Wave Propagation in Structures Mechanical Engineering Series James F Doyle Online Ebook Texxtbook Full Chapter PDF

Wave Propagation in Structures
(Mechanical Engineering Series) James

F. Doyle
Visit to download the full and correct content document:
https://ebookmeta.com/product/wave-propagation-in-structures-mechanical-engineeri
ng-series-james-f-doyle/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Radio Wave Propagation Fundamentals, Second Edition

Artem Saakian
https://ebookmeta.com/product/radio-wave-propagation-
fundamentals-second-edition-artem-saakian/
Advanced Electromagnetic Wave Propagation Methods 1st

Edition Gonzalez
https://ebookmeta.com/product/advanced-electromagnetic-wave-
propagation-methods-1st-edition-gonzalez/
Wave Fields in Real Media: Wave Propagation in

Anisotropic, Anelastic, Porous and Electromagnetic
Media 4th Edition José M. Carcione
https://ebookmeta.com/product/wave-fields-in-real-media-wave-
propagation-in-anisotropic-anelastic-porous-and-electromagnetic-
media-4th-edition-jose-m-carcione/
Nonlinear Wave and Plasma Structures in the Auroral and

Subauroral Geospace 1st Edition Evgeny Mishin
https://ebookmeta.com/product/nonlinear-wave-and-plasma-
structures-in-the-auroral-and-subauroral-geospace-1st-edition-
evgeny-mishin/
Electromagnetic Vortices: Wave Phenomena and
Engineering Applications 1st Edition
https://ebookmeta.com/product/electromagnetic-vortices-wave-
phenomena-and-engineering-applications-1st-edition/
Aircraft Structures for Engineering Students (Aerospace

Engineering) 7th Edition Megson
https://ebookmeta.com/product/aircraft-structures-for-
engineering-students-aerospace-engineering-7th-edition-megson/
Fear No Evil Alex Cross 29 1st Edition James Patterson
https://ebookmeta.com/product/fear-no-evil-alex-cross-29-1st-
edition-james-patterson/
Design of Steel-Concrete Composite Structures Using

High-Strength Materials (Woodhead Publishing Series in
Civil and Structural Engineering) 1st Edition Richard
Liew
https://ebookmeta.com/product/design-of-steel-concrete-composite-
structures-using-high-strength-materials-woodhead-publishing-
series-in-civil-and-structural-engineering-1st-edition-richard-
liew/
Mastering Uncertainty in Mechanical Engineering 1st

Edition Peter F. Pelz
https://ebookmeta.com/product/mastering-uncertainty-in-
mechanical-engineering-1st-edition-peter-f-pelz/
Mechanical Engineering Series
James F. Doyle
Wave Propagation
in Structures
Third Edition
Series Editor
Francis A. Kulacki
Department of Mechanical Engineering
University of Minnesota
Minneapolis, MN, USA
The Mechanical Engineering Series presents advanced level treatment of topics
on the cutting edge of mechanical engineering. Designed for use by students,
researchers and practicing engineers, the series presents modern developments
in mechanical engineering and its innovative applications in applied mechanics,
bioengineering, dynamic systems and control, energy, energy conversion and energy
systems, fluid mechanics and fluid machinery, heat and mass transfer, manufacturing
science and technology, mechanical design, mechanics of materials, micro- and
nano-science technology, thermal physics, tribology, and vibration and acoustics.
The series features graduate-level texts, professional books, and research mono-
graphs in key engineering science concentrations.
More information about this series at http://www.springer.com/series/1161

James F. Doyle
Wave Propagation
in Structures
Third Edition
James F. Doyle
School of Aeronautics & Astronautics
Purdue University
West Lafayette, IN, USA
ISSN 0941-5122 ISSN 2192-063X (electronic)

ISBN 978-3-030-59678-1 ISBN 978-3-030-59679-8 (eBook)
https://doi.org/10.1007/978-3-030-59679-8
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland
AG 2021
2nd edition: © Springer 1997
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
This book is dedicated to
my father and mother,
Patrick and Teresa Doyle.
Thanks for the dreams.
Had I the heavens’ embroidered cloths,

