Последња измена: 2020-04-07 19:47:29

Determination of the presence of microstructure in a soil using a seismic

dilatometer
Dušan Berisavljević, Zoran Berisavljević

Дигитални Репозиторијум Рударско-геолошког факултета Универзитета у Београду

[Др.РГФ]

Dušan Berisavljević, Zoran Berisavljević | Determination of the presence of microstructure in a soil using a seismic
dilatometer | Bulletin of Engineering Geology and the Environment | 2018

DOI: 10.1007/s10064-018-1234-5

URI: dr.rgf.bg.ac.rs/s/repo/item/5073

Дигитални репозиторијум Рударско-геолошког факултета The Digital repository of The University of Belgrade
Универзитета у Београду омогућава приступ издањима Faculty of Mining and Geology archives faculty
Факултета и радовима запослених доступним у слободном publications available in open access, as well as the
приступу. Претрага репозиторијума доступна је на employees' publications. The Repository is available at:
dr.rgf.bg.ac.rs dr.rgf.bg.ac.rs
 D. Berisavljević, Z. Berisavljević

et al. 2006; Yun and Santamarina 2005) and is more affected

by the presence of microstructure than penetration resistance.
This is mainly due to the fact that at small shear strain levels
(less than 10−5–10−6) the bonds at interparticle contact are pre-
served (Diaz-Rodriguez and Santamarina 2001). The different
level of sensitivity makes it possible to combine measured or
interpreted large-to-medium strain parameters such as cone pen-
etration resistance (qc) or horizontal stress index (KD) with small
strain parameter G0 in order to reduce the uncertainty in the
application of the common correlations used to derive soil pa-
rameters. In order to do so, it is important to identify the presence
of microstructure in a soil. A useful chart for identifying the
presence of cementation in a soil was proposed by Cruz et al.
Fig. 1 Evolution of microstructure in normally consolidated clay (2012). Based on the large international database for sedimentary
(Leroueil and Vaughan 1990) σ’vI - effective stress corresponding to void
soils and a limited number of datapoints for granitic residual
ratio eI σ’vB - apparent preconsolidation stress after secondary compres-
sion (ageing) σ’P - apparent preconsolidation stress after cementation eI- soils, Cruz et al. (2012) interpreted that cemented residual soils
void ratio after deposition e0 - void ratio after secondary compression can be discerned from sedimentary low or non-cemented soils
based on consideration of the G0/ dilatometer modulus (ED)
considered to give an apparent increase in preconsolidation versus material index (ID) or G0/ DMT constrained modulus
stress from σvI’ to σvB’. If cementation due to, for example, (MDMT) versus KD correlations. Another useful chart which
deposition of carbonates, iron, and aluminum oxides occurs, may have a potential to identify the presence of microstructure
the apparent preconsolidation stress increases from σvB’ to in soil was proposed by Rivera-Cruz et al. (2012). The chart
σP’. Suction also influences the apparent preconsolidation consists of seven zones representing different soil behavior types,
stress, where for larger suction (lower saturation degree) the as shown in Fig. 2. The major drawback of this chart is the need
elastic stress range increases and the soil behaves more stiffly to measure the p2 pressure, which significantly extends the time
(e.g., Alonso et al. 1990; Jotisankasa, et al. 2007). This appar- needed to perform the test.
ent preconsolidation stress or yield stress represents the point Both the previously mentioned charts consider either the
where irreversible strains start to develop upon either a reduc- G0/ED or G0/MDMT ratio together with one of the intermediate
tion in suction or an increase in the net total stress. parameters in order to determine the presence of microstruc-
Throughout this paper, the term microstructure is used in- ture. Robertson (2015) presented a DMT-based soil behavior
stead of structure and it is meant to describe soils that have
Bunusual characteristics^ compared with Bideal soils^ that
have little or no microstructure (Robertson 2015). The major
causes for the development of microstructure in soils include
cementation, aging, stress, and strain history.

Soil behavior type (SBT) charts

for microstructure identification

In situ penetration tests, such as the cone penetration test (CPT)

or dilatometer test (DMT), are frequently used to determine the
in situ mechanical response of a soil to cone penetration during
CPT testing, or blade penetration and membrane expansion dur-
ing DMT testing. Penetration resistance in sands is controlled by
the relative density, effective stress state, compressibility and, to
a lesser degree, by cementation and aging (Jamiolkowski et al.
2001; Robertson 2012). On the other hand, due to the larger
number and area of grain-to-grain contact, G0 tends to increase
with increasing cementation in soil (Fernandez and Santamarina
2001; Rinaldi et al. 1998; Rinaldi and Santamarina 2008; Fig. 2 Seismic dilatometer (SDMT)-based soil behavior type chart
Schnaid 2005; Schneider and Moss 2011; Viana de Fonseca (Rivera-Cruz et al. 2012)
Determination of the presence of microstructure in a soil using a seismic dilatometer

type chart that considers intermediate DMT parameters ID and detect soils that behave differently from Bideal^ soils, i.e., soils
KD with the addition of the contours of the G0/σv’ ratio. In this without significant presence of microstructure. Figure 4 shows
paper the DMT-based soil behavior type (SBT) chart present- a comparison between the measured and estimated (from cor-
ed by Robertson (2015) is extended and validated for various relations shown in Fig. 3) Vs profiles for five sites.
soil types. The presence of microstructure is estimated indirectly based
This research is based on two concepts. First is the mea- on geological evidence and the history of the deposit, and di-
sured to estimated G0 ratio (MEGR), which takes into consid- rectly based on core sample inspection and laboratory test re-
eration the measured to estimated G0 (G0/σv’) ratio for micro- sults when available. During drilling, visual inspection and HCl
structure identification. This concept is similar to the mea- acid was used on core samples in order to check the presence of
sured to estimated velocity (Vs) ratio (MEVR) used to correct carbonates. A brief description and estimation of the level of
liquefaction resistance for aged sands presented by Andrus microstructure for the first site (Vrbas I) located near the town
et al. (2009) and Hayati and Andrus (2009). The second con- of Vrbas is given here. The topmost layer (first meter) is desic-
cept is based on the idea presented by Robertson (2015) at the cated crust, and according to the Unified Soil Classification
recently held DMT’15 conference. The DMT performance System (USCS) it is classified as CL with a degree of saturation
and derivation of parameters can be found in Marchetti of 90%. Below the desiccated crust, 1.0 m of aquatic loess is
(2001). found underlain by 2.0 m of saturated clayey silt of low plas-
ticity. Below the clayey silt, silt of low plasticity, 2.5 m thick, is
found. The water level is at a depth of 4.0 m from the ground
Estimation of shear wave velocity (VS) surface. Beneath the low plastic silts, fine silty sand with 15%
from seismic dilatometer (SDMT) correlations fines is encountered. The thickness of this layer is on average
1.8 m and it is underlain by a layer of fine-to-medium coarse,
When no seismic data are available during geotechnical inves- quartz, alluvial sand, which is normally consolidated with an
tigations, it is possible to estimate shear wave velocity (Vs) MDMT/qc ratio of approximately 6. Its overall thickness is large
from the seismic dilatometer (SDMT)-based correlations de- and it was not determined during field investigations.
rived by Marchetti et al. (2008), which are shown in Fig. 3. From Fig. 4a it can be observed that the estimated velocity
Most empirical correlations for penetration tests are based on closely follows the measured velocity except at depths from
case histories from soils with little or no microstructure 6.0 to 8.0 m, where the fine silty sand layer is encountered. In
(Robertson 2015); thus, when using such correlations engineers this layer, the average relative error between the measured and
need to be cautious since G0 in highly structured soil can be estimated Vs is 40%, which is twice the average error reported
underestimated. Subsequent validation of these correlations has by Marchetti (2014). It can be assumed that this difference is
shown that the relative error in the estimated Vs (from ID, KD, due to the presence of microstructure in the fine silty sand.
and MDMT) is on average 20% (Marchetti 2014). As previously Two possible factors may have caused microstructure to de-
mentioned, the measured Vs (G0) is highly sensitive to the velop in this layer. Firstly, it was deposited in a slow-moving
presence of microstructure in soil. Thus, comparison between or stagnant water environment, which is evident from the
the measured and estimated velocity can be directly used to presence of organic matter and finer particles than the sand
layer below, which is typical river bed sediment of the Danube
River. Secondly, it is believed that precipitation of carbonates
from upper layers caused cementation of the soil particles.
Figure 4b shows the measured and estimated Vs for another
site located near the town of Vrbas (Vrbas II site). The location
of this site is approximately 3.5 km from the Vrbas I site. It
consists of alluvial deposits. The upper 9 m consist of silts
interbedded with fine silty sands. The lower soil layer consists
of predominantly quartz sand, which belongs to the same geo-
logical unit as the sand described for the previous site. At this
site, no significant presence of microstructure is found. From
Fig. 4b, it is observed that the measured and estimated veloci-
ties closely match, except where local peaks in the estimated
velocities are encountered. These peaks are usually difficult to
detect from the measured velocity, since it is average in 50 cm
while KD and ID are determined at 20 cm depth intervals.
Fig. 3 Ratio of G0/MDMT vs. KD (overconsolidation ratio [OCR]) for Figure 4c presents the results of measured and estimated Vs
various soil types (Monaco et al. 2009). G0 small strain stiffness for the Loznica site, where terrace sediments (upper 10.0 m)
 D. Berisavljević, Z. Berisavljević

