Sie sind auf Seite 1von 34

Soil-Structure Interaction Research Project

"A Practical Subgrade Model for


Improved Soil-Structure Interaction Analysis:
Parameter Assessment"

Manhattan College Research Report No. CEEN/GE-2011-1


by
John S. Horvath, Ph.D., P.E., M.ASCE
Professor of Civil Engineering
in collaboration with
Regis J. Colasanti, M.S., P.E., M.ASCE
Consulting Engineer IV
URS Energy & Construction Division - Denver, Colorado

Manhattan College
School of Engineering
Civil and Environmental Engineering Department
Bronx, New York, U.S.A.

January 2011

ii

2011 by John S. Horvath. All rights reserved.


Prof. John S. Horvath, Ph.D., P.E., M.ASCE
Manhattan College
Civil and Environmental Engineering Department
Bronx, New York
U.S.A.
email: john.horvath@manhattan.edu
voice: +1-718-862-7177
fax: +1-718-862-8035

Limitations
The information contained in this report is the result of original research and is made available to
the public solely as a contribution to the general state of civil engineering knowledge. As such,
this report does not constitute a design manual or standard, and should not be interpreted, used, or
referred to as such. The authors and Manhattan College assume no liability for the performance of
any structure constructed using the analytical and design methodologies presented and discussed
in this report.
In addition, any reference, either direct or indirect, to a specific manufacturer, material, or
product is made solely for the purposes of identification and documentation relevant to the study
presented in this report and does not constitute an endorsement or promotion of that
manufacturer, material, or product by the authors or Manhattan College.
Distribution and Use
Authorization is hereby granted by the copyright holder to copy, distribute, and/or print this
report without restriction provided this report is copied and/or printed in its entirety, distributed at
no cost to recipients, and used for personal education and individual reference purposes only.
This report may not be sold or used in whole or in part in any manner for commercial purposes of
any type, including manufacturer's product or marketing literature, without prior written consent
of the copyright holder.
Those wishing to make a paper copy of this report are advised that it is formatted for doublesided printing. A gutter has been provided in the page formatting to allow space for binding along
the inner side if desired. It is suggested that the printer settings in Adobe Acrobat be set to Fit
to Printable Area and Auto-Rotate and Center for optimum readability.

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

iii
Contents
List of Figures................................................................................................................................. v
List of Tables .................................................................................................................................. v
Preface........................................................................................................................................... vii
Executive Summary ...................................................................................................................... ix
Introduction.................................................................................................................................... 1
Subgrade Models
Basic Concepts................................................................................................................... 1
Background ....................................................................................................................... 2
Recent Research................................................................................................................ 3
Modified Kerr-Reissner Hybrid Subgrade Model
Introduction....................................................................................................................... 3
Theoretical Basis ............................................................................................................... 3
Constituent Models
Modified Kerr/Horvath-Colasanti Model.......................................................... 4
Reissner Simplified Continuum Model.............................................................. 5
Synthesis ............................................................................................................................ 6
Parameter Assessment
Overview ............................................................................................................... 6
Estimation of Depth to Rigid Base ..................................................................... 7
Estimation of Equivalent Single-Layer Elastic Parameters............................. 8
Example Applications Using Case Histories
Introduction and Overview................................................................................. 9
Case A: Georgia State University (GSU) Building B........................................ 9
Massachusetts Institute of Technology (MIT)
Introduction and Overview.................................................................. 11
Site Characterization............................................................................ 12
Mat Weight............................................................................................ 13
Case History B: Whitaker Laboratory ............................................... 15
Case History C: Chemistry Building .................................................. 17
Summary and Conclusions.......................................................................................................... 19
Recommendations ........................................................................................................................ 19
Acknowledgements ...................................................................................................................... 20
Appendix I - Parameters for MK-R Model
No Plate or Perfectly-Smooth Plate-Subgrade Interface............................................. 20
Perfectly-Rough Plate-Subgrade Interface................................................................... 21
Appendix II - Extension of MT Method .................................................................................... 21
Soil-Structure Interaction Research Project
"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis - Part III: Parameter Assessment"
Manhattan College Department of Civil and Environmental Engineering Report No. CEEN/GE-2011-1

iv
References..................................................................................................................................... 23

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

v
List of Figures
Figure 1. Modified Kerr/Horvath-Colasanti mechanical subgrade model............................... 4
Figure 2. Reissner Simplified Continuum subgrade model ....................................................... 5
Figure 3. Steps in MK-R Model parameter evaluation .............................................................. 7
Figure 4. Comparison of calculated versus measured results for Case History A................. 11
Figure 5. Assumed mat loading for Case History B.................................................................. 16
Figure 6. Comparison of calculated versus measured results for Case History B ................. 16
Figure 7. Assumed mat loading for Case History C ................................................................. 17
Figure 8. Comparison of calculated versus measured results for Case History C................. 18
List of Tables
Table 1. Synthesis and overall hierarchy of subgrade models (Horvath and Colasanti 2011a)
......................................................................................................................................................... 2

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis - Part III: Parameter Assessment"
Manhattan College Department of Civil and Environmental Engineering Report No. CEEN/GE-2011-1

vi

This page intentionally left blank.

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

vii
Preface
This is the first in a new series of reports related to geotechnical engineering research performed
by the faculty at Manhattan College. It is the third such series created by me during my 24 years
as a member of the full-time faculty of the College's School of Engineering. The first series,
designated "CE/GE", was created the year I started at the College (1987) under the auspices of
the former Civil Engineering Department. A total of 22 reports, each authored solely by me, were
issued between 1987 and 2000. This was followed by a second series created in 2001 and
designated "CGT" under the auspices of the now-defunct Manhattan College Center for
Geotechnology (CGT) for which I was the Founding Director. This organization was intended to
be a multidisciplinary one that encompassed both traditional geotechnical as well as
geoenvironmental research (at that time civil and environmental engineering were separate
departments at the College) and was thus under the broader jurisdiction of the School of
Engineering. A total of 11 reports, again each authored solely by me, were issued between 2001
and 2004. This third and latest series, designated "CEEN/GE", is now created as we begin the
second decade of the 21st century. Reports in this latest series are being issued under the auspices
of the Civil and Environmental Engineering Department that was formed in 2004.
I feel it is fitting that this inaugural CEEN/GE report is another in a series of contributions to the
Soil-Structure Interaction Research Project. The topic of soil-structure interaction is one that has
interested me from literally the beginning of my professional engineering career in 1972 (as a
result of being given an assignment in practice to determine Winkler's Coefficient of Subgrade
Reaction for a project application) and was the basis of my doctoral studies during the 1973-1979
timeframe. It is thus a topic that has engaged my research interest for virtually my entire
professional career that is now in its 39th year. In addition, this project is one of three that I
formally identified, established, and conducted under the auspices of the aforementioned CGT.
Although that organization is now defunct the work established under its jurisdiction has never
stopped but has been continued by me up to the present in an ongoing, less-formal manner.
The particular technical content of this report focuses on one of three aspects related to the new
Modified Kerr-Reissner (MK-R) hybrid subgrade model that was developed by me in
collaboration with Mr. Regis J. Colasanti, P.E. who is also a collaborating author of this report. I
cannot overemphasize the important role Mr. Colasanti played in making the MK-R Model a
practical reality. The concept of what eventually became the MK-R Model has been with me
since I completed my doctoral studies more than 30 years ago. However, the analytical and
computational technology combined with the practical experience and know-how to properly
execute and implement the structural engineering aspects of the MK-R Model are something that
eluded me until my fortuitous collaboration with Mr. Colasanti that began almost three years ago
to the date, in February 2008. I consider this collaboration and its successful outcome in the form
of the MK-R Model one of the crown jewels of my professional career.
Because this report focuses on only one aspect of the MK-R Model (the practical geotechnical
aspects of its implementation) it is necessary for those interested in using this model to, as a
minimum, read two additional recent publications devoted to the theoretical basis and
development of this model (Horvath and Colasanti 2011a) and guidelines for how to implement
this model in commercially-available structural analysis software (Colasanti and Horvath 2010).
Copies of these publications can be accessed at my personal website (www.jshce.com).
John S. Horvath, Ph.D., P.E., M.ASCE
Bronx, New York, U.S.A.
15 January 2011
Soil-Structure Interaction Research Project
"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis - Part III: Parameter Assessment"
Manhattan College Department of Civil and Environmental Engineering Report No. CEEN/GE-2011-1

viii

This page intentionally left blank.

