in-situ methods (CPTU
results are shown in Fig. 9. At some depths, the scatter of and DMT). Moreover, the number of tests and their loca-
data was significant, but the general trend agrees with the tions must be chosen reflecting the particular geological
CPTU measurements. The material index in Fig. 9a sug- setting.
gests the existence in some test points of more fine- A method is described how the modulus number, m,
grained (clay/silt) layers down to about 5 m depth. This can be derived from a stress-adjusted cone resistance. As
observation is confirmed by the relatively high excess this concept is based on tests on normally consolidated
pore water pressure readings in Fig. 3d). soils, significantly lower (conservative) values of the
The horizontal stress index is an important parameter constrained modulus can be obtained in overconsolidated
for the calculation of the DMT modulus, ED, and the con- soils.
strained modulus, Mt, cf. Eq. (19). The constrained mod- The modulus number is affected by pre-loading. This
ulus, Mt, and the modulus number determined according effect can be taken into account using Eq. (17). The un-
to Eq. (21) are shown in Fig. 10a and b, respectively. loading modulus number is about 3 to 7 times higher than
It is interesting to compare the modulus number deter- the modulus number during virgin loading, cf. Fig. 2. The
mined from CPT (Fig. 6b and 7a) with the DMT modulus pre-loading effect is higher in loose soils with a low vir-
number (Fig. 10b). Note the difference in scale between gin modulus number.
the figures. There is good agreement between the modu- The constrained modulus derived from CPT (Fig. 7b)
lus number determined from the CPTs and DMTs in the and measured by DMT (Fig. 10a) are in good agreement,
normally consolidated layer between 3 and 5 m depth. provided that the pre-loading effect is considered.
However, in the overconsolidated layers, the modulus
number derived from the DMT is significantly higher 10 References
than from the CPT for normally consolidated conditions. [1] C.F.P.B. Proceedings of the 3rd Bolivian International Conference
However, when the preconsolidation effect is accounted on Deep Foundation, (2017), available from conference web site:
for, the modulus number from CPT (Fig. 7a) is in good http://www.cfpbolivia.com/web/page.aspx?refid=163. (accessed:
agreement with the modulus numbers back-calculated 25 August 2019).
[2] Fellenius, B.H., Terceros H.M., and Massarsch, K.R. “Bolivian
from DMT. This comparison shows the importance of Experimental Site for Testing”. Proceedings, 3rd Bolivian Interna-
adjusting the modulus number for pre-loading effect ac- tional Conference on Deep Foundations, Santa Cruz de la Sierra,
cording to Eq. (17) as shown in Fig. 2. Bolivia, April 27-29, 2017. Vol. 2, Publisher: Omnipress, Madi-
son WI, USA: pp. 3–31.
[3] Agaiby, S. and Mayne, P.W. “Interpretation of geotechnical pa-
9 Conclusions rameters using in-situ data for Bolivian Experimental Site for
Testing (B.E.S.T.)”, Proceedings, 3rd Bolivian Conference on
Calculation of settlement in granular soils is an im- Deep Foundations, Santa Cruz de la Sierra, April 27-29, 2017.
portant task as the design of foundations on such soils Vol. 2, Publisher: Omnipress, Madison WI, USA: 69–89.
[4] Massarsch, K.R., Perez, N., Terceros M. A., and Terceros H.M.
usually is governed by total and differential settlement. “Interpretation of Seismic Tests. Proceedings”, 3rd Bolivian Inter-
However, even when employing sophisticated analytical national Conference on Deep Foundations, Santa Cruz de la Si-
methods, often very crude assumptions are made when erra, Bolivia, April 27-29, 2017. Vol. 2, Publisher: Omnipress,
estimating soil compressibility and stress history. The Madison WI, USA: 47–67.
[5] Massarsch, K.R. “Settlement analysis of compacted fill”. Proceed-
concept of the tangent modulus method is simple and ings 13th ICSMFE, New Delhi, India, January 5-10, 1994, Vol. 1,
transparent and can be applied in all soil types. Publisher: CRC Press; 1 edition (January 1, 1994), pp. 325-328.
In granular soils, the most reliable in-put data – alt- [6] Mayne, P.W., Coop, M.R., Springman, S., Huang, A-B., and
hough never perfect – can be derived from in-situ tests, Zornberg, J. “State-of-the-Art Paper (SOA-1): Geomaterial
Behavior and Testing”. Proceedings 17th ICSMFE 2009
such as the CPTU or the DMT. It is important to note the (Alexandria, Egypt), Vol. 4, Millpress/IOS Press Rotterdam, the
difference in soil testing between the two methods. The Netherlands: 2777–2872.
CPTU causes soil failure during probe penetration. Thus, [7] Robertson, P. K., and Campanella, R. G. “Interpretation of Cone
deformation properties are only indirectly related to the Penetration Tests - Part I (Sand)”. 2011. Canadian Geotechnical
Journal 20(4):718–733. DOI: 10.1139/t83-078.
measured cone resistance. The advantage, however, is [8] Jáky, J. “Pressure in silos”. Proc. 2nd ICSMFE. Rotterdam, June
that a relatively continuous soil profile is obtained. In the 21–30, 1948, Vol. 1. Publisher: ISSMGE, London, UK, pp. 103–
case of a DMT, the deformation properties of the soil ad- 107.
jacent to the blade can be measured without causing soil [9] Kulhawy, F.H., and Mayne, P.W. “Manual on Estimating Soil
Properties for Foundation Design”, Report EL-6800, 1990. Elec-
failure. Another important advantage of the DMT is that tric Power Research Institute, Palo Alto, 306 p.
it measures soil properties in the lateral direction, which [10] Massarsch, K.R., and Fellenius, B.H. “In situ tests for settlement
preserve the effect of stress history more realistically than design of compacted sand”. Proceedings of Institution of Civil En-
the CPTU. gineers – Geotechnical Engineering, ISSN 1353-2618 | E-ISSN
1751–8563, Volume 172, Issue 3, June, 2019, pp. 207–217
https://doi.org/10.1680/jgeen.18.00046.
 PRESSURE, p0 (kPa) PRESSURE, p1 (kPa)
0 200 400 600 800 1000 0 500 1000 1500 2000
0 0