Enwrought with golden and silver light,
The blue and the dim and the dark cloths
Of night and light and the half light,
Would spread the cloths under your feet:
But I, being poor, have only my dreams;
I have spread my dreams under your feet;
Tread softly because you tread upon my dreams.
—W.B. Yeats [1]
Preface
The study of wave propagation seems very remote to many engineers, even to those
who are involved in structural dynamics. One of the reasons for this is that the
examples usually taught in school are either so simple as to be not applicable
to real world problems or so mathematically abstruse as to be intractable. This
book contains an approach, spectral analysis or frequency domain synthesis, that
I have found to be very effective in analyzing waves. What has struck me most
about this approach is how I can use the same analytical framework to examine
the experimental results as well as to manipulate the experimental data itself.
As an experimentalist, I had found it very frustrating having my analytical tools
incompatible with my experiments. For example, it is experimentally impossible to
generate a step function wave and yet that is the type of analytical solution often
available.
Spectral analysis is very encompassing—it touches on analysis, numerical
methods, and experimental methods. I want this book to do justice to its versatility,
so many subjects are introduced. As a result, some areas may seem a little thin, but I
do hope, nonetheless, that the bigger picture, the unity, comes across. Furthermore,
spectral analysis is not so much a solution technique as it is a different insight into
the wave mechanics; consequently, in most of the examples an attempt is made to
make the connection between the frequency domain and time domains.
In writing the second edition, I strived to keep what was good about the
first edition—that is, the combination of experimental and analytical results—but
incorporated more recent developments and extensions at that time. The question
not fully articulated in the first edition is: What should be different about a book
on waves in structures ? This is the question that guided my reorganization of the
material as well as the selection of new topics. It had become clearer to me that
the essence of a structure is the coupling of systems and, this should be the central
theme of a book on waves in structures. There are two readily recognized forms
of coupling: mechanical coupling “at the ends” such as when two bars are joined
at an angle, and differential coupling as when two bars are connected uniformly
along their lengths by springs. Both couplings are intimately related to each other as
seen from the example of a curved beam: it can be modeled as a collection of small
vii
viii Preface
straight segments connected end to end, or directly in terms of coupled differential

equations. The two approaches, ultimately, give the same results but, at the same
time, give quite different insights into the system behavior. The former leads to
richer system response functions and its ultimate form is in the spectral element
method. The latter leads to richer differential equations, which is manifested in
very interesting spectrum relations. The variety of examples were chosen so as to
illustrate and elaborate on this dual aspect of coupling.
In writing this third edition, again I strived to keep intact what is good about the
second edition but add recent developments. The two most important developments
over the last two decades is the introduction of engineered materials in the form of
metamaterials and nanostructures. They add a new level to the meaning of structure.
The other development is the almost universal adoption of finite element (FE)
methods as the “go to” tool for solving structural dynamics problems. I see spectral
analysis as a tool to give insight into data be it obtained from physical experiment
or computer experiment. In this edition, some tools are developed to extend the
postprocessing of FE data using spectral analysis ideas.
The organization of the chapters is mostly similar to that of the second edition,
but each chapter is revised and updated with some new examples and references to
the literature. One completely new chapter has been added. Chapter 9 deals with
discrete and discretized structures. Spectral analysis methods are a very natural way
to analyze the behavior of these systems. Examples are taken from nanotechnology
and molecular dynamics.
In recent years, desktop computers have become incredibly powerful and very
affordable. Thus, problems that would not be tackled in the past can be accom-
plished very quickly, and solution schemes rejected in the past are now feasible. For
example, in the first edition, the double summation was barely introduced, but now
it plays a central role in the dynamics of plate and shell structures. Similarly, the
eigenvibration analysis of large structures was limited to the lowest modes but now
can be applied to the very high-frequency modes of relevance to wave propagation
in discrete structures. Therefore, not only is the use of a computer implicit in all
the examples, the solution strategies and techniques are also computer oriented. In
a similar vein, I have tried to supplement each chapter with a collection of pertinent
problems plus specific references that can form the basis for further study.
A book like this is impossible to complete without the help of many people,
but it is equally impossible to properly acknowledge all of them individually.
However, I would like to single out helpers on the first and second editions:
Brian Bilodeau, Albert Danial, Sudhir Kamle, Lance Kannal, Matt Ledington,
Mike Martin, Steve Rizzi, Gopal Srinivasan, and Hong Zhang. Thank you guys.
The errors and inaccuracies in all editions have been purely my own doing.
West Lafayette, IN, USA James F. Doyle

August, 2020
Contents
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1 Spectral Analysis of Wave Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.1 Fourier Transforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.1.1 Continuous Fourier Transforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.1.2 Discrete Fourier Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.1.3 Fast Fourier Transform Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.1.4 Space Distributions Using Fourier Series . . . . . . . . . . . . . . . . . . . . 20
1.2 Applications Using the FFT Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.2.1 Explorations of the Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.2.2 Experimental Aspects of Wave Signals. . . . . . . . . . . . . . . . . . . . . . . 26
1.3 Spectral Analysis of Wave Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
1.3.1 General Functions of Space-Time and Spectrum Relations . 31
1.3.2 Some Wave Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
1.3.3 Group Speed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
1.3.4 Summary of Wave Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
1.4 Propagating and Reconstructing Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
1.4.1 Basic Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
1.4.2 Integration of Signals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
Further Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2 Longitudinal Waves in Rods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.1 Elementary Rod Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.1.1 Equation of Motion and Spectral Analysis . . . . . . . . . . . . . . . . . . . 50
2.1.2 Basic Solution for Waves in Rods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
2.2 Dissipation in Rods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
2.2.1 Distributed Constraint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
2.2.2 Viscoelastic Rod . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2.3 Coupled Thermoelastic Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
2.3.1 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
ix
x Contents
2.3.2 The Spectrum Relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