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

ix
Executive Summary
Subgrade models are an essential analytical tool used by both geotechnical and structural
engineers in many types of soil-structure interaction (SSI) analyses performed for both routine
foundation engineering practice as well as research. With few exceptions, the subgrade model
used has always been Winkler's Hypothesis which originated in the 19th century. This has been
done out of need and not choice as it has been appreciated for the better part of a century that
Winkler's Hypothesis, although easy to visualize and implement mathematically, is a very poor
model of actual subgrade behavior. The fundamental theoretical flaw inherent in Winkler's
Hypothesis is that it does not replicate the important behavioral mechanism of shear that occurs
within an actual subgrade. This theoretical shortcoming is usually referred to as a lack of spring
coupling in light of the typical visualization of Winkler's Hypothesis as representing an
assemblage of independent axial springs.
As a result of this inherent, serious theoretical deficiency, the use of Winkler's Hypothesis always
introduces significant complications that most users do not seem to fully understand and
appreciate. Specifically, in order to obtain 'correct' results when using Winkler's Hypothesis the
analyst must correctly anticipate the subgrade-shearing (spring-coupling) effects beforehand and
incorporate those effects in the input values of the Winkler Coefficient of Subgrade Reaction
(WCSR). Moreover, research has shown that the Coefficient of Subgrade Reaction (CSR), which
is fundamentally a generic calculated result of a SSI analysis and not an input parameter, varies
not only with the geotechnical properties of the subgrade but also the structural properties and
loading of the structural element in contact with the subgrade. In addition, in real applications the
CSR never has a constant magnitude along a structure-subgrade interface. Consequently,
developing and inputting appropriate values of the both the magnitude and distribution of the
WCSR is a daunting task as it essentially means the correct results must be known beforehand so
that the correct input parameters can be developed and used to produce the desired correct results.
While this may seem illogical it nevertheless correctly reflects the consequences of using
Winkler's Hypothesis and also explains why the decades-long search to link the WCSR to
fundamental soil properties has been an exercise in futility and, to a certain extent, a wasted
effort.
Subgrade models more advanced than Winkler's Hypothesis that inherently better replicate actual
soil subgrade behavior and, most importantly, do not introduce the same inherent complications
as Winkler's Hypothesis have existed at least since 1940. A hallmark of all subgrade models more
advanced than Winkler's Hypothesis is that they inherently incorporate soil shearing (springcoupling) in their mathematical formulation. The variation among and between advanced
subgrade models is simply the mathematical sophistication of the spring-coupling and thus how
closely it replicates the shear behavior of an actual subgrade. However, despite the inherent
superiority of these advanced subgrade models compared to Winkler's Hypothesis there have
always been pragmatic obstacles to using such advanced models. Either the correlation of model
parameters to actual soil properties has been unclear or it has been practically impossible to
implement the models in commercially-available computer software. The net result is that
Winkler's Hypothesis continues in widespread use to the present because of the practical and
pragmatic issues related to implementing more-advanced subgrade models.
The last few years have seen the development by the authors of a new advanced subgrade model
called the Modified Kerr-Reissner (MK-R) Model that is based on a unique hybrid derivational
approach that results in a subgrade model that eliminates all of the traditional implementation
difficulties and roadblocks of advanced subgrade models without sacrificing any of their
Soil-Structure Interaction Research Project
"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis - Part III: Parameter Assessment"
Manhattan College Department of Civil and Environmental Engineering Report No. CEEN/GE-2011-1

x
theoretical rigor and performance. This report is the planned third in a series of three publications
by the authors with the collective goal to disseminate not only knowledge about the MK-R Model
but provide both engineering practitioners and researchers alike with specific guidelines on how
to implement this model at the present time using structural and geotechnical engineering
methodologies that are already routinely available. Thus the MK-R Model is not just another
theoretical, but impractical, exercise but is ready for immediate use in practice and research.
Previous, recent publications by the authors have addressed the theoretical basis and development
of the MK-R Model as well as provided specific guidelines for its structural-engineering
implementation using commercially-available structural analysis software. These prior
publications are necessary reading to fully understand the contents of this report. This report
focuses on geotechnical-engineering implementation issues and illustrates the overall application
of the MK-R Model using three case histories of mat-supported buildings.

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

1
INTRODUCTION
Recent years have seen significant growth in the capabilities of computer hardware and
software that have allowed numerical modeling and analysis of two- and three-dimensional
geotechnical continua on a routine basis for an ever-increasing variety of applications. Such
continuum models allow use of constitutive models that have the potential to replicate a wide
range of soil behavior, including time dependency.
However, despite this growth in technology there are still many applications in routine practice
where analytical software utilizing subgrade models is still preferred. Subgrade models are not
complete constitutive models as they attempt to replicate only certain aspects of soil behavior
using an arithmetic equation involving two spatial dimensions to approximate three-dimensional
behavior (Horvath 1979, 1988, 1989, 2002). The situations where a subgrade model as opposed to
a complete constitutive model still predominate in practice are primarily foundation engineering
applications involving relatively flexible, plate-like structural elements such as mats (rafts) and
base slabs of cut-and-cover tunnels and water/wastewater treatment tanks where it is desirable or
even necessary to model the combined superstructure, foundation, and underlying ground (which
is usually referred to as the subgrade in such problems) as a single system. No known
geotechnical engineering software exists at the present time for such whole-system analysis so
structural engineering software must be used. This places significant restrictions on the
theoretical sophistication of the geotechnical components of the system that can be implemented
in such software (Colasanti and Horvath 2010).
Additional applications where subgrade models have proven to be useful include other
relatively flexible structural elements in contact with or embedded in the ground such as slabs-ongrade, 'rigid' (Portland-cement concrete or PCC) pavements, and deep foundations subjected to
lateral and/or moment loading. Collectively, these myriad and diverse applications involving
relatively flexible structural elements in direct contact with the ground are referred to as soilstructure interaction (SSI) problems. They reflect situations where a reasonably accurate
assessment of the overall structure-subgrade system requires satisfying not only static force and
moment equilibrium but also a complementary and compatible displacement and deformation
pattern for both the structure and subgrade along their interface(s).
SUBGRADE MODELS
Basic Concepts
As discussed in detail in Horvath (1979, 1988, 1989), all subgrade models have the same
general form as expressed in the following equation:

p + 2 p + 4 p + ... = W + 2W + 4W + ... .

(1)

where p is called the subgrade reaction and is the vertical normal stress applied at the subgrade
surface either directly or by the structural element in contact with the subgrade and W is the
normal displacement (e.g. settlement in the case of a mat foundation) of the surface to which p is
applied. Note that the coefficients multiplying the p and W terms in Eq. 1, which will vary
depending on the specific subgrade model, have been omitted here for simplicity. In addition,
depending on the specific subgrade model not all of the higher-derivative terms beyond the first
term on each side of Eq. 1 will appear. This is shown in Table 1 that is taken from Horvath and
Colasanti (2011a) and which summarizes all subgrade models found in the literature by the
primary author to date. This table will be referred to and further explained throughout this report.

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis - Part III: Parameter Assessment"
Manhattan College Department of Civil and Environmental Engineering Report No. CEEN/GE-2011-1

2
Table 1. Synthesis and overall hierarchy of subgrade models (Horvath and Colasanti 2011a).
Subgrade model
Mechanical
Simplified elastic continuum
Hybrid
Winkler's Hypothesis
Winkler-Type Simplified Continuum
[trivial]
Filonenko-Borodich
Pasternak-Type Simplified Continuum
[none to date]
Loof's Hypothesis
Vlasov and Leont'ev (1 layer)
Pasternak's Hypothesis
Modified Kerr/Horvath-Colasanti
Reissner Simplified Continuum
Modified KerrModified Pasternak/Kerr
Vlasov and Leont'ev (2 layers)
Reissner
Haber-Schaim
[none to date]
[none to date]
[none to date]
[none to date]
[none to date]
Hetnyi
[none to date]
[none to date]
Rhines
[none to date]
[none to date]
Note: Italicized models are those referred to in this report.

Derivative Derivative
order
order
of p(x,y)
of W(x,y)
0 2 4 0 2 4
X
X
X
X
X
X
X
X

X
X
X

X
X

X
X
X
X

Background
Because there is still very much a significant role for subgrade models in current geotechnical
and structural engineering practice and research this means the long-unfulfilled need for a
subgrade model that is a theoretically sound yet practical-to-use improvement to the inherently
and seriously flawed Winkler's Hypothesis, which originated in the 19th century, still exists1. Past
research has shown that efforts to develop such an improved model date back to at least circa
1940 and are thus more than 70 years old at this point in time (Horvath 1979, 1988, 1989).
As discussed in detail in Horvath (1979, 1988, 1989) and in summary in Horvath (2002) and
Horvath and Colasanti (2011a), subgrade model development has occurred more or less
contemporaneously using two completely different and independent conceptual and theoretical
approaches. The more common and well known is the mechanical approach in which various
mechanical elements such as axial springs, tensioned membranes, shear layers, and flexural layers
are arbitrarily combined to produce a physical model that has some well-defined mathematical
behavior. Winkler's Hypothesis as conceptually visualized as a layer of independent axial springs
is the classic example of a mechanical subgrade model and is, in fact, not only the simplest of
such models but also the basic element and starting point of every other more-advanced
mechanical subgrade model identified to date. This can be seen in Table 1 where the zero-order
derivative relationship between p and W that defines Winkler's Hypothesis appears in all moreadvanced, higher-order mechanical models.
The attraction of subgrade models developed using the mechanical approach is that they are
physically obvious and thus reasonably straightforward to model and thus implement in
commercially-available structural analysis software. However, the drawback of such models has
1

The theoretical flaws and shortcomings inherent in Winkler's Hypothesis have been discussed in
numerous publications over the years. However, as noted in the Executive Summary of this report the
subtle implication that these flaws have on the generic Coefficient of Subgrade Reaction (CSR), which is
defined simply as the ratio p(x,y)/W(x,y) and is fundamentally a calculated result in SSI analyses, appears
not to have been fully understood and appreciated by users. Colasanti and Horvath (2010) present a detailed
discussion of not only the well-known flaws and shortcomings of Winkler's Hypothesis but also clearly
explain the significant implications these shortcomings have on turning the CSR from an initially-unknown
problem result to a required problem input that must be known correctly beforehand if correct results are to
be obtained using Winkler's Hypothesis.
Soil-Structure Interaction Research Project
"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