Fig. 4 Comparison between the measured and estimated shear wave velocity (Vs) profiles: (a) Vrbas I site, (b) Vrbas II site, (c) Loznica site, (d)
Obrenovac site, and (e) Zemun loess
Determination of the presence of microstructure in a soil using a seismic dilatometer

are underlain by stiff marls. The water level is at a depth of – the larger the difference between measured and estimated
10.0 m. The measured and estimated velocities differ signifi- velocity, the greater is the presence of microstructure in
cantly, particularly below the depth of 4.0 m. Below this depth, soil; and
field investigations of terrace sediments have confirmed the – correlations shown in Fig. 3 should be used with caution
presence of organic matter and carbonates. The influence of when there is an indication that the soil is structured.
the suction on the measured velocities and DMT pressures in
the layers above the water table is not known, but it is believed When SDMT data for gravels exist, the reverse methodol-
that it can be significant. The average relative error in the esti- ogy may apply. It is possible to estimate KD from thrust mea-
mated velocity for marls is more than 25%. This is an expected surements (FDMT) at the surface using the relation proposed by
result for marl, which is a soil with a significant presence of Campanella and Robertson (1991) between the effective lift-
microstructure due to cementation. It is worth noting that dur- off pressure (p0-u0) and FDMT. Then, the measured Vs can be
ing testing of these marls a limit pressure of 80 bars was used to derive G0 by introducing the mass density (which is
reached on the high pressure gauge of the DMT control unit. usually not difficult to estimate correctly). Knowing both KD
It could be postulated that the high effective stress state reduces and G0, it is possible to estimate MDMT from Fig. 3 (for ID >
the effect that cementation has on the measured Vs (Rinaldi 1.8). It is believed that this approach is more appropriate for
et.al. 1998; Yun and Santamarina 2005; Pestana and Salvati the estimation of MDMT in gravels, even if correlations be-
2006; Rinaldi and Santamarina 2008) and that the difference tween (p0-u0) and FDMT and G0/MDMT and KD are derived
between the measured and estimated Vs could be expected to be for sands. This approach is supported by the fact that the total
even greater in less overconsolidated marls. In order to confirm thrust (blade penetration resistance qD) increases when the
that, the results for another site (Obrenovac) where marls were blade is penetrating through gravels, while p0 decreases com-
tested are shown in Fig. 4d. The ground profile at this site pared to sands in the same geological unit. This approach is
consists of an 8.0 m thick alluvial deposit underlain by marl. strongly influenced by the mean grain size diameter (D50).
From the figure it can be seen that the measured and estimated Nevertheless, the proposed approach needs further
velocities diverge more rapidly at the depth where marls are investigation.
encountered (approximately 8.0 m). This confirms that the
measured velocities are highly influenced by the presence of
microstructure in the soil. The difference in the measured and Microstructure-soil behavior type (M-SBT)
estimated velocities from 6.0 to 8.0 m is possibly due to the fact chart
that at those depths the tests were performed in sandy gravels
where the use of DMT may not be suitable (Marchetti et al. Figure 5 shows the proposed microstructure-soil behavior
2001). The derived DMT parameters (MDMT and KD) thus yield type (M-SBT) chart.
a poor prediction of Vs in gravelly soils. Figure 4e shows the On the chart, four broad groups of soil behavior are dis-
velocity profile obtained in loess. A description of this site can tinguished: FC—fine-grained contractive, FD—fine-
be found in Berisavljevic et al. (2014 and 2015). By its nature, grained dilative, CC—coarse-grained contractive, and
this loess is highly structured. The measured Vs profile is ap- CD—coarse-grained dilative. These groups differentiate
proximately constant with depth, with no difference between soil types that differ in their mechanical response to blade
the buried soil and loess horizons. On the other hand, the penetration and membrane expansion during a DMT. For
estimated velocity profile gradually increases from the ground instance, when testing highly overconsolidated clays, the
surface and has several marked peaks. Closer inspection of the tendency of the clay to dilate is restricted due to low perme-
correlation used to estimate Vs reveals that these peaks are a ability, which can cause negative (lower than hydrostatic)
consequence of the greater difference between p1 and p0, i.e., a pore pressures to develop, thus increasing the effective
larger DMT modulus ED. It was found by Berisavljevic et al. stress in the soil and mobilizing the frictional shearing re-
(2014) that p0 (KD) is very low and consequently any variation sistance. A soil with this response will fall into the FD zone.
in p1 may produce a significant difference in ED and thus Based on the value of ID, the soil can be classified as fine-
MDMT, which is used to derive G0 from the correlations grained (ID < 0.6) or coarse-grained (ID > 1.8). Fine-grained
shown in Fig. 3. A much better prediction of Vs is obtained soils would have an undrained response during a DMT,
when the correlation proposed by Berisavljevic et al. (2014) is while coarse-grained soils would have a drained response
used, as shown in Fig. 4e. during a DMT. Partial drainage may take place in transition-
The observed results indicate that: al soils, i.e., in the intermediate permeability range (0.6 <
ID < 1.8), as suggested by Schnaid and Odebrecht (2015).
– correlations for estimating Vs from DMT results are de- The boundary between dilative and contractive behavior is
rived for soils without a significant presence of proposed by Robertson (2012) for soils with little or no
microstructure; microstructure. For fine-grained soils, the boundary
 D. Berisavljević, Z. Berisavljević