2.3.3 Blast Loading of a Rod . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
2.4 Reflection and Transmission of Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
2.4.1 Reflection from an Elastic Boundary . . . . . . . . . . . . . . . . . . . . . . . . . 69
2.4.2 Reflection from an Oscillator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
2.4.3 Concentrated Mass Connecting Two Rods . . . . . . . . . . . . . . . . . . . 72
2.4.4 Interactions at a Distributed Elastic Joint . . . . . . . . . . . . . . . . . . . . 74
2.5 Distributed Loading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
2.5.1 Periodically Extended Load Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
2.5.2 Connected Waveguide Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
3 Flexural Waves in Beams. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
3.1 Bernoulli–Euler Beam Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
3.1.1 Equations of Motion and Spectral Analysis . . . . . . . . . . . . . . . . . . 86
3.1.2 Basic Solution for Waves in Beams . . . . . . . . . . . . . . . . . . . . . . . . . . 89
3.1.3 Beam with Axial Load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
3.2 Bernoulli–Euler Beam with Constraints. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.2.1 Beam on an Elastic Foundation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
3.2.2 Coupled Beam Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
3.3 Reflection and Transmission of Flexural Waves . . . . . . . . . . . . . . . . . . . . . . 104
3.3.1 Reflections from Simple Ends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
3.3.2 Reflections and Transmissions at a General Joint . . . . . . . . . . . . 107
3.4 Curved Beams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
3.4.1 Deformation of Curved Beams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
3.4.2 Spectrum Relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
3.4.3 Impact of a Curved Beam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
4 Higher Order Waveguide Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
4.1 Waves in Infinite and Semi-Infinite Media . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
4.1.1 Navier’s Equations and Helmholtz Potentials . . . . . . . . . . . . . . . . 124
4.1.2 Constructing Potentials Appropriate for Boundaries . . . . . . . . 127
4.1.3 Forced Response of a Semi-Infinite Plane. . . . . . . . . . . . . . . . . . . . 132
4.1.4 Free-Edge Waves: Rayleigh Surface Waves . . . . . . . . . . . . . . . . . . 135
4.2 Waves in Doubly Bounded Media . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
4.2.1 Forced Responses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
4.2.2 Spectrum Relations for Free-Wave Responses . . . . . . . . . . . . . . . 141
4.2.3 Discussion of Lamb Waves in Structural Waveguides . . . . . . . 145
4.3 Variational Formulation of Dynamic Equilibrium . . . . . . . . . . . . . . . . . . . . 146
4.3.1 Work and Strain Energy in a General Body . . . . . . . . . . . . . . . . . . 146
4.3.2 Virtual Work and Hamilton’s Principle . . . . . . . . . . . . . . . . . . . . . . . 148
4.3.3 Illustrative Application of the Ritz Semi-direct Method. . . . . 152
Contents xi
4.4 Refined Beam Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155

4.4.1 Timoshenko Beam Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
4.4.2 Adjustable Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
4.4.3 Spectrum Relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
4.4.4 Impact of a Timoshenko Beam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
4.5 Refined Rod Models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
4.5.1 Love One-Mode Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
4.5.2 Mindlin–Herrmann Two-Mode Model . . . . . . . . . . . . . . . . . . . . . . . 167
4.5.3 Three-Mode Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
4.6 N-Mode Rod Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
4.6.1 Kinematic Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
4.6.2 Spectrum and Dispersion Relations . . . . . . . . . . . . . . . . . . . . . . . . . . 177
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
5 The Spectral Element Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
5.1 Structures as Connected Waveguides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
5.2 Spectral Element for Rods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
5.2.1 Shape Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
5.2.2 Dynamic Stiffness for Rods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
5.2.3 Simple Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
5.2.4 Nodal Representation of Distributed Loadings . . . . . . . . . . . . . . 194
5.3 Spectral Element for Beams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
5.3.1 Shape Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
5.3.2 Dynamic Stiffness for Beams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
5.4 General Frame Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
5.4.1 Member Stiffness Matrix Referred to Global Axes . . . . . . . . . . 202
5.4.2 Structural Stiffness Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
5.4.3 Some Programming Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . 205
5.4.4 An Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
5.4.5 Periodic Structures and the Transfer Matrix . . . . . . . . . . . . . . . . . 208
5.4.6 Long Truss Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
5.5 Spectral Super-Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
5.5.1 Formulation of a Spectral Super-Element . . . . . . . . . . . . . . . . . . . . 214
5.5.2 Assemblage of Super-Element Stiffnesses . . . . . . . . . . . . . . . . . . . 217
5.5.3 Example of an Angle Joint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
5.5.4 Recovery of Internal Displacements and Forces . . . . . . . . . . . . . 222
5.5.5 Periodic Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
5.6 Impact Force Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
5.6.1 Wave Propagation Responses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
5.6.2 Frequency Domain Deconvolution . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
5.6.3 Examples of Force Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
xii Contents
6 Waves in Plates and Cylinders. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237