X
X

X
X
X
X

3
always been how to rationally correlate and evaluate the spring stiffnesses, etc. of the various
mechanical elements to traditional soil properties on a site- and application-specific basis.
The other method for subgrade model development is the simplified elastic continuum (a.k.a.
simplified continuum) approach as pioneered by Reissner (1958). In this approach the starting
point is always the complete system of partial differential equations (equilibrium, compatibility,
constitutive) defining the stress-strain response of an isotropic, homogeneous elastic layer of
finite thickness to some surface load. Arbitrary simplifying assumptions are then applied to this
system of equations to produce a reduced system of simplified equations that are then solved for
the desired result.
The attraction of models developed using this approach is that the coefficients of the equation
defining the load-displacement response of the simplified elastic layer are always well defined in
terms of problem parameters. The drawback of such models is that the overall simplified elastic
continuum is not readily and easily modeled using commercially-available structural analysis
software.
The net result of the individual positive and negative aspects of both approaches for subgrade
model development has been that the numerous advanced subgrade models inherently more
accurate than Winkler's Hypothesis that have been identified over the past 70 years (see Table 1)
have seen relatively little practical application. As a result, practitioners and researchers alike
continue to struggle with using Winkler's Hypothesis as a subgrade model simply out of necessity
and not for its technical superiority or even adequacy.
Recent Research
In an attempt to overcome of the implementation impasse of the past 70 years and advance the
state-of-practice with subgrade models and SSI analysis to better match the state-of-knowledge,
the last few years have seen a new collaborative research effort between the authors of this report.
The key aspect of this effort was to approach subgrade-model development from a new
conceptual perspective to produce a hybrid model that has the best features but none of the
drawbacks of the traditional mechanical and simplified-continuum approaches when used alone.
This new model is called the Modified Kerr-Reissner (MK-R) hybrid subgrade model. Because of
its unique hybrid origins the MK-R Model is believed to be the first advanced subgrade model
that is practical for immediate use using routine structural and geotechnical engineering tools
available at the present time to both the average practitioner as well as researcher.
MODIFIED KERR-REISSNER HYBRID SUBGRADE MODEL
Introduction
A detailed presentation of the theoretical background and development of the MK-R Model can
be found in Horvath and Colasanti (2011a). Only key developmental aspects are discussed in this
report. The focus of this report is the geotechnical aspects of implementing this model in practice
(structural engineering issues in terms of software implementation are covered in detail in
Colasanti and Horvath 2010).
Theoretical Basis
Development of the MK-R Model was made possible by the observation that subgrade models
of identical relative accuracy and form of their governing load-displacement equation (expressed
in a generic, generalized manner by Eq. 1) can be developed using both the mechanical and
simplified-continuum approaches for subgrade modeling (Horvath 1979, 1988, 1989). This can be
seen in Table 1 that contains all known subgrade models developed to date. In the specific case of
Soil-Structure Interaction Research Project
"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis - Part III: Parameter Assessment"
Manhattan College Department of Civil and Environmental Engineering Report No. CEEN/GE-2011-1

4
developing the MK-R Model, the newly-identified Modified Kerr (a.k.a. Horvath-Colasanti)
mechanical subgrade model that was identified for the first time in the published literature in
Horvath and Colasanti (2011a) was synergistically combined with the Reissner Simplified
Continuum (RSC) Model. Both the Modified Kerr/Horvath-Colasanti and RSC models are two
orders of accuracy greater than Winkler's Hypothesis ("accuracy" in this case relates to how many
terms in Eq. 1 beyond p and W appear in the load-displacement equation defining a model's
behavior). Most importantly, the all-important spring-coupling that is lacking in and is the most
serious shortcoming of Winkler's Hypothesis is inherently included in the MK-R Model.
Constituent Models
Modified Kerr/Horvath-Colasanti Model
The Modified Kerr/Horvath-Colasanti Model (referred to hereinafter as simply the Modified
Kerr Model) is shown in Fig. 1 and consists of, from top to bottom, an upper layer of independent
axial springs of stiffness ku underlain by a membrane under constant tension T underlain by a
lower layer of independent axial springs of stiffness kl. (note that this lower spring layer is the
Winkler subgrade that always forms the first building block for all advanced mechanical subgrade
models). The equation governing the relationship between the subgrade reaction, p, that
represents the vertical stress applied normal to the subgrade surface and normal displacement
(settlement for the model orientation shown in Fig. 1) of the subgrade surface, W, is
T
p
ku + kl

2
k k
p = u l

ku + kl

Tk u
W

ku + kl

2
W .

(2)

Note that Eq. 2 conforms to the general form of all subgrade-model equations (Eq. 1). Note also
the point made previously that although the physical composition of this model is clear and could
thus be easily implemented in commercially-available structural analysis software how the
parameters ku, kl, and T could be correlated to fundamental subgrade properties in a given
application is neither intuitive nor obvious. As noted previously, it is this latter issue that has
constrained the use of more-advanced mechanical subgrade models in routine practice even
though they have existed since at least 1940.

Figure 1. Modified Kerr/Horvath-Colasanti mechanical subgrade model.

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

Figure 2. Reissner Simplified Continuum subgrade model.


Reissner Simplified Continuum Model
The RSC Model is shown in Fig. 2 and consists of an isotropic, homogeneous layer of linearelastic material with Young's modulus, E, and finite thickness, H, underlain by a rigid base. The
original solution developed by Reissner (1958) was for the problem variant shown in Fig, 2(a)
and assumed only the subgrade reaction, p, applied directly to the subgrade surface. The resulting
equation that defines the load-displacement behavior of this model is

GH 2 2
E
GH 2
p = W
p
W
H
3
12 E

(3a)

Note again that Eq. 3a conforms to the general form of all subgrade-model equations (Eq. 1).
Note also the point made previously that although the physical parameters appearing in the
coefficients of Eq. 3a are clear and could thus be readily evaluated (in concept) on a site- and
application-specific basis implementation of this model in commercially-available structural
analysis software is either difficult or impossible depending on the specific software package
used. It is this latter issue that has constrained the use of simplified-continuum subgrade models
in routine practice even though such models have been available since the 1950s. The primary
author has experienced this practical implementation constraint firsthand as a result of extensive
research into the RSC going back to the 1970s (Horvath 1979).
Although a linear-elastic plate overlying the elastic layer as shown in Fig. 2(b) is not an
essential component of the RSC such a plate is present in many applications of practical interest.
As it turns out, the presence of a plate influences boundary conditions and the resulting equation
governing the behavior of the RSC Model. The solution for a perfectly-smooth interface between
Soil-Structure Interaction Research Project
"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis - Part III: Parameter Assessment"
Manhattan College Department of Civil and Environmental Engineering Report No. CEEN/GE-2011-1

6
the underside of a plate and the surface of the elastic layer is the same as Eq. 3a (Horvath 1983).
The other limiting case of a perfectly-rough plate-elastic layer interface was derived in Horvath
(1979) and the governing equation for the subgrade component only is

GH 2 2
GH 2
t
E
p = W
p

H
2 3 2 H
12 E

2
W .

(3b)

Synthesis
The MK-R Model is developed by noting that Eq. 2 defining the behavior of the Modified Kerr
Model is identical in form to Eqs. 3a/3b defining the behavior of the RSC Model. This allows the
coefficients of these equations to be equated so that the abstract mechanical parameters in Eq. 2
can be expressed explicitly in terms of the elastic and problem-geometry parameters in Eqs.
3a/3b. These relationships are given in Appendix I of this report. The important result is that the
abstract mechanical elements ku, kl, and T of the Modified Kerr Model (Fig. 1) can be defined
explicitly and exactly on a site- and application-specific basis using the fundamental subgrade
properties E, G, and H (Fig. 2). This is the primary contribution of the RSC Model to the MK-R
Model, to allow rational conceptual and numerical assessment of subgrade model parameters on a
site- and application-specific basis.
The contribution of the Modified Kerr Model to the MK-R Model is that the physical model
defining the behavior of the Modified Kerr Model (Fig. 1) can be easily implemented in
commercially-available structural analysis software. Specific guidelines on how to do this are
presented in detail in Colasanti and Horvath (2010) for ANSYS Version 11.0.
In summary, the new MK-R hybrid subgrade model consists of the mechanical visualization
shown in Fig. 1 being implemented into commercially-available structural analysis software (or
special application-specific software if desired) with model coefficients and parameters defined
by the physical model shown in Fig. 2 and evaluated per Appendix I. Specific suggestions for
performing the parameter-assessment phase of a project are the primary focus of this report and
are addressed in the following sections using three mat-foundation case histories for illustration.
Parameter Assessment
Overview
The first step in any project in which the MK-R Model will be used is to perform a sitecharacterization study to define subsurface stratigraphy and several key soil properties (unit
weight, elastic parameters) for each stratum. This is shown qualitatively in Fig. 3(a). As will be
seen using case-history applications, as an alternative to evaluating the relevant elastic properties
directly when fine-grain soils are involved it is possible as an alternative to use interpreted results
from standard oedometer (one-dimensional consolidation) tests to calculate the necessary elastic
properties, at least for an analysis of drained (fully-consolidated) conditions.
The results of the site-characterization study (Fig. 3(a)) are then transformed using engineering
judgment and calculations into the idealized multi-layer linear-elastic system shown qualitatively
in Fig. 3(b). This system consists of an arbitrary n number of artificial layers, each of which is
assumed to consist of an isotropic, homogeneous, linear-elastic material. The Young's modulus,
Ei, and shear modulus, Gi, for the material in each artificial layer i is determined from the
compressibility data obtained during the site-characterization phase.