Fig. 5 Proposed microstructure-

soil behavior type (M-SBT) chart

between dilative and contractive behavior at large strains in Fig. 3 and correlations for MDMT as proposed by Marchetti
occurs when the over consolidation ratio (OCR) ≈ 4. (1980) and Marchetti et al. (2001). The contours of the G0/σv’
Based on Marchetti et al. (2001), the correlation value of ratio are plotted on the M-SBT chart shown in Fig. 5.
OCR = 4 corresponds to KD ≈ 5. For coarse-grained soils,
the boundary between dilative and contractive behavior at
large strains occurs for state parameter ψ ≈ −0.05 Validation of M-SBT chart
(Robertson 2012). Derivation of the state parameter to be
used for constructing the boundary between dilative and Figures 6, 8, and 9 show results obtained in various soil types
contractive behavior for coarse-grained soils is a two-step plotted on the M-SBT chart. On each chart a brief description
process. First, the state parameter is derived from CPT- of the tested soil is included. Minimum and maximum values
based correlations with clean sand equivalent normalized as well as the arithmetic mean (given in parentheses) of the
cone resistance (Qtn,cs) (Robertson 2010). Secondly, the link measured G0/σv’ ratio for the particular soil are given. The
between KD and Qtn,cs is used with the Qtn,cs – ψ correlation arithmetic mean and the range of values give one possible
(Robertson 2012) to obtain the following equation: way to compare the measured and estimated data for a partic-
ular soil. The stricter and more precise approach would be to
compare the measured and estimated values for each
Ψ ¼ 0:56−0:33∙logð25∙K D Þ ð1Þ
datapoint, but this is less convenient for presentation in graph-
ical form when a large amount of data are evaluated. It should
In Eq. 1, KD = 3 yields ψ ≈ −0.05, which defines the con- be mentioned that G0 evaluated from SDMT is obtained from
tractive–dilative boundary for coarse-grained soils. As sug- a measured Btrue interval^ velocity (every 0.5 m), so the
gested by Robertson (2015), this boundary is not defined by equivalent value of ID and KD should also be taken as an
a single value of KD for all soils. It should be mentioned that average of three successive readings at a depth corresponding
the contractive–dilative boundary is linearly extrapolated with to the G0 evaluation. The position of each point shown on the
the same slope from ID = 10 to the practically impossible value M-SBT chart is determined from the average ID and KD values
of ID = 100. This extension of the boundaries of the SBT chart of three successive readings, which differ by less than 10%.
given by Robertson (2015) is mainly due to the clearer pre-
sentation of the data and completeness of the chart. The pres- Fine-grained soils
ence of microstructure in the soil is estimated by comparing
the measured and estimated G0/σv’ ratios. The contours of the Figure 6 presents the results for fine-grained soils. It is inter-
G0/σv’ ratio are derived by linking the correlations presented esting to compare the Fiumicino clay and Smederevo silty
Determination of the presence of microstructure in a soil using a seismic dilatometer

Fig. 6 Validation of proposed

microstructure-soil behavior type
(M-SBT) chart: fine-grained soils
(ID < 0.6) and transitional soils

clay both plotted in the FC region lying relatively close to each presence of microstructure. It is believed that the microstruc-
other. The datapoints for both soils plot on the same ture is caused by cementation and aging, as indicated in the
Bestimated^ G0/σv’ lines but the Smederevo silty clay has a legend of Fig. 6. Cementation could be a consequence of the
significantly higher Bmeasured^ G0/σv’ ratio. This may indi- high carbonate content (24%) measured in the laboratory.
cate that the behavior of the silty clay is influenced by the Marchetti et al. (2001) suggested that cemented clays can be

Fig. 7 Profiles of intermediate

dilatometer test (DMT) parame-
ters for Smederevo silty clay
(depth below 9.0 m)
 D. Berisavljević, Z. Berisavljević

recognized by a constant value of KD (KD = 3–4) with depth. variability should still be allowed due to the local soil condi-
For Smederevo silty clays, KD is approximately equal to 3 and tions. For the Vrbas II sand, the estimated and measured G0/
is constant with depth, as shown in Fig. 7. σv’ ratios closely match, as can be observed from Fig. 8. This
As presented on the M-SBT chart in Fig. 6, points for marl is expected and consistent with the trends shown in Fig. 4b.
(HOC marl) plot in the FD region, which could be a conse- Most datapoints for both sands lie above the contractive–dila-
quence of the high apparent overconsolidation. The estimated tive boundary, indicating a dilative behavior (negative state
average G0/σv’ ratio for both marls is more than two times parameter) during drained loading. The NC silty sand site is
lower than the average measured G0/σv’ ratio. This is expect- located near the city of Subotica in the northern part of Serbia.
ed due to the presence of cementation in marls. Based on the The results correspond to the data obtained at depths ranging
previous two groups of soil discussed, the M-SBT chart shows from 19 to 25 m below ground level measured in two SDMTs.
a strong potential to accurately identify soil types based on Subotica sand is rich in carbonates, as was observed from
their mechanical response to blade penetration in the DMT. inspection of core samples using HCl acid. As a consequence
The last soil included in the chart given in Fig. 6 is Obrenovac of a high carbonate content, the sand is colored white. Lower
stiff silty clay, which is found above water level at relatively cone penetration resistances (qc) and thrust measurements
shallow depths (3–7 m) from ground elevation. Subsequent (i.e., qD) at a similar stress level (i.e., similar depth of testing)
laboratory testing on samples retrieved from boreholes indi- may indicate the higher compressibility of Subotica sand than
cated almost full saturation of this soil (Sr > 95%). Thus, it previously described quartz sands, which on the other hand
could be expected that suction has a negligible influence on may indicate that this sand has a different mineralogy.
the measured Vs and G0. Datapoints for this soil plot in the Unfortunately, no systematic research has been conducted in
transitional zone, e.g., 0.6 < ID < 1.8, which could indicate that order to inspect the mineralogical composition of this sand
the measured DMT pressures are influenced by partial drain- more closely. The field observations and SDMT results indi-
age of pore pressures during the test (Schnaid and Odebrech, cate that the silty sand is cemented but that its structure is very
2015). Unfortunately, during the test, the C reading and/or sensitive to disturbance produced by large shear strains im-
short dissipation tests (as suggested by Marchetti 2015) were posed during blade (or cone) penetration. On the other hand,
not performed in order to determine the influence of partial small strains imposed during propagation of shear waves keep
drainage on the measured DMT pressures. However, high KD the structure intact. These observations are reflected in the
values could indicate the dilative nature of this stiff clay. The large difference (on average 2.3 times) between the measured
measured and estimated G0/σv’ ratios differ on average by and estimated G0/σv’ ratios. Even if the contractive–dilative
20%, which is much less than for the previously presented soil boundary, shown in Fig. 8, is determined for young soils with-
types. This is consistent with field observations and laboratory out significant microstructure (Robertson 2015), it is believed
testing, where no significant presence of microstructure is that the contractive behavior of Subotica sand determined
found in this soil. from the M-SBT chart corresponds to the expected behavior.