6.1 Models of Plates in Flexure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
6.1.1 Mindlin Plate Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
6.1.2 Flexural Behavior of Very Thin Plates . . . . . . . . . . . . . . . . . . . . . . . 241
6.1.3 Spectral Analysis of Thin Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
6.1.4 Constructing Simpler Waveguide Models . . . . . . . . . . . . . . . . . . . . 250
6.2 Arbitrarily Crested Waves. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
6.2.1 Point Impact of a Plate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
6.2.2 Point Impact Using Wavenumber Summations (pEL
Method) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
6.3 Reflection and Scattering of Flexural Waves . . . . . . . . . . . . . . . . . . . . . . . . . . 263
6.3.1 Waves Reflected from a Straight Edge . . . . . . . . . . . . . . . . . . . . . . . 263
6.3.2 Free-Edge Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
6.3.3 Scattering of Flexural Waves. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
6.4 Waves in Cylinders and Curved Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
6.4.1 Deformation of Cylindrical Shells . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
6.4.2 Wave Propagation Along a Cylinder . . . . . . . . . . . . . . . . . . . . . . . . . 280
6.4.3 Curved Plate Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
6.4.4 Spectrum Relation for Propagation in the Hoop Direction . . 286
6.4.5 Donnell Shell Equations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
7 Thin-Walled Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
7.1 Spectral Elements for Flat Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
7.1.1 Membrane Spectral Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296
7.1.2 Flexure Spectral Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
7.1.3 A Simple Example. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
7.2 Folded Plate Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
7.2.1 Structural Stiffness Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
7.2.2 Computer Program Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
7.2.3 Structural Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
7.3 Spectral Elements for Curved Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
7.3.1 Impact of an Infinite Curved Plate . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
7.3.2 General Shape Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318
7.3.3 Dynamic Stiffness Relation for a Curved Shell Element . . . . 321
7.3.4 Point Loading of a Complete Cylinder . . . . . . . . . . . . . . . . . . . . . . . 323
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
8 Structure–Fluid Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
8.1 Plate–Fluid Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
8.1.1 Linearized Acoustic Wave Equations. . . . . . . . . . . . . . . . . . . . . . . . . 328
8.1.2 Incident Plane Wave on a Plate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332
Contents xiii
8.2 Panel Excitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336

8.2.1 Line Loading of an Infinite Plate in a Fluid . . . . . . . . . . . . . . . . . . 336
8.2.2 Double Panel Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 338
8.2.3 Cylindrical Cavity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342
8.2.4 Comparison of Fluid Loadings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345
8.3 Waveguide Modeling of Distributed Pressures. . . . . . . . . . . . . . . . . . . . . . . . 347
8.3.1 Free Wave Response. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
8.3.2 Modified Spectrum Relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
8.4 Radiation from Finite Plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
8.4.1 Finite Plate Response. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
8.4.2 Fluid Response from a Finite Plate . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
8.4.3 Far-Field Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
9 Discrete and Discretized Structures. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
9.1 Wave Propagation in 1D Discrete Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
9.1.1 Beaded String . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
9.1.2 Two-Mass String. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
9.1.3 Chains with Internal Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
9.2 Wave Propagation in Anisotropic Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 374
9.2.1 Elastic Anisotropic Solids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375
9.2.2 2D Anisotropic Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 379
9.3 Atomic Lattice Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 382
9.3.1 Lennard-Jones Potentials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
9.3.2 Atomistic Models of Elasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
9.3.3 Embedded Atom Models (EAM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391
9.3.4 Continuum Models for Discrete Systems . . . . . . . . . . . . . . . . . . . . 395
9.4 Spectral Analyses of FE Discretized Systems . . . . . . . . . . . . . . . . . . . . . . . . 400
9.4.1 Spectrum Relations from Element Stiffness . . . . . . . . . . . . . . . . . 401
9.4.2 Real-Only Spectrum Relations and Spectral Shapes. . . . . . . . . 404
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
Afterword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 411
Appendix A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
A.1 Bessel Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
A.1.1 Bessel Equations and Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413
A.1.2 Limiting Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 417
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419
Notation
Roman letters
a Radius
A Cross-sectional area
Â, B̂, Ĉ, D̂ Frequency dependent coefficients
b Thickness, depth
B Bulk modulus
cg Group speed
√
co Longitudinal wave speed, EA/ρA
cP , cS , cR Primary, Secondary, and Rayleigh wave speeds
CE Specific heat
D, D̄ Plate stiffness, D = Eh3 /12(1 − ν 2 ), D̄ = GH 3 /12
E, E ∗ , Ê Young’s modulus, E ∗ = E/(1 − ν 2 ), viscoelastic modulus
EI Beam flexural stiffness
F Member axial force
ĝi (x) Element shape functions
G(t), Ĝ Structural unit response, frequency response function
h Beam or rod height, plate thickness
Hn Hankel function, Hn = Jn ± iYn
√
i Complex −1
I Second moment of area, I = bh3 /12 for rectangle
In Modified Bessel functions of the first kind
Jn Bessel functions of the first kind
k, k1 , k2 Waveguide spectrum relations
kx , ky , kz 2-D wavenumbers
K Stiffness, thermal conduction
[ k ], [ K ] Stiffness matrix
Kn Modified Bessel functions of the second kind
xv
xvi Notation
L Length of element, distance to boundary