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

Figure 3. Steps in MK-R Model parameter evaluation.


The next and final step is to convert the system shown in Fig. 3(b) to the one shown in Fig. 3(c)
which is an equivalent single layer of isotropic, homogeneous, linear-elastic material with
constant elastic parameters and underlain by a rigid base. The thickness of this layer, H, is
defined as the depth below foundation level to some effective rigid base. Additional comments as
well as detailed suggestions for transforming the idealized model shown in Fig. 3(b) to that
shown in Fig. 3(c) are given in subsequent sections of this report.
Once the parameters shown in Fig. 3(c) have been determined the equations in Appendix I are
used to calculate the unique values of the mechanical elements for the Modified Kerr Model
shown in Fig. 3(d) as it is the physical model shown in Fig. 3(d) that is implemented into
commercially-available structural analysis software using the guidelines presented in Colasanti
and Horvath (2010). Alternatively, application-specific software could be created to solve the
problem, e.g. Horvath (2009a).
Estimation of Depth to Rigid Base
Experience has shown that a reasonably accurate determination of the thickness, H, of the layer
shown in Fig. 3(c) is critical to the successful implementation of the MK-R Model. In practice,
this depth can be less than the depth to an actual, physical, relatively-rigid base identified during
the site-characterization process. In addition, there will be situations where a relatively rigid
stratum many not be clearly apparent. Consequently, this particular issue will be discussed here in
some detail and both situations are illustrated subsequently using case histories.
It is important to note that this sensitivity of results obtained using the MK-R Model to the
assumed depth-to-rigid-base should not be perceived as a unique disadvantage of this model. It is
well known that that calculated settlements obtained using even exact theory of elasticity
Soil-Structure Interaction Research Project
"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis - Part III: Parameter Assessment"
Manhattan College Department of Civil and Environmental Engineering Report No. CEEN/GE-2011-1

8
solutions are sensitive to the assumed depth to rigid base (Mayne 2005). Thus this is an issue that
comes up in geotechnical applications using other analytical methodologies. As will be seen,
because this is such an important issue for successful application of the MK-R Model particular
attention has been paid to developing a rational analytical methodology for its assessment.
For the purposes of the MK-R Model, the effective rigid base is defined as the depth at which
settlements caused by subgrade loading from the foundation element (e.g. mat as in the case
histories presented herein) can be taken to be zero. Thus this depth will be the smaller of either
the depth to an actual physical stratum of material that is relatively rigid compared to overlying,
more-compressible materials or an equivalent rigid base determined by some analytical method.
Geotechnical research in recent years has clearly demonstrated that traditional rules-of-thumb
concerning the relative depth below a shallow foundation (footing, mat, etc.) at which settlement
caused by the foundation is for all practical purposes zero are not valid. These rules-of-thumb
were, in general, based on a) theory-of-elasticity solutions for systems where Young's modulus is
constant with depth (a situation not typical of actual subgrades) and b) a consideration of stress
change only and not absolute stresses. For example, for decades it has been assumed that the
'depth-of-influence' of a shallow foundation (conceptually equivalent to the depth-of-rigid-base as
defined herein for the MK-R Model) is twice the width of a square loaded area and four times the
width of an infinite strip. However, recent research (Burland and Burbridge 1985, Berardi and
Lancellotta 1991, Charles 1996) has shown collectively and unequivocally that this depth is not
constant for a given loaded-area shape but highly dependent on both the relative (to overburden
stresses) magnitude of the applied load as well as the absolute width of the loaded area, with
decreasing relative depth of influence with decreasing load and/or increasing width. So, for
example, a square mat foundation that is, say, 20 metres (66 ft) wide might have a depth-ofinfluence of perhaps only one-half its width (10m (33 ft)) which is significantly less than the
traditional rule-of-thumb of twice its width (40 m (131 ft)).
After some investigation of alternatives, the MT Method (Charles 1996) was chosen for the
purpose of estimating the depth H in Fig. 3(c) although some enhancements to and extension of
this method were made by the primary author as discussed subsequently and described in detail in
Appendix II of this report. Use of the MT Method for this purpose should be considered tentative,
interim, and conditional and subject to future research and verification. This is because of both
the relative novelty of the MK-R Model and the fact that the MT Method was developed
primarily for use with large areal fills and not SSI applications. However, the results obtained
using the MT Method for the case histories presented subsequently yielded estimates of H
consistent with what would be expected from results given by both Burland and Burbridge (1985)
and Berardi and Lancellotta (1991) for work more directly related to shallow foundations. In
addition, the calculated settlements obtained using the MK-R Model with the MT Method to
estimate the depth-to-rigid-base compared favorably to measured settlements for these case
histories as will be shown subsequently.
From a practical perspective, solution of the equations presented in Charles (1996) is greatly
facilitated by computer solution. A proprietary code named MTH developed by the primary
author (Horvath 2009b) was used for this purpose for the results reported in this report.
Estimation of Equivalent Single-Layer Elastic Parameters
The MT Method does not explicitly address the issue of determining the equivalent single-layer
elastic parameters (Fig, 3(c)) for a layered system as shown in Fig. 3(b). Thus the MT Method
was extended by the primary author to do this using a strain-weighted approach that is an
accepted and successful methodology in similar applications in geotechnical engineering, e.g.
Fraser and Wardle (1976). The details for how this is done are given in Appendix II. This
extended MT Method is referred to as the MTH Method.
Soil-Structure Interaction Research Project
"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

9
Example Applications Using Case Histories
Introduction and Overview
Despite the widespread use of mat foundations in practice and the publication of numerous case
histories over the years (Hemsley (2000) provides a good summary of numerous previouslypublished cases) there are surprisingly few case histories that are usable for studies such as the
present one. Unlike the situation with spread footings or, especially, deep foundations where
individual foundation elements can be load tested relatively easily and quickly to provide data for
analytical comparisons, an adequately documented mat foundation is a very extensive and
relatively expensive effort that requires monitoring numerous points not only on the mat but,
depending on the subsurface conditions, within the underlying subgrade as well during all phases
of construction and sometimes for several years afterward as in two of the case histories
discussed subsequently. A further and significant complication with finding suitable mat
foundation case histories for assessment involves the fact that mat loading is very complex, not
only because there are numerous load points to identify and consider (not just one load as with a
footing or deep foundation) but also because the loads themselves are not intuitively obvious due
to superstructure interaction effects. It has long been established that for all but the shortest
buildings or simplest non-building structures that the superstructure supported by a mat will have
a significant effect on mat loads (both magnitude and distribution) and resulting mat settlements
due to load redistribution that results from vertical displacements and deformations of the
superstructure (Burland et al. 1977, The Institution of Structural Engineers 1989, Banavalkar
1995). Very few published mat case histories provide even the most basic information concerning
the superstructure to allow this important factor to be considered.
Despite these difficulties, three case histories have been found that are usable to demonstrate
the specific details for parameter assessment and application of the MK-R Model in practice. All
calculated results presented herein were obtained using this model implemented in ANSYS
Version 11.0 following the detailed guidelines presented in Colasanti and Horvath (2010).
Case History A: Georgia State University (GSU) Building B
This 13-story dormitory building located in Atlanta, Georgia was the subject of a paper by
Mayne (2005). The reported site geologic profile consists of residual soils (fill underlain by
undisturbed) transitioning to weathered bedrock and finally sound bedrock. The groundwater
table is approximately 7.5 m (25 ft) below grade. The mat supporting this building is 18 m (60 ft)
by 104 m (340 ft) in plan dimensions, 1070 mm (3.5 ft) thick, and founded approximately 2.3 m
(7.5 ft) below grade.
This structure was apparently constructed in the early 1990s and settled significantly more
during and after construction (approximately 250 mm (10 in) maximum) than had been
anticipated by the original designers (reportedly 45 mm (1.8 in) maximum). The focus of Mayne's
paper was an independent, after-the-fact reassessment of estimated settlements using an analytical
methodology based on the theory of elasticity (Mayne and Poulos 1999) and Young's moduli
measured in-situ using a flat-plate dilatometer (DMT).
Because of the specific analytical methodology used, Mayne's calculated results were relatively
sensitive to the assumed depth to a rigid base, a situation identical conceptually to that faced
when using the MK-R Model as was discussed previously. Defining an actual, physical rigid base
was particularly difficult for this site because of geologic conditions that are inherently
gradational and variable, a situation quite typical whenever residual soils grading to sound
bedrock are encountered as in the Piedmont region of the eastern U.S.A. within which this site is
located Because of this uncertainty, Mayne presented results for several different assumed
depths-to-rigid-base.
Soil-Structure Interaction Research Project
"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis - Part III: Parameter Assessment"
Manhattan College Department of Civil and Environmental Engineering Report No. CEEN/GE-2011-1