Coarse-grained soils Loess and loess-like soils

The results obtained in coarse-grained soils (ID > 1.8) are pre- Figure 9 presents a M-SBT chart including the results obtained
sented in Fig. 8. The three sands presented are normally con- for various soil types. These results are taken either from the
solidated, two of which are clean quartz alluvial sands (≈88% literature or from the author’s database.
quartz) belonging to the same geological unit (Vrbas I and In light of the DMT results, collapsible loess behaves as a
Vrbas II). For quartz sands, a close match between the mea- sand-like soil (Berisavljevic et al. 2014; Lutenegger and
sured and estimated G0/σv’ ratios is observed. The average Donchev 1983; Mulabdic and Minazek 2015). On the M-
estimated G0/σv’ ratio for the sand at the Vrbas I site is ap- SBT chart data, points for loess plot in the CC region far below
proximately 800, while the average measured ratio is 1070. the contractive–dilative boundary. This indicates that at higher
The relative difference between the average measured and strain (stress) levels, which cause the loess structure to col-
average estimated G0/σv’ ratios is approximately 25%. lapse, loess has a contractive behavior. Most of the datapoints
When assessing the presence of microstructure in soil from of the estimated G0/σv’ ratio lie below the G0/σv’ = 750 line,
the G0/σv’ ratio, a larger relative difference should be allowed while the measured ratios are significantly higher, ranging
than when estimating it from the Vs ratio, since G0 is derived from 1090 to 4380. The high measured G0/σv’ ratios are a
from the square of the measured velocity. Thus, it is believed consequence of the larger grain-to-grain contact area produced
that the 25% relative difference is common and to be expected by cementation of the larger silt and sand particles with clay
for soils without any significant presence of microstructure. minerals and CaCO3. Thus, large differences in the measured
The trends shown in Fig. 3 are derived for the global database and estimated ratios are a clear indicator of the presence of
(Marchetti et al. 2008; Monaco et al. 2009), but some microstructure in the soil. A description of the loess sites
Determination of the presence of microstructure in a soil using a seismic dilatometer

Fig. 8 Validation of proposed

microstructure-soil behavior type
(M-SBT) chart: coarse-grained
soils (ID > 1.8)

presented herein can be found in Berisavljevic et al. (2014 and area allows determination of the estimated G0/σv’ ratio. It can
2015). Beside Serbian loess, other loess-like soils are included be observed that the estimated and measured G0/σv’ ratios
on the M-SBT chart shown in Fig. 9. closely match, which is consistent with findings reported by
The results reported by Mlynarek et al. (2015) represent Mlynarek et al. (2015). Datapoints before and after wetting
non-collapsible, normally consolidated loess silts with G0/σv’ for Spanish loess-like soil reported by Devincenzi and
ratios ranging from 500 to 850. The area marked in Fig. 9 Canicio (2001) are included in the chart. The presented results
indicates the range of ID and KD values estimated from charts indicate that soil before wetting behaves as coarse-grained di-
reported by Mlynarek et al. (2015). The location of the marked lative (CD region), while after wetting it behaves as coarse-

Fig. 9 Validation of proposed

microstructure-soil behavior type
(M-SBT) chart: various soil types.
G0 small strain stiffness
 D. Berisavljević, Z. Berisavljević

grained contractive (CC). The change of position on the M- extent of lateritic soil from the measured G0/σv’ ratio. Based
SBT chart is a direct consequence of soil structure collapse on the findings reported by Giacheti et al. (2006) for the Bauru
caused by wetting of the soil. In the pre-wetted state, the cohe- site, it seems that the G0/σv’ ratio is a better parameter than G0/
sion (cementation) contribution causes high apparent ED for distinguishing lateritic from saprolitic soil.
overconsolidation, which is reflected in the dilative behavior A final observation regarding the position of collapsible soil
similar to dense sand. After saturation, this cementation is lost, on the M-SBT chart can be made. Datapoints which are known
which causes the soil to behave as NC loose sand. It is interest- to represent collapsible soil from the author’s database as well
ing to note that the Serbian loess and loess silts reported by as soils reported in the literature are bonded by a red line. For
Devincenzi and Canicio (2001) behave as a sand-like soil even Brasilia porous clay, the datapoints for the top 8 m are believed
if both soil types have high fines content (> 85%). When tested to represent potentially collapsible soil. Collapsibility is
at natural moisture content (for both soils the degree of satura- assessed from the dry unit weight, void ratio, and moisture
tion is approximately 50%), these soils show significantly dif- content taken or evaluated from figures reported by Ortigao
ferent behavior, as can be seen from their positions on the M- (1994) and Ortigao et al. (1995). For lateritic soil, it is assumed
SBT chart, i.e., the loess-like silts plot far above the contrac- that the topmost 4 m are collapsible. From Fig. 9, it can be
tive–dilative boundary, while Serbian loess plots far below it for observed that no single position of collapsible soil exists in
a similar range of ID (ID > 1.8). A possible explanation of this the M-SBT chart. This signifies the importance of geotechnical
opposite behavior of these two loess soils can be sought in the investigations, site characterization, and complexity in various
different values of their natural void ratio (e0). Devincenzi and soil type identifications based on their mechanical behavior
Canicio (2001) reported that e0 for tested loess-like soil ranges during penetration testing. It should be mentioned that all the
from 0.6 to 0.7, while for Serbian loess e0 ranges from 1 to 1.2. datapoints presented on the M-SBT charts taken from the liter-
This may imply that Serbian loess is much more sensitive to ature have approximate positions, since they are read or digi-
disturbance caused by blade penetration. Thus, in sensitive tized from graphs reported in the mentioned references.
loess the soil around the blade collapses, and is displaced fur-
ther away from the blade. Consequently, the pressures taken in
the DMT reflect the result of a soil tested in a completely dis- Trends of small strain stiffness (G0)
turbed state providing excessively low values of KD. On the with intermediate dilatometer test (DMT)
other hand, DMT pressures in less sensitive loess are taken in parameters
soil with partially preserved structure, giving a result that rep-
resents Breal^ soil behavior. The well-documented trend that G0 normalized with respect to
one of the DMT-derived parameters, such as ED or MDMT, de-
Tropical soils creases with increasing KD or ID has been previously reported
(Marchetti 2008; Monaco et al. 2009; Cruz et al. 2012; Rivera-
The results reported by Ortigao (1994), Ortigao et al. (1995), and Cruz et al. 2012). Figure 10a shows trends of the ratio of mea-
Rocha et al. (2015) for tropical soils are shown on the M-SBT sured to estimated small strain shear moduli (Gmeasured/Gestimated)
chart given in Fig. 9. Brasilia porous clay falls into the transi- with KD for soils with ID > 1.8. Datapoints for Aveiro sand taken
tional soil region, which may indicate difficulties in the deriva- from graphs reported by Amoroso et al. (2015) are added beside
tion of geotechnical parameters from the usual interpretation the previously described sites. The following trends emerge from
formulae (e.g., Marchetti et al. 2001). This soil has a high void Fig. 10a: in normally consolidated sands with no presence of
ratio (1.7) and low dry densities produced by the laterization microstructure, Gmeasured/Gestimated is almost constant with KD
process, particularly in the top layers. As reported by Ortigao and ranges from 0.8 to 1.45, but most datapoints are grouped
et al. (1995), this clay is collapsible, which is indicated by the red around 1. For soils with a significant presence of microstructure,
boundary on the SBT chart. G0 was not reported for this site . no particular trend between Gmeasured/Gestimated and KD is ob-
Recently, Rocha et al. (2015) presented the results of SDMT served. For these soils, Gmeasured/Gestimated is larger than 1.8. In
performed in lateritic and saprolitic soils. Datapoints for the Fig. 10b, the relationship between Gmeasured/Gestimated and ID is
Bauru RS2 site are shown in Fig. 9. Rocha et al. (2015) reported shown. It can be observed that no particular trend exists between
that the mechanical behavior of lateritic and saprolitic soil differs Gmeasured/Gestimated and ID for any type of soil.
significantly. The SBT chart recognizes this difference in behav- The presented results indicate that the boundary separating
ior, which can be seen from the different locations of points soils with and without a significant presence of microstructure
representing both soil types. Moreover, the measured G0/σv’ can be drawn at a ratio of Gmeasured/Gestimated of 1.5. This is
ratio is much higher in lateritic than in saprolitic soil. The trend consistent with a reported 20% average relative error (e.g.,
of G0/σv’ to decrease with depth is similar to the evolution of the Marchetti, 2014) in the estimated Vs from mechanical DMT,
laterization process from the ground surface toward deeper i.e., uncertainty emerging from local site conditions and soil
layers. This similarity shows the potential of determining the variability are included in the proposed constant Gmeasured/
Determination of the presence of microstructure in a soil using a seismic dilatometer