M, Mx Moment
n Frequency counter
N Number of terms in transform
p(t), p̂ Acoustic pressure
P (t), P̂ Applied force history
qu , qv , qw Distributed load
q Heat flux
r Radial coordinate
R Radius
t Time
T Time window, temperature
u(t) Response; velocity, strain, etc.
u, v, w Displacements
V Member shear force
W Space transform window
x, y, z Rectilinear coordinates
Yn Bessel functions of the second kind, shape functions
Greek letters
α Coefficient of thermal expansion
β [ω2 ρA/EI ]1/4 , [ω2 ρh/D]1/4
ij k Permutation symbol
δij Kronecker delta, small quantity
Determinant
η Viscosity, damping
κ Plate curvature
θ Angular coordinate
ν Poisson’s ratio
μ Shear modulus
λ Lamê constant
ρ Mass density
σ, Stress, strain
ξ Space transform variable
φ, φx , φy Rotation
, Hz Helmholtz functions
ψ Lateral contraction, stress function
ωn Angular frequency
ωc Cut-off, coincidence, critical frequency
Notation xvii
Special Symbols
Rn Random noise
∂2 ∂2
∇2 Differential operator, ∂x 2
+ ∂y 2
V Volume
[ ] Square matrix
{ } Vector
Subscripts
a Acoustic medium
m Space wavenumber counter
n Frequency counter
1, 2 Sensors, modes
P , S, R Primary, Secondary, and Rayleigh waves
, (Comma) derivative with respect to indicated variable
Superscripts
∗ Complex conjugate
¯ Bar, local coordinates
˙ Dot, time derivative
ˆ Frequency domain (transformed) quantity
˜ Wavenumber domain (transformed) quantity
Prime, derivative with respect to argument
Abbreviations
BC, pBC Boundary condition, periodic BC

DoF, SDoF Degree of freedom, single DoF
CFT Continuous Fourier transform
CST Constant strain triangle element
DKT Discrete Kirchhoff triangular FE element
EAM Embedded atom model
EoM Equation of motion
EVP Eigenvalue problem
FB Free body
FE Finite element
FRF Frequency response function
FST Fourier series transform
MD Molecular dynamics
MRT Membrane with rotation triangular FE element
pEL Periodically extended load
TMM, NMM Three-mode model, N-mode model
xviii Notation
Reference
1. Yeats, W.B.: Collected Poems of W.B. Yeats. Macmillan and Co., London (1960)
Introduction
We must gather and group appearances, until the scientific imagination discerns their hidden
laws, and unity arises from variety; and then from unity we must re-deduce variety, and force
the discovered law to utter its revelations of the future.
W.R. Hamilton [4]
This book is an introduction to the spectral analysis method as a means of solving

wave propagation problems in structures. The emphasis is on practical methods
from both the computational and applications aspects, and reference to physical
and computer FE experimental results is made whenever possible.
While it is possible to solve structural dynamics problems by starting with the
partial differential equations of motion and integrating, the task is horrendously
large even for the biggest computers available. This would not be a good idea
anyway because (and this is a point very often overlooked) a useful solution to a
problem is one that also puts organization and coherence onto the results. It is not
sufficient to be able to quote, say, the strain history at some location or even at
thousands of locations; the results must be placed in some higher-order context, be
seen as part of some larger unity. Notice that this aspect of the problem is present
even when interpreting experimental results and is not just associated with analysis.
One of the goals of this book is to provide such a unifying framework for the
analysis of waves in structures. By consistently using the spectral analysis method
for all problems a unity emerges. This unity is not only in the formulation but
(when coupled with the fast Fourier transform FFT) is also among the formulation,
the solution procedure, the solutions themselves, and the post-manipulations of the
results.
© The Editor(s) (if applicable) and The Author(s), under exclusive license to 1
Springer Nature Switzerland AG 2021
J. F. Doyle, Wave Propagation in Structures, Mechanical Engineering Series,
https://doi.org/10.1007/978-3-030-59679-8_1
2 Introduction
What Is Spectral Analysis?
Over the years many analytical techniques have been developed for treating wave
propagation problems. Central among these is the method of Fourier synthesis (or
spectral analysis), where the behavior of a signal is viewed as a superposition of
many infinitely long wave trains of different periods (or frequencies). The actual
response is synthesized by a judicious combination of these wave trains. Thus the
problem of characterizing a signal is transformed into one of determining the set of
combination coefficients. These coefficients are called the Fourier transform of the
signal. While the problem being tackled invariably simplifies when it is expressed
in terms of the Fourier transform (Sneddon’s book [6], for example, shows a wide
range of applications to both static and dynamic problems) the last step in the
analysis involves performing an inverse transform (reconstructing the signal) and
this, generally, is very difficult to do in an exact analytical manner. Consequently,
many approximate and asymptotic schemes have usually been resorted to. These are
quite adequate for determining the remote behavior (as is required in seismology,
say) but can lose too much information when applied to structural impact problems.
It should also be pointed out that analytical transforms are feasible only if the
function to be transformed is mathematically simple—unfortunately this is not the
case in any situation of practical interest and is certainly not true when dealing
with experimental data. This inversion problem is the biggest impediment to a more
widespread use of the transform methods.
Spectral analysis forms the basis of this book, but the approach is different
from the classical method in that, from the outset, the transforms are approximated
by the discrete Fourier transform (DFT). In contrast to the continuous transform,
this represents the signal by a finite number of wave trains and has the enormous
advantage in that the fast Fourier transform (FFT) computer algorithm can be
used for economically computing the transforms. Being able to do transforms
and inversions quickly adds great heuristic value to the tool in which the waves
can be actually “seen” and iterated on, and realistic signals (even ones that are
experimentally based) can be treated. It should be pointed out that while the method
uses a computer, it is not a numerical method in the usual sense, because the
analytical description of the waves is still retained. As a consequence, the very
important class of problems called inverse problems can be tackled.
The approach presented in this book takes advantage of many of the techniques
already developed for use in time series analysis and for the efficient numerical
implementation of them. (Chatfield’s book [1], for example, is a readable introduc-
tion to the area of signal processing.) In fact, this aspect of spectral analysis is really
part of the more general area of digital processing of the signals and herein lies
one of the unifying advantages of the present approach—the programming structure
is then already in place for the subsequent post-processing of the data. This is
especially significant for the manipulation of experimental data.
Introduction 3
Structures and Waveguides
A structure can be as simple as a cantilevered diving board or as complicated