10
However, the most significant aspect of this case history in terms of its use for the study
reported in this report is that very simplistic modeling assumptions were made by Mayne. First of
all, there was no attempt to define and model actual column locations and loads or even model the
actual stiffness of the mat. Rather, a perfectly-flexible loaded area with a uniform applied stress
of 150 kPa (3 ksf) was assumed. In addition, mat-superstructure interaction effects were
neglected. Based on published research cited previously in this report, these effects would be
expected to be noticeable for a 13-story residential-type building that is presumably of some type
of frame construction. Nevertheless, the results presented by Mayne were useful because they
demonstrated that the order-of-magnitude of settlements actually observed could have been
anticipated even using the very simple analytical model used as long as relatively accurate
estimates of subgrade compressibility (Young's moduli obtained with appropriate in-situ testing
in this case) are made.
The analyses performed for the present study using the MK-R Model utilized Mayne's
interpretation of the DMT data for this site that Young's modulus was reasonably constant with
depth with a magnitude of 8500 kN/m2 (180 ksf). Because there was insufficient information
provided in Mayne's paper to estimate actual column locations and loads or incorporate
superstructure interaction effects a uniformly loaded mat was assumed. However, mat stiffness
was included in the analyses and assumed to be one-quarter of the stiffness of the uncracked, asbuilt section reported in Mayne's paper. A reduced stiffness was used based on calculations by the
contributing author that showed there was likely cracked-section behavior under the assumed
uniform loading as well as to account for time-dependent reduction of the mat stiffness due to
creep and effective Young's modulus reduction of the PCC.
One of the benefits of using the primary author's MTH Method for parameter assessment for
the MK-R Model is that it is not necessary to assume a depth-to-rigid-base beforehand as was
necessary for Mayne to do using the Mayne and Poulos (1999) methodology. The MTH
algorithm estimates an equivalent depth-to-rigid-base and this is compared to the depth to an
actual, physical rigid base that may exist in a particular case. The lesser of the equivalent and
actual depths is used for H (Fig. 3(c)) in the MK-R Model. This approach of solving for a depthto-rigid-base as opposed to assuming one a priori is particularly useful in cases such as at this site
where the actual depth to a rigid base is ill-defined because of the geology inherent in sites such
as this underlain by residual soils that transition gradationally and not abruptly to sound
underlying bedrock.
For this case history, the effective depth to rigid base was calculated using the MTH Method to
be approximately 20 m (66 ft) below the ground surface or 17.7 m (58 ft) below the mat-subgrade
interface. This depth falls within the range of 12 m (40 ft) to 24 m (79 m) depths below the matsubgrade interface analyzed by Mayne and is remarkably close to the 18 m (59 ft) depth that
Mayne found provided the best agreement overall between calculated and measured settlements.
Fig. 4 is a plot of measured settlements as reported by Mayne versus those calculated for this
study. Note that the section through the mat is taken on a diagonal as opposed to in a moreconventional transverse or longitudinal direction. This was done to be consistent with the
information presented in Fig. 9 of Mayne's paper. The agreement between calculated and
measured results is considered to be very good considering all the approximations involved with
the way mat loading was simulated and the fact that mat-structure interaction effects were
neglected. Note that settlements were also calculated assuming a perfectly-flexible mat to not
only provide a direct comparison with Mayne's analytical assumptions but also to illustrate the
sensitivity of the calculated results to mat-stiffness assumptions, at least for this case history.

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

11

Figure 4. Comparison of calculated versus measured results for Case History A.


Massachusetts Institute of Technology (MIT)
Introduction and Overview
DeSimone and Gould (1972) presented an unusually well-documented and detailed discussion
of two mat-supported buildings (the Whitaker Laboratory and Chemistry Building) that were
constructed at MIT in Cambridge, Massachusetts during the same general timeframe (1960s).
Their published work remains a model for a thorough, well-presented mat foundation case history
paper even four decades later.
Although these two buildings are broadly similar in terms of geometry, design details, and
construction process their foundation loadings are qualitatively quite different. As a result, one
structure (Whitaker Laboratory) exhibited the classical sagging (dish-shaped) settlement pattern
whereas the other (Chemistry Building) displayed the less-common hogging pattern. Because of
the distinctly different loading and settlement patterns for these two structures they were treated
as separate case histories for the purposes of the present study and separate analyses were
performed for each. However, because the two buildings are located relatively close to each other

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis - Part III: Parameter Assessment"
Manhattan College Department of Civil and Environmental Engineering Report No. CEEN/GE-2011-1

12
and share an essentially identical site geology before discussing the two structures independently
issues common to both will be discussed first.
Site Characterization
The portion of the MIT campus within which these two buildings are located was apparently
once located either in the bed of the Charles River and/or within the marginal wetlands along the
north bank of the river. However, by the time these structures were constructed in the 1960s the
area had apparently been elevated above river level by filling some time earlier.
Fill and underlying organic soils (organic clay and peat) extend down to approximately 10 m
(30 ft) below the existing ground surface. Below that lies the well-known Boston Blue Clay
(BBC). Both buildings were essentially founded at the top of the BBC stratum. At the site of the
Whitaker Laboratory the thickness of the BBC stratum is uniform and approximately 21 m (70 ft)
thick. However at the Chemistry Building site the thickness of the BBC stratum varies somewhat,
from about 21 m (70 ft) to 25 m (80 ft) going from north to south toward the present-day channel
of the river.
At both sites the BBC stratum is underlain by a relatively thick sequence of coarse-grain soils.
These are underlain in turn by bedrock. Groundwater in the area is relatively shallow,
approximately 3 m (10 ft) below the existing ground surface.
From the perspective of analyzing the behavior of the mats supporting these two structures the
top of the coarse-grain stratum underlying the BBC stratum acts as a reasonably well-defined
physical rigid base. Therefore part of the parameter development for the MK-R Model was to
determine if the rigid base for analytical purposes was this actual physical rigid base or an
effective rigid base located at a shallower depth. Note that this is conceptually different than Case
History A where a physical rigid base was ill-defined for the site.
Settlements of both structures were influenced solely by compressibility of the BBC. Data
provided in DeSimone and Gould indicate that prior to construction this stratum was
overconsolidated at its top, with the yield stress decreasing with depth until the soil became
essentially normally consolidated in its lower portions. This is a very classic pattern of stress
history for fine-grain soil strata such as the BBC that are associated with Pleistocene glaciation
and subsequent Holocene sea-level rise and tectonic isostasy in the northeastern U.S.A. However,
both structures under discussion were designed so that the net vertical effective stress applied at
foundation level after construction would be slightly less than the vertical effective overburden
stress that existed prior to construction. Therefore all vertical stresses imposed on the BBC by
these structures were within the recompression range of the BBC. This was likely an important
design criterion for these structures in order to keep post-construction total and differential
settlements within acceptable limits.
For the purposes of developing the MK-R Model parameters, the Young's moduli within the
BBC stratum were calculated for the present study based on laboratory testing as opposed to
being measured directly by in-situ testing as was the case for Case History A. Thus the procedure
used for the MIT case histories (referred to subsequently as "B" and "C") illustrate another
conceptual approach that can be used for evaluating the MK-R Model parameters.
The procedure used in the present study to determine the required soil properties began by first
dividing the BBC stratum into artificial layers (5 ft (1.5 m) thick was used in this case) as shown
conceptually in Fig. 3(b). Then the theoretical relationship given in Table 12.2 of Lambe and
Whitman (1969) relating the recompression ratio (as given by DeSimone and Gould based on
laboratory oedometer data for the MIT site as well as field observations on other, earlier projects
in the Boston area) and average applied vertical effective stress estimated for that layer (based on
overburden stresses and applied stresses from the building) to the drained constrained modulus
was evaluated numerically for each layer. Note that an essential part of the process is to make an
accurate assessment of the average operative vertical effective stress magnitude for each artificial
Soil-Structure Interaction Research Project
"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

13
layer so that appropriate operative values of moduli are determined for each artificial layer. The
drained Young's modulus for each artificial layer was then calculated based on the drained
constrained modulus and an assumed value of Poisson's ratio. Note that only the drained Young's
moduli were calculated and used for the present study as calculated-versus-measured settlement
comparisons were made only for the long-term case several years after construction when
piezometric data at the site indicated primary consolidation was completed. On other projects
involving fine-grain soils it may be desirable to perform a separate set of analyses assuming
undrained Young's moduli to assess behavior in a near-term timeframe.
Once the operative value of drained Young's modulus was determined for each artificial layer
the MTH methodology outlined in Appendix II was used to determine the equivalent Young's
modulus, E, and depth to rigid base, H, for the overall system (Fig. 3(c)) so that the MK-R Model
parameters (Fig. 3(d)) could be evaluated as summarized in Appendix I.
Mat Weight
Before discussing the specific analyses and results for the two MIT case histories there needs to
be a discussion concerning mat weight, specifically, how it should be handled in general in mat
analysis and design. As will be seen, the issue of mat weight turned out to be significant for both
MIT case histories when it came to comparing measured and calculated settlements. Research and
experience indicates this can be an important factor for mat foundations in general (Burland et al.
1977) and thus the role of mat weight should always be considered carefully and explicitly on a
site- and application-specific basis. As will be seen, for most mat-supported structures there are
usually several distinct issues related to mat weight and sometimes these can be complex and
conflicting so that multiple analyses are required for a given structure in order to properly
consider all of the conditions that may exist at various times during its life.
The specific issues considered with regard to mat weight involve a complex relationship
between numerous factors including the:

time it takes to excavate and dewater (as necessary) a site down to the planned foundation
level;

magnitude of subgrade heave that may occur during the excavation and dewatering process
which, in turn, depends on both the compression and consolidation characteristics of the
subgrade soils;

time it takes to form and pour the mat PCC;

time it takes for the mat PCC to transition from a fluid to a solid capable of resisting flexural
stresses;

compression and consolidation characteristics of the subgrade soils under the vertical stress
imposed by the weight of the mat PCC;

structural composition of the superstructure frame (steel versus PCC) as this will affect
whether or not the frame components have material properties and concomitant flexural
stiffnesses that are time independent or dependent;

construction of the superstructure frame that will both affect and be affected by the
superstructure-mat interaction;

