CE731A : Risk and Reliability in Geotechnical Engineering | Instructor : Dr. Prishati Raychowdhury
I. INTRODUCTION
A. Spatial Soil Variability
One of the distinctive features of soil and earth
Abstract— Natural soil variability can result in a
materials is their inherent variability even within
range of pile capacities at a specific test site. Soil
short distances which makes Geotechnical
variability is ideally determined using in-situ testing
Engineering an art to study in civil engineering. The
techniques such as the Cone Penetration Test.
variation of property from one location to another is
Spatial correlation structures are usually used while
called spatial variability.
modelling inherent uncertainty associated with soil
In addition to mean and standard deviation of soil
properties. To retrieve this information a sufficient
parameters, the spatial correlation has been
number of measurements and equal intervals
increasingly recognized as an influential property in
between measurements or samples are always
determining probabilistic outcomes. Various
required. However, in practice, it is not usual to
models have been proposed in last decades to
conduct site investigation in that way subject to a
identify the factors affecting overall variability and
limited budget. The main topic of this paper is to
magnitude of each source of uncertainty. The
estimate the spatial correlation of a given soil
Inherent soil variability arises due to complex geo-
deposit sample. Spatial correlation can be an
physical phenomenons such as sedimentation,
important factor in reliability analysis of
weathering, development of stress over time. Other
geotechnical systems, yet it is rarely estimated
types of error that occur are measurement errors and
during routine site investigations. In this report,
transformation errors. Measurement errors occur
spatial structure of cone penetration data is
from equipment and operator errors while
estimated based on the parameters qc (cone tip
transformation error occurs when lab and field
resistance) and friction sleeve resistance of a
measurements are transformed into various models.
vertical soil profile at a given site. Two different
Lithology is another cause of variability in soil
method for calculation of Auto correlation using
parameters manifested in form of soil layers and
Autocorrelation function and Variogram model is
pockets of different stiffness embedded in soil
suggested. In addition to this autocorrelation of
strata.
given CPT trace is quantified using a scale of
fluctuation. The SOF is also identified and
compared using various methods suggested in To date the analysis of the spatial variability of
literature like Variation reduction function, geotechnical materials has focussed on two
integration of sample autocorrelation function etc. mathematical techniques, namely, time series
Such spatial variability analysis can be utilized in analysis, and geostatistics. In this paper , we are
various domains of geotechnical engineering for focusing of time series analysis to determine the
different design and soil classification parameters. autocorrelation. In geotechnical properties,
however, the time domain is replaced by space or
distance domain.
B. Cone Penetration Test
Inherent variability in geotechnical properties can The cone penetrometer test (CPT) is a method used
be modelled by the given equation in which a depth to determine the geotechnical engineering
dependent geotechnical property (x) is broken down properties of soils. The test method consists of
into a deterministic component, (t) and a fluctuating pushing an instrumented cone, with the tip facing
component, (w) along with an error component (e) down, into the ground at a controlled rate
that totally represent the inherent soil variability. (controlled between 1.5 -2.5 cm/s accepted). The
resolution of the CPT in stratigraphic layers is
x(z) = t(z) + w(z) + e(z) related to the size of the cone tip, with typical cone
tips having a cross-sectional area of either 10 or 15
cm², corresponding to diameters of 3.6 and 4.4 cm.
The CPT is able to detect fine changes in the
stratigraphy and spatial variability of soil properties.
VI. GRAPHS
IV. VARIOGRAMS
The variogram function is a more appropriate way
to describe spatial relations primarily because it is
not related to the sample mean or sample variance.
Measure of the degree of spatial dependence
between samples along a specific support is a
variogram function. It is defined as the variance
between data at a particular lag distance.
Average of the CPT data– 10876.72 kpa Average of the CPT data – 173.70 kpa
Standard deviation of the CPT data –8084.83 kpa Standard deviation of the CPT data – 125 kpa
Variance of the CPT data – 65364411.60 square kpa Variance of the CPT data– 15632.11 square kpa
Fig. 5 Autocorrelation plot vs distance lag (Tip Resistance)
Fig. 4. Autocorrelation plot vs distance lag (Tip Resistance)
The auto covariance distance from the variogram The auto covariance distance from the variogram
function plot is the distance at which the spatial function plot is the distance at which the spatial
variance reduces by 1/e times (37%). The 67% variance reduces by 1/e times (37%). The 67%
variance line is plotted to find the auto covariance variance line is plotted to find the auto covariance
distance which is approximately 0.65 metres. distance which is approximately 0.50 metres.
Fig. 10. Autocorrelation comparison using direct formula and Fig. 11. Autocorrelation comparison using direct formula and
variogram method (Tip Resistance) variogram method (Friction Resistance)
VII. RESULTS & APPLICATIONS