as an airplane fuselage constructed as a combination of thin plates and frame
members. We view such structures as a collection of waveguides with appropriate
connectivities. A waveguide directs the flow of wave energy and, in its elementary
form, can be viewed as a hydraulic network analog, however, the entities transported
are more complicated than water or oil.
Perceiving a structural member as a waveguide is not always an easy matter.
For example, it is reasonable to expect that a narrow bar struck along its length
conducts longitudinal waves and intuition says that if it is struck transversely it
generates flexural motion. However, on closer examination it turns out not to be that
simple. When the bar is first impacted transversely, the waves generated propagate
into a semi-infinite body and behave as if there is only one free surface. Only after
some time has elapsed do the waves experience the other lateral surface, where they
then reflect back into the body. On a time scale comparable to many transits of the
wave, it is seen that a particle at some location further down the guide experiences
a complex superposition of the initial wave plus all the new waves generated by
reflections. This obviously is not flexural motion. The question then is: At what
stage (both time and position) does the response resemble a flexural wave?
The answer comes in two parts. First, it can be said that the transition depends on
such factors as the duration of the pulse, the distance between bounding surfaces,
and the transit time. That is, the longer the pulse and the smaller the depth, the
sooner (both in time and position) the response resembles a flexural wave. However,
it never does become a flexural wave. This leads to the other part of the answer. The
point and success of waveguide analysis is to forego a detailed analysis of the waves
and replace the three-dimensional model by a simpler one that has embedded in it
the essential characteristics of the behavior as well as a reasonable approximation of
the lateral boundary conditions. This model usually involves resultants on the cross
section and is valid (within itself) for all time and positions not just at large times
and distances.
There are various schemes for establishing the waveguide model that ranges from
the purely ad hoc “Strength of Materials” approaches, to reduced forms of the 3D
equations, to using exact solutions; Redwood [5] gives a good survey of waveguide
analysis for both solids and fluids. The approach taken in this book is to begin the
analysis using an elementary model and then to add complexity to it—the formal
procedure is via Hamilton’s variational principle combined with the Ritz method.
This approach has the advantage of being quite intuitive because it is based on a
statement of the deformation; in addition, it can show the way to approach as yet
unformulated problems.
Some exact solutions for waveguides are also developed. Generally, these are too
cumbersome to be of direct use in structural dynamics, but they do aid considerably
in gaining a deeper understanding of the nature of approximation used in the more
familiar waveguide models.
4 Introduction
Wave Propagation and Vibrations
A major aspect of this book is the persistent treatment of the effect of boundaries and
discontinuities on the waves because real structures have many such terminations.
This can be done efficiently because the quantities used in the analysis of the
waveguide are also used to set up the connectivity conditions. As a result, wave
solutions for structures more interesting than simple rods and beams can be pieced
together successfully. Moreover, the way to solve problems of structures with many
members and boundaries is then available.
A connection not often investigated when studying structural dynamics is the
relationship between wave propagation and vibrations. For many engineers, these
are two separate areas with quite distinct methods of analysis. However, another
advantage of the spectral approach to dynamics is that the close connection between
waves and vibrations becomes apparent. Even the same language can be used,
terms such as power spectral density, filtering, spectral estimation, convolution,
and sampled waveforms have the same meaning. Consequently, many of the
technologies developed over the last four decades for vibrations and modal analyses
are directly applicable to the spectral analysis of waves.
An exciting possibility (and one of the motivations for writing this book) is
to facilitate the reverse process, that is, to transfer many of the wave ideas into
vibration analysis. This should lead to a richer understanding of such topics as
impulse testing, transient vibrations, and filtering in periodic structures. A number of
examples throughout the book demonstrate the “evolution” of resonance as multiple
reflections are included in the analysis.
Themes and Threads
This book concentrates on wave propagation in the basic structural elements of rods,
beams, and plates. These form a rich collection of problems and the intent is to