Fig. 10 Ratio of Gmeasured/

Gestimated versus (a) KD for
coarse-grained soils; and (b) ID
for various soil types

Gestimated value that separates soils with and without a signif- that this type of graph is not non-dimensional, since G0 is in
icant presence of microstructure. MPa. This makes the ordinate sensitive to changes in scale
Figure 11 shows the plot of the G0/KD ratio versus KD on a when different units for G0 are used. However, the general
log–log scale for all values of ID. There is a clear tendency of trend of the G0/KD ratio decreasing with increasing KD would
decreasing G0/KD with increasing KD. be the same. No particular trend can be observed between G0/
It can be observed that coarse-grained soils tend to have a KD and ID, except for the previously mentioned fact that
higher G0/KD ratio than fine-grained soils for the same value coarse-grained soils have a higher G0/KD ratio than fine-
of KD. However, the graph should not be used alone for de- grained soils.
termination of the presence of microstructure, since different
soil types can overlap. An example is the position of
Smederevo silty clay (ID < 0.6) which has a high presence of Influence of microstructure on derived
microstructure plot within the region of NC Aveiro sand (ID > parameters
1.8) without microstructure. When compared, NC sands have
different rates of decreasing G0/KD with KD, with a higher rate This section shows the results of comparison between the dif-
for a sand in a looser state (higher ψ). It should be mentioned ferent methods used to derive axisymmetric effective peak
 D. Berisavljević, Z. Berisavljević

Fig. 11 Ratio of G0/KD versus

KD for various soil types. G0
small strain stiffness

friction angles (φp’) and the lateral stress coefficient at rest (K0) – The DMT- and CPT-based correlations predict similar
for soils with and without significant presence of microstruc- values. The correlation proposed by Marchetti (1997) gives
ture. Friction angles are derived from common correlations the lowest values of friction angles. This is consistent with
used for DMT and CPT interpretation in sands. The correlations the findings that φp’ derived from KD represents a lower
with the respective equations used are shown in Table 1. Two bound value (see Marchetti et al. 2001; Mayne 2015).
DMT-based, two CPT-based, and one Vs-based correlations are – The highest φp’ is predicted by the Vs-based approach.
used. The Vs was determined from an SDMT. The distance Uzielli et al. (2013) derived the correlation shown in
between the SDMT and CPT is approximately 1 m. All the Table 1 based on 12 datasets for the stress-normalized
mentioned correlations have been derived predominantly for Vs in the range of 125–225 m/s. For the Vrbas II site,
quartz-silica sands from the results obtained in the calibration the majority of stress-normalized Vs are higher than
chamber (CC) or triaxial testing on undisturbed samples. 225 m/s or close to the upper boundary of 225 m/s. This
Figure 12 shows the results obtained from the Vrbas II site. may be one of the reasons for the inconsistency between
For this site, CPT was performed to a depth of approxi- φp’ predicted from Vs and other applied correlations.
mately 18 m. Figure 12 indicates the following: – The difference between φp’ derived from D&M theory
(Schmertmann, 1988) and from qc (Kulhawy and Mayne,
– All correlations give a similar distribution of φp’ with 1990) increases with increase in sand density. This differ-
depth. ence is largest at depths from 11.6 to 13.2 m. Briaud and

Table 1 Origin of equations used for φp’) determination (key in Figs. 12 and 13)

Approach based on Equation Reference

KD (DMT) φp′ = 28° + 14.6 log (KD) − 2.1[log(KD)]2 Marchetti (1997)

qD (DMT) qf = qD = γsBNγqξγq Durgunoglu and Mitchell (1973); Schmertmann (1988)
Dr (CPT)a φp’ = 34.5 + 0.1{0.32 ln [(qc/pa)/[17.68(σv0 ’ /pa)0.5]]} Jamiolkowski et al. (2001); Schmertmann (1978)
0
qc (CPT) φp ¼ 17:6 þ 11log½ðqc =pa Þ =ðσv ’ =pa Þ 0:5 Kulhawy and Mayne (1990)
Vs (SDMT) φp’ (degrees) = 3.9°[Vs/(σv ′ /pa)0.25]0.44 Uzielli et al. (2013)

φp’ secant axisymmetric effective peak friction angle, σv’ vertical effective stress, γs average effective unit weight for the soil above the DMT blade, ξγq
shape factor, B thickness of DMT blade, CPT cone penetration test, DMT dilatometer test, KD horizontal stress index, Nγq bearing capacity factor, pa
atmospheric pressure, qc cone resistance, qD DMT penetration stress, qf D&M Durgunoglu & Mitchell bearing capacity, SDMT seismic dilatometer, , Vs
shear wave velocity
a
General form of this equation is φp’ = a + b{Dr} (Schmertmann 1978). Coefficients a and b are 34.5 and 0.1, respectively (see Fig. 12 from
Jamiolkowski 2001). Relative density (Dr) is estimated based on Eq. 5 proposed by Jamiolkowski et al. (2001). Coefficients C0, C1, and C2 are
17.68, 0.5, and 3.1 respectively. These coefficients (see Table 4 in Jamiolkowski et al. 2001) represent average values for NC silica sands
Determination of the presence of microstructure in a soil using a seismic dilatometer