1 1

2 2

3 3

4 4

5 5

DEPTH (m)
DEPTH (m)

6 6

7 7

8 8

9 9

10 10

11 11

12 12
a) b)
Figure 8. Pressure readings obtained from six DMT tests.
MATERIAL INDEX, ID HORIZONTAL STRESS INDEX, KD
0,1 1 10 0 5 10 15 20
0 0
CLAY SILT SAND
1 1

2 2

3 3

4 4

5 5
DEPTH (m)

DEPTH (m)

6 6

7 7

8 8

9 9

10 10

11 11

12 12
a) b)
Figure 9. Material index and horizontal stress index from DMT measurements.
 CONSTRAINED MODULUS, MDMT (MPa) MODULUS NUMBER, m
0 20 40 60 80 100 0 500 1000 1500 2000
0 0

1 1

2 2

3 3

4 4

5 5
DEPTH (m)

DEPTH (m)
6 6

7 7

8 8

9 9

10 10

11 11

12 12
a) b)
Figure 10. Constrained modulus and modulus number determined from DMT.

[11] Schmertmann J.H. “Measure and use of the in-situ lateral stress”. [16] Marchetti, S. “Some 2015 Updates to the TC16 DMT Report".
The Practice of Foundation Engineering, A Volume Honoring Jorj Proc. 3rd Intern. Conf. on the Flat Dilatometer, DMT'15. Rome,
O. Osterberg”. R.J. Krizek, C.H. Dowding, and F. Somogyi, Eds. Italy, (2015), pp. 43-65.
Department of Civil Engineering, Technological Institute, North- [17] Ohde, J. „Grundbaumechanik“ (Ground engineering mechanics),
western University, Evanston, USA, (1985), pp. 189-213. Publisher: A. V. Hütte, Berlin. Band 3, 27. Auflage, (1951), pp.
[12] Lunne, T. and Christophersen, H.P. “Interpretation of cone pene- 886-902.
trometer data for offshore sands”. Proceedings of the Offshore [18] Janbu, N. “Soil compressibility as determined by oedometer and
Technology Conference, Richardson, Texas, USA, (1983). Paper triaxial tests”. III. European Conference on Soil Mechanics and
No. 4464. Foundation Engineering, ECSMFE, Wiesbaden, October 15 -18,
[13] Jamiolkowski, M., Ghionna, V.N., Lancelotta, R. and Pasqualini, 1963, Vol. 1, pp. 19-25, and Discussion contribution, Vol. 2, Pub-
E. “New correlations of penetration tests for design practice”. lisher: Deutsche Gesellschaft für Erd-und Grundbau e.V., pp. 17-
Proc. of Symposium on Penetration Testing. ISOPT-1. Edited by 21.
J. de Ruiter, Orlando, USA, March 22-24, 1988, A.A. Balkema, [19] Canadian Foundation Engineering Manual (CFEM), Canadian
Rotterdam, the Netherlands, pp. 263–296. Geotechnical 3rd Edition (1992), Canadian Geotechnical Society.
[14] Massarsch, K.R. and Fellenius, B.H. “In situ tests for settlement BiTech Publishers Ltd. Richmond, BC, Canada, 512 p.
design of compacted sand”. Proceedings of the Institution of Civil [20] Janbu, N. “Soil models in offshore engineering”. Géotechnique
Engineers – Geotechnical Engineering, (2018) 172(3) 207–217. 35(3) 1985. pp. 241-281. doi.org/10.1680/geot.1985.35.3.241
[15] Marchetti S., Monaco P., Totani G., and Calabrese M. “The Flat [21] Marchetti, S. “In-situ tests by flat dilatometer”. ASCE Journal of
Dilatometer Test (DMT) in Soil Investigations”. Proc. In situ Geotechnical Engineering Division 106(GT3), (1980), pp. 299-
2001, Int. Conf. On In situ Measurement of Soil Properties, Bali, 321.
Indonesia, May 2001, 41 pp. (Reprinted in Proc. DMT 2006,
Washington D.C., USA).