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis - Part III: Parameter Assessment"
Manhattan College Department of Civil and Environmental Engineering Report No. CEEN/GE-2011-1

14

construction of the floors that will be affected by differential settlement;

application of the architectural finishes (curtain wall/cladding, interior partitions) that will be
affected by differential settlement; and

occupancy and/or use of the structure for its intended purpose, and whether or not the loads
resulting from that use are reasonably constant or significantly variable over time.

As a result of this very complex sequence of events that is, in general, unique for a given site
and structure several distinct states can be defined as a function of time for the overall
construction process of a mat-supported structure:

After excavation to the planned foundation level the mat is formed and PCC poured.
Depending on the size of the mat, the actual pour may extend over a period of many hours.
The subgrade soils begin to compress immediately under the weight of the fluid PCC
producing settlement at the mat-subgrade interface. However, initially while the PCC is still
fluid no flexural stresses are generated within the mat as a result of any differential
settlements induced by the weight of the still-fluid PCC. How much total settlement occurs
during this stage depends on the mat dimensions as well as the compression and
consolidation characteristics of the subgrade soils.

The mat PCC begins to set and then cure. How quickly this occurs is a complex function of
mat dimensions, temperature, temperature controls used (if any), and the chemistry of the
PCC mix itself in terms of cement type plus additives used (if any). Depending on the
characteristics of the subgrade soils further settlement may occur during this stage even
though the load applied to the subgrade is essentially constant during this time. Once the PCC
begins to set any differential settlements that occur after that point will create flexural stresses
within the mat.

The mat PCC is sufficiently cured so that construction of any below-grade walls and erection
of the superstructure to be supported on the mat can begin. Note that this is typically when
settlement measurement points are established on the top of the mat. This means that any
excavation heave and subsequent recompression settlement under mat weight that has
occurred prior to this time will be unknown unless geotechnical instrumentation has not only
been placed at appropriate locations on and within the subgrade soils but has survived the
construction process up to this point in time.

As the superstructure is erected (framing, floors, and architectural walls/partitions in the case
of a building) and concomitant loads are applied to the mat additional total settlements will
occur. Any differential settlements that develop will affect both the mat as well as the
components of the superstructure. Note that the relationship between load application and
concomitant settlement will always be structure-specific depending on the compression and
consolidation characteristics of the subgrade soils. In addition, if temporary dewatering has
been performed as part of the excavation to foundation level then somewhere during this
stage the dewatering is likely to be stopped allowing the piezometric regime around and
below the mat to begin to return to pre-construction levels, affecting the effective stress
regime of the subgrade in the process. How quickly this piezometric recovery takes will be
affected by several factors, the primary one being the permeability of the subgrade soils.
Changes in effective stresses within the subgrade soils as well as uplift water pressures on the

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

15
bottom of the mat will complicate the progress and interpretation of any measured
settlements.

Finally, as the structure is occupied or otherwise used for its intended purpose additional total
and differential settlements will occur. Again, how rapidly this progresses will depend on the
compression and consolidation characteristics of the subgrade soils. Depending on the
specific use of the structure and the relative magnitude and time-dependent variability of liveto-dead loads on the mat it is possible that there can be cycles of mat heave (rebound) and
settlement throughout the life of the structure.

The point and conclusion of this extended discussion is that it is rarely, if ever, clear-cut as to
whether mat weight should be unilaterally included or neglected in the analysis or design of a
mat-supported structure. In part this is because the answer to this question depends on the
intended use of the calculated results. For example, it should be apparent from the preceding
discussion that the relevance of mat weight will vary not only as a function of the compression
and consolidation characteristics of the subgrade soils but will also depend on whether the
analysis is performed to evaluate bending moments in the mat; compare calculated-to-measured
settlements of the mat; or evaluate the superstructure for differential settlements. Given the
uncertainties involved it is likely that in most cases that analyses both without and with mat
weight would or at least should be performed for a given structure and the results used to present
a range of behaviors whether for design or comparison to observations. In fact this is what was
done for the present study for the two MIT case histories as the time-dependent behavior of the
BBC subgrade supporting these mats greatly complicated interpretation of both calculated and
measured results. However, for each structure an opinion and basis for that opinion is presented
by the authors as to which case (without or with mat weight) is likely to be closer to the measured
settlements for these particular mats.
Case History B: Whitaker Laboratory
This is an eight-story building with two basement levels that is 18.3 m (60.0 ft) by 67.1 m
(219.7 ft) in plan dimensions. The mat supporting this building is 1140 mm (3.75 ft) thick and
founded 10.3 m (33.7 ft) below grade. Fig. 5 shows a transverse cross-section through the mat
with the estimated dead-plus-live service loads from the superstructure as given by DeSimone
and Gould that were used in the present study. Note that although the stiffness of the mat was
included in the analyses performed for the present study most mat-superstructure interaction
effects were neglected as there was insufficient information concerning the superstructure
presented by DeSimone and Gould that would have allowed such effects to be modeled.
However, it was possible to at least approximate the rotational restraint to the mat caused by the
below-grade exterior walls that extend in the longitudinal direction along each edge of the mat
and this was included in the structural model used for the present study. These restraints were
modeled as rotational springs set 1 ft (300 mm) in from each edge of the mat and had a noticeable
effect on the calculated results.
The actual, physical rigid base (assumed to be the interface between the BBC and underlying
coarse-grain stratum) was, as noted previously, approximately 21 m (70 ft) below the matsubgrade interface at this site. Using the MTH Method outlined in Appendix II, the depth to the
effective rigid base was estimated to occur at a slightly shallower depth, 19.2 m (63 ft) below
foundation level, with the equivalent homogeneous, isotropic, single-layer Young's modulus
estimated to be 21 MN/m2 (430 ksf). These properties were used to calculate the MK-R Model
parameters using the relationships given in Appendix I.

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis - Part III: Parameter Assessment"
Manhattan College Department of Civil and Environmental Engineering Report No. CEEN/GE-2011-1

16

Figure 5. Assumed mat loading for Case History B.


Analyses were performed using the MK-R Model both with and without mat weight. The
calculated results were compared with long-term measured settlements made at numerous
locations on the top of the mat. The authors believe the no-mat-weight calculated results are
likely to be closer to the measured as field measurements given in DeSimone and Gould suggest
the BBC had recompressed to nearly 100% primary consolidation under the mat weight before
measurement points on the mat were established. This opinion is supported by the settlement data
plotted in Fig. 6. The agreement between calculated and measured settlements is quite good,
especially for the no-mat-weight analysis case. It is interesting to note the relative significance of
mat weight in this case which emphasizes the point made previously that mat weight is something
that should be considered carefully on each project.

Figure 6. Comparison of calculated versus measured results for Case History B.


Soil-Structure Interaction Research Project
"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

17
Case History C: Chemistry Building
This is a five-story building with two basement levels that is 20.0 m (65.5 ft) by 67.1 m (279.5
ft) in plan dimensions. The mat supporting this building is relatively thin, only 762 mm (2.5 ft)
thick, and founded 9.3 m (30.4 ft) below grade. Fig. 7 shows a transverse cross-section through
the mat with estimated dead-plus-live superstructure service loads as given by DeSimone and
Gould that were used in the present study. Note that the foundation loading for this structure is
somewhat atypical in that the superstructure loads from above ground level are transferred across
almost the entire width dimension of the building to columns located at and close to the
longitudinal sides of the structure. Again, the analyses performed for the present study considered
mat stiffness but neglected mat-superstructure interaction except for the rotational restraint
provided by the below-grade exterior walls that extend in the longitudinal direction.

Figure 7. Assumed mat loading for Case History C.