Fig. 12 Comparison of φp’ derived from various methods (see Table 1)

for Vrbas II site. φp’ secant axisymmetric effective peak friction angle,
KD horizontal stress index, qc cone resistance, qD dilatometer test pene-
tration stress, Vs shear wave velocity, Dr relative density Fig. 13 Comparison of φp’ derived from various methods (see Table 1)
for Subotica site. φp’ secant axisymmetric effective peak friction angle,
KD horizontal stress index, qc cone resistance, qD dilatometer test pene-
Miran (1992), based on the research conducted by tration stress, Vs shear wave velocity, Dr relative density
Jamiolkowski et al. (1988), mentioned that the difference
between φp’ laboratory (LAB) and φp’ (D&M) increases
with increasing sand density. Mayne (2014) and Mayne The presented results indicate that it is difficult to access
(2015) showed that the equation proposed by Kulhawy the ultimate shear strength from tests that induce small shear
and Mayne (1990) correlates well with φp’ (LAB) mea- strains in soil, such as when using Vs in order to determine
sured in triaxial compression tests. Thus, if φp’ derived φp’. The error is increased with increase in the presence of
from Kulhawy and Mayne (1990) correlation is assumed microstructure in the soil due to the sensitivity of Vs to it. In
to represent the LAB triaxial value, the greater difference contractive sands (Subotica sand) cementation has a small
between φp’ obtained from qc (LAB) and D&M in more influence on the measured penetration resistances (qc, qD,
dilative sand layers is confirmed. and KD), which is governed by the sand density and stress
– No correction for the failure envelope curvature is intro- state.
duced in any of the presented φp’ values. Thus, all values Some aspects of the derivation of K0 in soils with a signif-
correspond to the mean normal effective stress existing icance presence of microstructure are given below. Various
around the blade or cone during penetration. methods for estimating K0 in clay can be found in the literature
(Kulhawy and Mayne 1990; Lacasse and Lunne 1988;
Figure 13 shows φp’ values derived from the correlations Larsson 1989; Lunne et al. 1990; Marchetti 1980; Powel and
shown in Table 1 for the NC Subotica silty sand. For this sand, Uglow 1988; Smith and Houlsby 1995; Sully 1991; Yu 2004).
the results indicate the following: Kulhawy and Mayne (1990) proposed the general form of the
equation for determining K0 as:
– The correlation proposed by Uzielli et al. (2013) signifi-
cantly overpredicts φp’ compared to any other used meth- K 0 ¼ ðK D =β K Þ0:47 −0:6 ð2Þ
od. This signifies the importance of detecting the pres-
where parameter βk depends on the soil type and geologic
ence of microstructure in soil before using existing Vs-
based empirical correlations. At this site, the Vs1 values
are significantly higher than 225 m/s, the upper bound Table 2 βk parameter related to soil type (Kulhawy and Mayne 1990)
value for which the method applies, as suggested by
βk Soil type
Uzielli et al. (2013).
– DMT-based methods (KD and qD) predict similar values 0.9 Fissured clays
of φp’ when compared to each other. Higher values are 1.5 (Marchetti 1980) Insensitive clays
obtained from CPT-based methods (Dr and qc) than from 2 Sensitive clays
DMT-based methods. This is most pronounced approxi- 3 Glacial till
mately between 18.5 and 20.2 m.
 D. Berisavljević, Z. Berisavljević

1.5, and Eq. 2 with βk depending on the microstructure, is

shown in Fig. 14. The reference value derived from an
oedometer with horizontal stress measurement is included in
Fig. 14. It can be seen that K0 predicted by Eq. 2 with varying
βk compares much better with the reference K0 than the orig-
inal Marchetti (1980) correlation. The proposed approach is
sensitive to variation of the Gmeasured/Gestimated ratio. At
12.0 m, the Gmeasured/Gestimated ratio is higher than 3. In this
case, βk equal to 3 should be used in Eq. 2, giving the lowest
K0 value in the profile of 0.35. Nevertheless, most data below
10.0 m are close to the reference value of 0.55.

Comparison of DMT and Cone penetration

test with pore pressure measurement (CPTu)
SBT charts in marls
Fig. 14 Comparison of the lateral stress coefficient at rest (K0) derived
from seismic dilatometer (SDMT) and oedometer test results Robertson (2016) showed that Seismic cone penetration test
with pore pressure measurement (SCPTu) results in soils with
origin (Kulhawy and Mayne 1990). For preliminary assess- significant microstructure can be described differently by soil
ment of βk, Kulhawy and Mayne (1990) recommend the behavior type-normalized (SBTn) Qtn–Fr and Qtn–U2 charts.
values given in Table 2. Marchetti et al. (2001) states that in To illustrate this, results for very stiff, HOC Belgrade marl are
highly cemented clays the original equation (βk = 1.5) can shown in Fig. 15 on Qtn–Fr and Qtn–U2 charts. The SCPTu
significantly overestimate K0, since part of KD is due to ce- data from 22 to 26 m plot in the dilative-sand region (SD) of
mentation. If we assume that βk increases with cementation, the Qtn–Fr chart and the claylike-contractive (CC) of the Qtn–
which can be discerned from Table 2, the influence of cemen- U2 chart. High Qtn value is caused by high apparent OCR,
tation on the K0–KD correlation can be assessed by introduc- suggesting possible dilative behavior, where high U2 shows
ing the ratio of Gmeasured/Gestimated to be equal to βk in Eq. 2. a more contractive behavior at high shear strains consistent
This would ensure reducing the derived K0 with the increasing with cemented soils. Results presented here are somewhere
presence of microstructure in clayey soil. Thus, Eq. 2 can be between results obtained in Cooper Marl and Fernando
used with bk = Gmeasured/Gestimated, with βk ranging from 1.5 to Siltstone (soils with reference number 28 and 29 from
3 (as per Table 2). It is suggested to use this approach, named Robertson 2016). On the other hand, SDMT data for marls
BBetta^, for soils with ID < 0.6. To illustrate this, a comparison (similar to Belgrade marl) plot in the fine-grained-dilative
between the original Marchetti (1980) correlation, using βk = (FD) region, as shown in Fig. 6. Thus, the two most common

Fig. 15 Normalized (CPTu) cone penetration test with pore pressure measurement data from 22 to 26 m obtained in HOC Belgrade marl
Determination of the presence of microstructure in a soil using a seismic dilatometer

field tests, SCPTu and SDMT, provide three different descrip- suggested that the G0measured/G0estimated ratio be used as a di-
tions of marls (SD, CC, and FD) based on their behavior visor in Eq. 2 instead of a constant value of 1.5. If no presence
response during penetration and testing. These differences of microstructure is found in the soil (βk = G 0measured/
can be attributed to a different operational shear strain level G0estimated = 1.5), Eq. 2 becomes identical to the original
induced by cone and the blade during penetration. It should be Marchetti (1980) equation. The lower and upper bounds of
emphasized that CPTu-based soil behavior type charts, which G0measured/G0estimated used in Eq. 2 are 1.5 and 3, respectively.
have been used for some time now, can give valuable insight This approach is applicable for soils with ID < 0.6.
into recently developed DMT-based soil behavior type charts.