show that they all can be analyzed within the same framework once the spectral
analysis approach is adopted. Because of the structural applications (and because
all structures are finite in extent), a primary theme is the interaction of the wave
with discontinuities such as boundaries, junctions, and attachments. Supplemental
themes involve the construction of the mechanical models, and the duality between
the time and frequency domains.
While the material of each chapter is reasonably self-contained, Graff’s book [3]
can be used as an excellent supplemental reference on elastic wave propagation.
The book by Elmore and Heald [2] gives a broader and simpler introduction to
waves. Chapter 1 recapitulates the essence of the continuous Fourier transform and
its approximation in the form of the discrete Fourier series. The factors affecting the
quality of the approximation (or spectral estimate) are elucidated. It also discusses,
in a general way, how spectral analysis can be used to solve differential equations
Introduction 5
and especially those associated with wave motion. Two concepts of significance
emerge from this analysis. One is the idea of the spectrum relation (that unique
relation between the frequency and the wavenumber) and is essentially the transform
equivalent of the space-time differential equation. The other is that of multi-mode
solutions. These are shown to play a fundamental role in the solution of actual
boundary value problems even though all are not necessarily propagating modes.
The following two chapters deal, respectively, with rods and beams, in nearly
the same format. First, the governing differential equations are derived and then,
by spectral analysis, the kernel solutions and spectrum relations are obtained.
Having initiated a wave, how it interacts with structural discontinuities is then
investigated. Each chapter includes an example of coupling both at the differential
and mechanical levels. Chapter 4 addresses the question of the adequacy of the
waveguide models and a foundation for the construction of higher-order waveguides
is provided that readily allows extensions to be made to other specialized problems.
Chapter 5 introduces the spectral element method as a matrix method approach
to structural dynamics that combines aspects of the finite element method with
the spectral analysis method. In a way, this chapter is the culmination of the
connected waveguide approach and clearly makes the study of wave propagation
in complicated frame structures practical.
Chapter 6 expands the analysis to multiple dimensions to cover flexural wave
propagation in plates and cylinders. Chapter 7 begins the development of a matrix
methodology for plated structures including cylindrical shells.
The final two chapters extend the spectral analysis to some specialist areas.
Chapter 8 deals with the problem of plate/fluid interactions; Chap. 9 is new to this
edition and focusses on discrete systems (atoms, for example), discretized systems
(FE models, for example), and periodic structures (grilles and frame, for example).
The examples are drawn from metamaterials, nanotechnology and focus on filtering
actions.
A number of threads run through all the chapters. The spectral methodology is an
ideal companion for experimental analysis, and throughout the chapters summaries
of some of its experimental applications are given. The emphasis is on how spectral
analysis can extend the type of information extracted from the experimental data—
for example, how a structural response can be used to infer the force history causing
it. Another thread is the integration of finite element (FE) methods into the spectral
analyses. They are used to enhance the examples through what are called computer
experiments and vice versa how spectral analysis can be used to enhance and explain
FE results. An important thread is the role played by elastic constraints. This is
first introduced in connection with the simple rod encased in an elastic medium
but reappears with increasing complexity in the subsequent chapters. Its ultimate
expression is found in the coupled modes of the curved plate.
Admittedly, a large range of problems have been left out even though most
of them are treatable by the spectral methods. Consequently, an effort is made
to supplement each chapter with a collection of pertinent problems plus specific
references that indicate extensions of the modeling and the applications.
6 Introduction
A Question of Units
The choice of any particular set of units is bound to find disfavor with some
readers. Nondimensionalizing all of the plots is an unsatisfactory solution especially
because experimental results are included. Furthermore, working with dimensional
quantities help give a better “feel” for the results. The resolution of this dilemma
adopted here is to do most of the examples using a nominal material with nominal
properties. In the two major systems of units, these are given in the following table:
Property Common (US) SI