As noted previously, the depth to the actual, physical rigid base (assumed to be the interface
between the BBC and underlying coarse-grain stratum) for this structure varied somewhat in the
longitudinal direction from approximately 21 m (70 ft) to 25 m (80 ft) below the mat-subgrade
interface at this site. Using the MTH Method outlined in Appendix II, it was estimated that the
effective rigid base was somewhat higher, 18.7 m (61 ft) below foundation level, with the
equivalent homogeneous, isotropic, single-layer Young's modulus estimated to be 18 MN/m2
(375 ksf). Note that these results are similar, but not identical, to those of the Whitaker
Laboratory (Case History B). These properties were used to calculate the MK-R Model
parameters using the relationships given in Appendix I.
Analyses were again performed using the MK-R Model both with and without mat weight
although the no-weight results are believed to better replicate the actual conditions for measured
settlements for the same reasons given previously for the Whitaker Laboratory. These calculated
results were again compared with long-term measured settlements of the mat that were made at
numerous locations located on the top of the mat as reported by DeSimone and Gould. The results
are shown in Fig. 8. Note that in this case the agreement between calculated and measured
settlements is not nearly as good as for the Whitaker Laboratory. Although the calculated
settlements, especially for the no-mat-weight case, are of the same magnitude and overall
hogging shape as the observed settlements the calculated settlements indicate significantly greater
differential settlements than were actually observed. Some discussion of the possible reasons for
this is warranted.

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis - Part III: Parameter Assessment"
Manhattan College Department of Civil and Environmental Engineering Report No. CEEN/GE-2011-1

18

Figure 8. Comparison of calculated versus measured results for Case History C.


To begin with, it is believed the calculated results obtained using the MK-R Model are correct
in terms of accurately representing outcomes based on the subsurface data and loads presented by
DeSimone and Gould. This opinion was reached after also analyzing this case history using a
proprietary computer code named G183 developed by the contributing author (Colasanti 2008)
that rigorously solves a layered elastic system. This program was used previously by the authors
to provide baseline results for evaluating the accuracy of the MK-R Model during an earlier phase
of the present study that was reported in Horvath and Colasanti (2011a). Although not shown
here, the results obtained for this case history using the G183 code essentially matched those
obtained using the MK-R Model that are depicted in Fig. 8. Thus the only conclusion that can be
reached is that there are other reasons for the difference between calculated and measured results
for this case history.
It is relevant to note that the Chemistry Building was built a few years after the Whitaker
Laboratory and the overall geotechnical and structural instrumentation, at least as presented by
DeSimone and Gould, for the Chemistry Building was not nearly as detailed and complete as that
for the Whitaker Laboratory. It is possible that because of the overall good experience with and
performance of the Whitaker Laboratory there was less perceived need for a similarly detailed
investigation for the Chemistry Building which was broadly similar except for the difference in
column layout and loading that produced a different settlement pattern. Thus it is possible the
measured-settlement record for the Chemistry Building is incomplete or deficient in some way. In
addition, it is possible the actual mat loading on the Chemistry Building differed from that
presented by DeSimone and Gould and used in the present study, or that the effects of matSoil-Structure Interaction Research Project
"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

19
superstructure interaction (neglected in the present analyses except as noted previously) had
significant influence on the actual performance of the Chemistry Building. This latter point is
thought to be quite possible given the relatively thin mat used for the Chemistry Building which
means the superstructure would likely have greater influence on the overall apparent mat stiffness
compared to the Whitaker Laboratory where the mat is more than three times stiffer than that of
the Chemistry Building. This superstructure-interaction influence would generally be to increase
the apparent mat stiffness and thus decrease differential settlements which is what was indeed
observed. In any event, regardless of which factor or combination of factors was/were the actual
cause(s) of the difference between calculated and measure results the results obtained using the
MK-R Model are considered to provide reasonable agreement with measured results considering
the uncertainties and analytical approximations involved.
SUMMARY AND CONCLUSIONS
For the better part of a century Winkler's Hypothesis has been recognized as being seriously
deficient as a subgrade model for SSI analysis. This is because it is inherently and seriously
flawed by not replicating the all-important mechanism of vertical shearing resistance (springcoupling) that always occurs within an actual subgrade. The result of this significant theoretical
shortcoming is that the Coefficient of Subgrade Reaction, which is normally a calculated result, is
forced to become a known input parameter called the Winkler Coefficient of Subgrade Reaction.
This is because the lack of inherent spring coupling in a Winkler subgrade must be compensated
by having the analyst estimate beforehand and input the effects of spring coupling rather than
letting the subgrade model determine the spring coupling as part of the solution process. This
fact, which appears to have been significantly underappreciated to date by both practitioners and
researchers, is why Winkler's Hypothesis has always had associated with it the fact that the
Winkler Coefficient of Subgrade Reaction defies rational theoretical evaluation.
The MK-R Model is believed to be the long-sought practical improvement to Winkler's
Hypothesis as a subgrade model for SSI analysis. It incorporates inherently the all-important
mechanism of spring coupling that replicates subgrade shearing into its mathematical formulation
but, more importantly, its unique hybrid derivation means that it is both straightforward to
implement in existing commercially-available structural analysis software as well as to evaluate
using state-of-practice geotechnical methodologies.
The theoretical basis of the MK-R Model was presented in detail in Horvath and Colasanti
(2011a) and detailed guidelines for implementing this model in commercially-available structural
analysis software were presented in Colasanti and Horvath (2010). This report focused on the
geotechnical aspects of parameter assessment for the MK-R Model using well-established site
characterization and analytical methodologies, and illustrated this using three case histories from
the published literature. It is felt the overall result of these several publications is that the validity
and practical utility of the MK-R Model has been reasonably established.
In addition, some of the peripheral concepts presented in this report such as the MTH Method
for estimating the depth-to-rigid-base in a layered system may prove useful for other analytical
applications such as the elasticity solution developed by Mayne and Poulos (1999) that was used
in Mayne (2005).
Finally, it is worth noting that the MK-R Model has potential for SSI applications beyond
traditional ones such as mat foundations to applications such planar geosynthetics used as tensile
reinforcement (Horvath and Colasanti 2011b) as well as deep foundations.
RECOMMENDATIONS
The presentation of the MK-R Model in this report has demonstrated that accurate results using
this model are highly dependent on accurate assessment of the model parameters, specifically, the
Soil-Structure Interaction Research Project
"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis - Part III: Parameter Assessment"
Manhattan College Department of Civil and Environmental Engineering Report No. CEEN/GE-2011-1

20
operative elastic moduli of the subgrade materials and depth-to-rigid-base for the overall
subgrade system. Thus it is strongly recommended that additional research be devoted to
continued investigation of how to best estimate the depth-to-rigid-base. Although the MTH
Method presented in this report appears to produce reasonable results that does not mean the
method could not be improved or perhaps even replaced completely by a better one.
In addition, although there has been substantial research in recent decades into sitecharacterization methodologies the importance for continued development and advancement of
these methodologies cannot be overstressed and is one made previously by the primary author for
other applications (Horvath 2011). In particular, estimating in-situ moduli, especially using in-situ
techniques for materials such as coarse-grain and residual soils that are essentially impossible to
sample and test using conventional laboratory methodologies, is a very important area of
continued research that is relevant to the use of the MK-R Model.
ACKNOWLEDGEMENTS
All ANSYS analyses performed for the analyses presented in this report utilized computer
hardware and software resources provided by the contributing author's employer, URS
Washington Division. This support is acknowledged with gratitude.
The use of software tradenames in this paper is for necessary identification purposes only and
does not constitute a recommendation or endorsement of that software by the authors or
Manhattan College.
APPENDIX I - PARAMETERS FOR MK- R MODEL
No Plate or Perfectly-Smooth Plate-Subgrade Interface
The original solution (Reissner 1958) developed for what would eventually be called the
Reissner Simplified Continuum (RSC) subgrade model assumed that a normal stress, herein
referred to as the subgrade reaction, p, was applied directly to the subgrade surface (Fig. 2(a)).
This is equivalent to saying there is no structural element (plate, etc.) on the subgrade surface. It
was subsequently shown (Horvath 1983)2 that the solution for an elastic plate (Fig. 2(b)) with the
limiting case of a perfectly-smooth plate-subgrade interface produces the same equation.
The equation defining the relationship between the subgrade reaction, p, and normal
displacement of the subgrade surface, W, for these cases is:

GH 2 2
E
GH 2
p = W
p
W .
H
3
12 E

(3a)

Equating the coefficients in Eq. 3a with those in Eq. 2:


T
p
ku + kl

2
k k
p = u l

ku + kl

Tk u
W

ku + kl

2
W

(2)

that defines the behavior of the Modified Kerr mechanical subgrade model produces the
following relationships:

Note that in this reference a coordinate system different from Reissner's was used and there are
typographical errors in portions of the derivation.
Soil-Structure Interaction Research Project
"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

21

ku =

4E
H

(4a)

kl =

4E
3H

(4b)

4GH
.
9

(4c)

T=

Eqs. 4a-c are what are used in practice to evaluate the parameters for the mechanical model
(Fig. 3(d)) implemented into commercially-available structural analysis software using the
outcomes of the geotechnical site characterization process shown conceptually in Fig. 3(a-c).
Perfectly-Rough Plate-Subgrade Interface
The other limiting boundary condition for the RSC of a perfectly-rough plate-subgrade
interface for an elastic plate of thickness t (Fig. 2(b)) was derived in Horvath (1979):

GH 2 2
GH 2
t 2
E
p = W
p

W .
H
2 3 2 H
12 E

(3b)

Equating the coefficients in Eq. 3b with those in Eq. 2

T
p
ku + kl

k k
2
p = u l
ku + kl

Tk u

W
ku + kl

2
W

(2)

for the Modified Kerr mechanical subgrade model produces the following relationships:

T=

ku =

E 4 H 3t

H H

(5a)

kl =

E 4 H 3t

3H H t

(5b)