Conclusion References

This paper indicates the possibility of using SDMT results for Alonso EE, Gens A, Josa A (1990) A constitutive model for partially
soil type evaluation and determining the presence of micro- saturated soils. Geotechnique 40(3):405–430
structure. The intermediate parameters ID and KD are used for Amoroso S, Rodrigues C, Viana da Fonseca A, Cruz N (2015)
soil type determination, while the G0/σv’ ratio is used to de- Liquefaction evaluation of Aveiro sands from SCPTU and SDMT
Tests. Proceedings of the 3rd International Conference on the flat
termine the presence of microstructure in a soil. Four broad dilatometer DMT'15, Roma, pp 293–300
groups of soil types, as shown in Fig. 5, are distinguished Andrus RD, Hayati H, Mohanan NP (2009) Correcting liquefaction re-
based on their response to DMT blade insertion and mem- sistance for Aged Sands using measured to estimated velocity ratio.
brane expansion. It is shown that the G0/σv’ ratio can effec- J Geotech Geoenviron 135(6):735–744
tively be used to distinguish between soils with and without Wen B-P, Yan Y-J (2014) Influence of structure on shear characteristics of
the unsaturated loess in Lanzhou, China. Eng Geol 168:46–58
the presence of microstructure. It is found that the difference Berisavljevic D, Berisavljevic Z, Čebašek V, Šušić N (2014)
between the measured and estimated G0/σv’ ratio increases Characterisation of collapsing loess by seismic dilatometer. Eng
with increasing presence of microstructure. The boundary that Geol 181:180–189
separates soils with and without microstructure is found to be Berisavljevic D, Rakic D, Susic N, 2015. SDMT – a tool for in situ
identification of collapsible soils. Proceedings of the 3rd
at the ratio of G0measured/G0estimated of 1.5. The ratio of
International Conference on the Flat Dilatometer DMT'15, Roma,
G0measured/G0estimated is highest in macroporous loess, lateritic pp 457–463
soils, and marls, which are soils well-known to have a high Briaud JL, Miran J (1992) The flat dilatometer test. Department of
presence of microstructure. The lowest ratio is found in NC Transportation - Federal Highway Administration, Washington,
Fiumicino clay and clean quartz sands. Direct comparison of DC. Publication No. FHWASA-91-044, p 102
Burland JB (1990) On the compressibility and shear strength of natural
measured and estimated Vs is an equally good way to detect
clays. Geotechnique 40(3):329–378
the presence of microstructure in a soil. If the relative error Campanella RG, Robertson PK (1991) Use and interpretation of research
between the measured and estimated Vs is more than 20%, the dilatometer. Can Geotech J 28:113–126
soil may have microstructure. Direct use of the correlations Cotecchia F, Chandler RJ (1997) The influence of structure on the pre-
proposed by Marchetti et al. (2008) for G0/MDMT evaluation failure behaviour of a natural clay. Geotechnique 47(3):523–544
using KD and ID is applicable only if the soil is known to have Cruz N, Rodrigez C, Viana da Fonseca A (2012) Detecting the present of
cementation structures in soils, based in DMT interpreted charts. In:
a low presence of microstructure. It is shown that the G0/KD Couthino, Mayne (eds) In geotechnical and geophysical site charac-
ratio decreases with increasing KD. Coarse-grained soils tend terization. Taylor and Francis group, London, pp 1723–1728
to have a higher G 0 /K D ratio for the similar range of Cuccovillo T, Coop MR (1997) Yielding and pre-failure deformation of
G0measured/G0estimated. The applicability of the Vs or G0/σv’ structured sands. Geotechnique 47(3):491–508
Devincenzi MJ, Canicio M (2001) Geotechnical characterization by in
ratio to detect the presence of microstructure in transitional
situ tests of a loess-like deposit in its natural state and after satura-
soils (0.6 < ID < 1.8) still needs to be confirmed. tion. Proceedings of the International Conference on In-Situ
For quartz-silica sands without microstructure, the com- Measurement of Soil Properties and Case Histories, Bali,
mon correlations (derived for quartz-silica sands) used to ob- Indonesia, 159–166
tain the effective peak axisymmetric friction angle (φp’) from Diaz-Rodriguez JA, Santamarina JC (2001) Mexico City soil behavior at
different strains: observations and physical interpretation. J Geotech
DMT and CPT can be used with confidence. This observation Geoenviron Eng 127(9):783–789
is based on the similarity and small variation of the estimated Durgunoglu TH, Mitchell JK (1973) Static penetration resistance of soils.
φp’ with all the correlations used. The Vs-based approach Prepared for NASA Headquarters, Washington, D. C. under NASA
predicts higher values of φp’ than the other methods used in Grant NGR 05–003-406, "Lunar Soil Properties and Soil
Mechanics"
quartz-silica sands. In sands with a high presence of micro-
Fernandez A, Santamarina JC (2001) Effect of cementation on the small
structure, φp’ predicted from Vs is overly high and should not strain parameters of sands. Can Geotech J 38(1):191–199
be used in design. We present an idea of reducing K0 by taking Gasparre A, Coop MR (2008) Quantification of the effects of structure on
into account the influence of microstructure on KD. It is the compression of a stiff clay. Can Geotech J 45:1324–1334
 D. Berisavljević, Z. Berisavljević