GH 4 H 3t 4 H 3t

+
.
12 H 3H 3t

(5c)

Again, Eqs. 5a-c are what are used in practice when implementing the MK-R Model in
commercially-available structural analysis software (Fig. 3).
APPENDIX II - EXTENSION OF MT METHOD
The MT Method (Charles 1996) for determining the depth to an equivalent rigid base is based
on a simplistic one-dimensional-compression model for the subgrade beneath a loaded area.
Clearly, the assumption of one-dimensional compression becomes a better approximation of
Soil-Structure Interaction Research Project
"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis - Part III: Parameter Assessment"
Manhattan College Department of Civil and Environmental Engineering Report No. CEEN/GE-2011-1

22
reality as the width of the loaded area becomes relatively large which was the application of
primary interest to Charles.
The primary author used this concept to extend the MT Method and develop a strain-weighted
methodology for estimating the equivalent single-layer elastic properties (Fig. 3(c)) for a layered
elastic system (Fig. 3(b)). An outline of how this was done is presented in this appendix. The
resulting extended methodology is called the MTH Method.
To begin with, Charles defined the vertical strain, v, beneath a loaded area as

v =

v v z
=
D
D

(6)

where D = the drained constrained modulus of the material beneath the loaded area (assumed
constant for the entire subgrade by Charles); z = the depth beneath the loaded area; = unit
weight of the soil beneath the loaded area (assumed constant for the entire subgrade by Charles);
v = vertical stress beneath the loaded area; and v = the change in vertical stress beneath the
loaded area. Note that the implication is that the overburden stresses are effective and the soil unit
weight should be that which causes effective stresses.
One key element of the primary author's extended methodology is that the requirements of
constant modulus and soil unit weight for a subgrade are relaxed. Eq. 6 is still assumed to be
applicable for calculating vertical strain but is now used only to calculate the strain, vi, for an
arbitrary layer i of arbitrary thickness beneath a site (as shown conceptually in Fig. 3(b)) and with
the important distinction that the drained constrained modulus, D, is allowed to vary from layer to
layer. The drained constrained modulus, Di, for a typical layer i is calculated using the following
relationship from the theory of elasticity:

Di =

Ei (1 i )
.
(1 + i )(1 2 i )

(7)

Note that Ei and i for each artificial layer would have been determined previously during the
site characterization phase of the project either directly using in-situ testing as was done for Case
History A presented in this report or by calculation using laboratory test data as was done for
Case History B and C in this report3. Alternatively, if shear moduli, G, had been determined from
the site characterization process they could be converted into Young's moduli using the wellknown relationship for an elastic material:

G=

E
.
2(1 + )

(8)

The equivalent average drained constrained modulus, D*, and Poisson's ratio, *, for the
overall system shown in Fig. 3(c) are obtained using the following relationships based on a
simple assumption of strain-weighting:

Note that when using oedometer data as the basis for subgrade stiffness parameters as was done for Case
History B and C the drained constrained moduli are already known from the procedure explained in the
body of the report so the step of converting the Young's moduli back to drained constrained moduli using
Eq. 7 can, in principle, be skipped.
Soil-Structure Interaction Research Project
"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

23
n

D* =

(D )
i vi

i =1

i =1

(9)

vi

* =

( )
i =1

i vi

vi

(10)

Finally, the equivalent average Young's modulus, E*, for the system shown in Fig. 3(c) is
calculated by rearranging Eq. 7 to solve for E* using the results from Eqs. 9 and 10. The
equivalent shear modulus, G*, for the system shown in Fig. 3(c) is then obtained using Eq. 8.
Note that the averages shown in Eqs. 9 and 10 are for n artificial layers between foundation
level (not the ground surface) and the effective rigid base at a depth H below foundation level.
The latter depth is unknown initially and always needs to be determined on a site- and
application-specific basis using the MT Method. For the case histories analyzed for this study this
was done on a trial-and-error basis incorporated within a proprietary computer code named MTH
(Horvath 2009) that integrated both the MT Method for determining H and the primary author's
extension of that method as outlined in this appendix to handle layered systems with variable soil
properties to determine the equivalent single-layer elastic parameters into a single, seamless
analysis that is called the MTH Method.
REFERENCES
Note: Copies of references marked with a "@" can be downloaded in PDF file format from the
primary author's personal website, www.jshce.com. Other references are also available on the
Web as noted.
Banavalkar, P. V. (1995). "Mat foundation and its interaction with the superstructure." Design
and performance of mat foundations; State-of-the-art review, ACI, 13-49.
Berardi, R. and Lancellotta, R. (1991). "Stiffness of granular soils from field performance."
Gotechnique, 41(1): 149-157.
Burland, J. B., Broms, B. B., and De Mello, V. F. B. (1977). "Behaviour of foundations and
structures." Proc., 9th Int. Conf. on Soil Mech. and Fndn. Engr., Tokyo, 495-546.
Burland, J. B. and Burbridge, M. C. (1985). "Settlement of foundations on sand and gravel."
Proc., Inst. of Civ. Engrs., Part 1, 78: 1325-1381.
Charles, J. A. (1996). "The depth of influence of loaded areas." Gotechnique, 46(1): 51-61.
Colasanti, R. J. (2008). Program G183PC mat foundation analysis and design; User's manual;
Version 5.2.
Colasanti, R. J. and Horvath, J. S. (2010). "A practical subgrade model for improved soilstructure interaction analysis: Software implementation." Practice Periodical on Structural
Des. and Const., ASCE, 15(4): 278-286.@
DeSimone, S. V. and Gould, J. P. (1972). "Performance of two mat foundations on Boston blue
clay." Proc., Specialty Conf. on Performance of Earth and Earth-Supported Structures, ASCE,
Vol. 1/Pt. 2: 953-980.
Fraser, R. A. and Wardle, L. J. (1976). Numerical analysis of rectangular rafts on layered
foundations. Geotechnique, 26(4): 613630.

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis - Part III: Parameter Assessment"
Manhattan College Department of Civil and Environmental Engineering Report No. CEEN/GE-2011-1

24
Hemsley, J. A. (2000). "Developments in raft analysis and design." Design applications of raft
foundations, Thomas Telford, 487-605.
Horvath, J. S. (1979). A study of analytical methods for determining the response of mat
foundations to static loads. Ph.D. dissertation, Polytechnic Inst. of N. Y., Brooklyn, NY.
Horvath, J. S. (1983). "New subgrade model applied to mat foundations." J. of Geotech. Engr.,
ASCE, (109)12: 1567-1587.
Horvath, J. S. (1988). Historical review and critique of mathematical models for plate- and beamtype foundation element subgrades. Rpt. No. CE/GE-88-4, Manhattan Coll., Sch. of Engr.,
Civil Engr. Dept., Bronx, NY.
Horvath, J. S. (1989). "Subgrade models for soil-structure interaction." Proc., Fndn. Engr. Cong.,
ASCE, 599-612.
Horvath, J. S. (2002). Soil-structure interaction research project; Basic SSI concepts and
applications overview. Rpt. No. CGT-2002-2, Manhattan Coll., Sch. of Engr., Ctr. for
Geotechnology, Bronx, NY.@
Horvath, J. S. (2009a). SSIH: A computer program for soil-structure interaction analysis of
horizontal foundation elements; User instructions and guidelines.
Horvath, J. S. (2009b). MTH: A computer program for calculating the equivalent depth to rigid
base and pseudo-homogenous/isotropic single-layer elastic parameters of a layered system
using an extended version of the Charles MT model; User instructions and guidelines.
Horvath, J. S. (2011). "Improved geotechnical analysis through better integration and dynamic
interaction between site characterization and analytical theory." Proc., Geo-Frontiers 2011,
ASCE Geo-Institute/IFAI/GMA/NAGS, (in press).@
Horvath, J. S. and Colasanti, R. J. (2011a). "A practical subgrade model for improved soilstructure interaction analysis: model development." Int. J. of Geomech., ASCE, 11(1): 59-64.@
Horvath, J. S. and Colasanti, R. J. (2011b). "New hybrid subgrade model for soil-structure
interaction analysis: Foundation and geosynthetics applications." Proc., Geo-Frontiers 2011,
ASCE Geo-Institute/IFAI/GMA/NAGS, (in press).@
Lambe, T. W. and Whitman, R. V. (1969). Soil mechanics. Wiley.
Mayne, P. W. (2005). "Unexpected but foreseeable mat settlements on Piedmont residuum." Int.
J. of Geoengr. Case Histories, 1(1): 5-17.<http://casehistories.geoengineer.org>
Mayne, P. W. and Poulos, H. G. (1999). "Approximate displacement influence factors for elastic
shallow foundations." J. of Geotech. and Geoenv. Engr., ASCE, 125 (6): 453-460.
Reissner, E. (1958). "A note on deflections of plates on a viscoelastic foundation." J. App. Mech.
(25)/Trans. ASME (80): 144-155.
The Institution of Structural Engineers (1989). Soil-structure interaction; The real behaviour of
structures. U.K.

Soil-Structure Interaction Research Project


"A Practical Subgrade Model for Improved Soil-Structure Interaction Analysis: Parameter Assessment"
Manhattan College Civil and Environmental Engineering Department Report No. CEEN/GE-2011-1

Das könnte Ihnen auch gefallen