Giacheti HL, Peixoto ASP, De Mio G, Carvalho D (2006) Flat dilatom- Mayne PW (2014) Interpretation of geotechnical parameters from seismic
eter testing in Brazilian tropical soils. In: Failmezger RA, Anderson piezocone tests. Proceedings of the 3rd International Symposium on
JB (eds). Second International Conference on the Flat Dilatometer, Cone Penetration Testing, Las Vegas, Nevada, 47–73
Washington. Vol. 1. ASCE, Washington, DC p 103–110 Mlynarek Z, Wierzbicki J, Manka M. (2015) Geotechnical parameters of
Hayati H, Andrus RD (2009) Updated liquefaction resistance correction loess soils from CPTU and SDMT. Proceedings of the 3rd
factors for Aged Sands. J Geotech Geoenviron 135(11):1683–1692 International Conference on the flat dilatometer DMT'15, Roma,
Heineck KS, Coop MR, Consoli NC (2005) Effect of microreinforcement pp 481–488
of soils from very small to large shear strains. J Geotech Geoenviron Monaco P, Marchetti S, Totani G, Marchetti D (2009) Interrelationship
131(8):1024–1033 between small strain modulus G0 and operative modulus. In:
Jamiolkowski M, Ghionna V, Lancellotta R, Pasqualini E (1988) New Kokusho T, Tsukamoto Y, Yoshimine M (eds) Performance-based
Correlations of Penetration Tests for Design Practice. Proc. ISOPT- Design in Earthquake Geotechnical Engineering- from case history
1, Orlando, Florida,z 1:263–296 to practice, proc. IS-Tokyo 2009, Tsukuba, Japan, June 15–17.
Jamiolkowski M, Lo Presti DCF, Manassero M (2001) Evaluation of Taylor & Francis Group, London (CD-Rom), pp 1315–1323
relative density and shear strength of sands from cone penetration Mulabdic M, Minazek K (2015) Use of dilatometer in unusual difficult
test (CPT) and flat dilatometer test (DMT). ASCE Geotechnical soils – a case study. Proceedings of the 3rd International Conference
Spec Publ 119:201–238 on the Flat Dilatometer DMT'15, Roma, pp 497–504
Jiang M, Zhang F, Hu H, Cui Y, Peng J (2014) Structural characterization Ortigao JAR (1994) Dilatometer tests in Brasilia porous clay. 7th
of natural loess and remolded loess under triaxial tests. Eng Geol International Congress of the International Association of
181:249–260 Engineering Geology, Lisbon, Portugal
Jotisankasa A, Ridley A, Coop MR (2007) Collapse behavior of Ortigao JAR, Cunha RP, Alves LS (1995) In situ tests in Brasilia porous
compacted Silty clay in suction-monitored Oedometer apparatus. J clay. Can Geotech J 33:189–198
Geotech Geoenviron 133(7):867–877 Pestana JM, Salvati LA (2006) Small-strain behavior of granular soils. I:
Kulhawy FH, Mayne PW (1990) Estimating soil properties for founda- model for cemented and Uncemented Sands and gravels. J Geotech
tion design. EPRI report EL-6800. Palo Alto, Electric Power Geoenviron 132(8):1071–1081
Research Institute, p 306 Powell JJM, Uglow IM (1988) The interpretation of the Marchetti
Lacasse S, Lunne T (1988) Calibration of dilatometer correlations. Dilatometer Test in UK clays. ICE Proceedings of the Conference
Proceedings of ISOPT-1, Orlando, FL, Vol. 1, 539–548 on Penetration Testing in the UK, University of Birmingham, Paper
Lambe TW, Whitman RV (1969) Soil mechanics. John Wiley & Sons No. 34, 269–273
Inc., New York Rinaldi VA, Redolfi ER, Santamarina JC (1998) Characterization of col-
Larsson R (1989) Dilatometerförsök (dilatometer tests) [in Swedish]. lapsible soils with combined geophysical and penetration testing.
Swedish Geotechnical Institute, Information No. 10, Linköping Proceedings of the First International Conference on Site
Leroueil S, Vaughan PR (1990) The general and congruent effects of Characterization, ISC ‘98, vol. 1, Atlanta (GA)
structure in natural soils and weak rocks. Geotechnique 40(3): Rinaldi VA, Santamarina JC (2008) Cemented soils: small strain stiffness.
467–488 In: Burns, Mayne and Santamarina (eds) IOS-Millpress.
Lunne T, Powell JJM, Hauge EA, Uglow IM, Mokkelbost KH (1990) Deformational Characteristics of Geomaterials 1:267–274
Correlation of dilatometer readings to lateral stress. Paper submitted Rivera-Cruz I, Howie J, Vargas-Herrera LA, Coto-Loria M, Luna-
to specialty session on measurement of lateral stress. 69th Annual Gonzalez O (2012) A new approach for identification of soil behav-
Meeting of the Transportation Research Board, Washington, DC iour type from seismic dilatometer (SDMT) data. In: Couthino,
Lutenegger AJ, Donchev P (1983) Flat dilatometer testing in some meta- Mayne X (eds) In geotechnical and geophysical site characteriza-
stable loess soils. Proc Int Symp In-Situ Testing Soil Rock tion. Taylor and Francis group, London, pp 947–954
Properties 2:337–340 Robertson PK (2010) Estimating in-situ state parameter and friction angle
Marchetti S (1980) In situ tests by flat dilatometer. J Geotech Eng 106(3): in sandy soils from CPT. 2nd International Symposium on Cone
299–321 Penetration Testing, CPT’10, Huntington Beach (CA)
Marchetti S (1997) The flat dilatometer: design applications. Proceedings Robertson PK (2012) Interpretation of in-situ tests – some insights. J.K.
of the Third International Geotechnical Engineering Conference, Mitchell Lecture, Proceedings of the 4th International Conference
Cairo University, Soil Mechanics and Foundation Research on Site Characterization, ISC’4, Porto de Galinhas, Brazil, 3–24
Laboratory, Egypt: 423–448 Robertson PK (2015) Soil behavior type using the DMT. Proceedings of
Marchetti S (2014) The seismic dilatometer for in situ soil investigations. the 3rd International Conference on the Flat Dilatometer DMT'15,
Proceedings of the Indian Geotechnical Conference IGC-2014, Roma, pp 243–250
Kakinada, India Robertson PK (2016) Cone penetration test (CPT)-based soil behaviour
Marchetti S (2015) Some 2015 updates to the TC16 DMT report 2001. type (SBT) classification system — an update. Can Geotech J 53:
Proceedings of the 3rd International Conference on the Flat 1910–1927
Dilatometer DMT'15, Roma, pp 43–65 Rocha BP, Castro BAC, Giacheti GL (2015) Seismic DMT Test in a non-
Marchetti S, Monaco P, Totani G, Calabrese M (2001) The flat dilatom- text book type geomaterial. Proceedings of the 3rd International
eter (DMT) in soil investigations (ISSMGE TC16). Proceedings of Conference on the Flat Dilatometer DMT'15, Roma, pp 505–512
the International Conference on In-Situ Measurement of Soil Sally JP (1991) Measurement of in situ lateral stress during full displace-
Properties and Case Histories, Bali, Indonesia, 95–131 ment penetration tests [Ph.D. Thesis]. University of British
Marchetti S, Monaco P, Totani G, Marchetti D (2008) In situ tests by Columbia, Vancouver, BC, pp 485
seismic dilatometer (SDMT). In Laier JE, Crapps DK, Hussein Schmertmann JH (1978) Guidelines for cone penetration test, perfor-
MH (eds), From research to practice in geotechnical engineering, mance and design. Report FHWA-TS-78-209, Washington, 145
ASCE Geotech. Special Publication No. 180 (honoring Dr. John H. Schmertmann JH (1988) Guidelines for using the CPT, CPTU and
Schmertmann), pp 292–311 Marchetti DMT for geotechnical design. U.S. Department of
Mayne PW (2015) Peak friction angle of undisturbed sands using DMT. Transportation, Federal Highway Administration, Office of
3rd International Conference on the Flat Dilatometer DMT'15, Research and Special Studies, Report No. FHWA-PA- 87 023+24,
Roma, pp 237–242 Vol. 3–4
Determination of the presence of microstructure in a soil using a seismic dilatometer

Schnaid F (2005) Geo-characterisation and properties of natural soils by geotechnical design codes of practice, advances in Soil Mechanics
in situ tests. Proceedings of the 16th International Conference on & Geotechnical Engineering (series), vol 1. IOS Millpress,
Soil Mechanics and Geotechnical Engineering, vol. 1, Osaka, Amsterdam, pp 214–227
Millpress, Rotterdam, pp 3–45 Viana da Fonseca A, Carvalho J, Ferreira C, Santos JA, Almeida F, Pereira
Schnaid F, Odebrech E (2015) Challenges in the interpretation of the E, Feliciano J, Grade J, Oliveira A (2006) Characterization of a profile
DMT in tailings. Proceedings of the 3rd International Conference of residual soil from granite combining geological, geophysical and
on the Flat Dilatometer DMT'15, Roma, pp 13–23 mechanical testing techniques. Geotech Geol Eng 24:1307–1348
Schneider JA, Moss RES (2011) Linking cyclic stress and cyclic strain Yamamuro JA, Wood FM, Lade PV (2008) Effect of depositional method
based methods for assesment of cyclic liquefaction triggering in on the microstructure of silty sand. Can Geotech J 45:1538–1555
sands. Geotechnique Lett 1:31–36 Yu HS (2004) In situ soil testing: from mechanics to interpretation. 1st J.
Smith MG, Houlsby GT (1995) Interpretation of the Marchetti dilatom- K. Mitchell Lecture, Proceedings of ISC-2, Porto. 1, 3–38
eter in clay. Proc 11th ECSMFE 1:247–252 Yun TS, Santamarina JC (2005) Decementation, softening, and collapse:
Uzielli M, Mayne PW, Cassidy MJ (2013) Probabilistic assessment of changes in small-strain shear stiffness in k0 loading. ASCE J
design strengths for sands from in-situ testing data, Modern Geotech Geoenviro Eng 131(3):350–358