Material Aluminum Aluminum
Young’s Modulus, E 10 × 106 psi 70 GN/m2
Shear Modulus, G 4 × 106 psi 28 GN/m2
Mass Density, ρ 0.25 × 10−3 lb·s2 /in.4 2800 kg/m3
Bar height, h 1 in. 25 mm
Bar depth, b 1 in. 25 mm
Bar length, L 100 in. 2500 mm
Plate thickness, h 0.1 in. 2.5 mm
The choice of shear modulus results in a Poisson’s ratio of ν = 0.25. One of the
recurring material parameters is the ratio

E
co = = 200 × 103 in./s = 5000 m/s
ρ
When a natural nondimensionalizing factor, such as co or the acceleration g,

presents itself we use it. In some examples, we deviate from these nominal values
and the dimensions are then made explicit. In other examples, where it is the relative
waveform (shape) and not the absolute magnitude that is important, we do not state
the units at all. Where ever possible, when FE results are quoted the complete
specification of the problem (mesh density, BCs, and so on) is given.
References
1. Chatfield, C.: The Analysis of Time Series: An Introduction. Chapman and Hall, London (1984)
2. Elmore, W.C., Heald, M.A.: Physics of Waves. Dover, New York (1985)
3. Graff, K.F.: Wave Motion in Elastic Solids. Ohio State University Press, Columbus (1975)
4. Hamilton, W.R.: The Mathematical Papers of Sir W.R. Hamilton. Cambridge University Press,
Cambridge (1940)
5. Redwood, M.: Mechanical Waveguides. Pergamon Press, New York (1960)
6. Sneddon, I.N.: Fourier Transforms. McGraw-Hill, New York (1951)
Chapter 1
Spectral Analysis of Wave Motion
It has long been known that an arbitrary time signal can be thought of as the
superposition of many sinusoidal components, that is, it has a distribution or
spectrum of components. Working in terms of the spectrum is called spectral
analysis. In wave analysis, the time domain for a motion or response is from minus
infinity to plus infinity. Functions in this domain are represented by a continuous
distribution of components which is known as its continuous Fourier transform
(CFT). However, the numerical evaluation and manipulation of the components
require discretizing the distribution in some manner—the one chosen here is by
way of the discrete Fourier transform (DFT). This has the significant advantage
that it allows the use of the very efficient fast Fourier transform (FFT) computer
algorithm.
The goal of this chapter is to introduce the discrete Fourier transform for the
efficient computation of the spectral content of a signal. Possible sources of errors
in using it on finite samples and ways of reducing their influences are discussed. The
Fourier transform is then applied to wave analysis. The crucial step is to set up the
connection between the spectral responses at different space locations—we do this
through the governing differential equations. In doing so, certain key ideas emerge
which recur throughout this book; central among these ideas are that of a wave
mode, its spectrum relation, and its phase and group speeds. In Fig. 1.1, movement
of the amplitude corresponds to the group speed, movement of the zero crossings
correspond to the phase speed.
1.1 Fourier Transforms
The continuous transform is a convenient starting point for discussing spectral

analysis because of its exact representation of functions. Only its definition and basic
© The Editor(s) (if applicable) and The Author(s), under exclusive license to 7
Springer Nature Switzerland AG 2021
J. F. Doyle, Wave Propagation in Structures, Mechanical Engineering Series,
https://doi.org/10.1007/978-3-030-59679-8_2
8 1 Spectral Analysis of Wave Motion
Fig. 1.1 Response showing the difference between the group and phase speeds. Movement of
the envelope corresponds to group behavior while movement of the zero crossings corresponds to
phase behavior
properties are given here; a more complete account can be found in Refs. [20, 21].
Background information on time series analysis can be found in Ref. [5].
The continuous Fourier transform is a powerful technique but has the drawback
that the functions (signals) must be known analytically over the complete domain.
This occurs in only rare cases making it unsuitable for practical situations, especially
if the signals are experimental in origin. Discrete Fourier transforms are more
suitable for our purposes.
1.1.1 Continuous Fourier Transforms
The continuous Fourier transform pair of a function F (t), defined on the time
domain from −∞ to +∞, is given as
∞ ∞
2π F (t) = Ĉ(ω)e+iωt dω , Ĉ(ω) = F (t)e−iωt dt (1.1)
−∞ −∞
where Ĉ(ω) is the continuous

√ Fourier transform (CFT), ω is the angular frequency,
and i is the complex −1. The first form is the inverse transform while the second
is the forward transform—this arbitrary convention arises because the signal to be
transformed usually originates in the time domain. The factor of 2π is necessary
so that a sequential use of the forward and inverse transforms recovers the original
function. However, it should be pointed out that other forms for this factor can be
found in the literature.
The process of obtaining the Fourier transform of a signal separates the waveform
into its constituent sinusoids (or spectrum) and thus a plot of Ĉ(ω) against frequency
represents a diagram displaying the amplitude of each of the constituent sinusoids.
1.1 Fourier Transforms 9
The spectrum Ĉ(ω) usually has nonzero real and imaginary parts. We use the
overhead “hat” to indicate the frequency domain spectrum of a function.
By way of a simple example of the application of Fourier transforms, consider a
rectangular pulse where the time function is given by
F (t) = Fo − a/2 ≤ t ≤ a/2
and zero otherwise. Substituting into the forward transform and integrating give
+a/2
sin (ωa/2)
Ĉ(ω) = Fo e−iωt dt = Fo a ≡ Ĉo (ω)
−a/2 ωa/2
In this particular case the transform is real-only and symmetric about ω = 0 as

shown in Fig. 1.2. The term inside the braces is called a sinc function, and has the
characteristic behavior of starting at unity magnitude and oscillating with decreasing
amplitude as its argument increases. It is noted that the value of the transform at
ω = 0 is the area under the time function. This, in fact, is a general result as seen
from
∞ ∞
Ĉ(0) = F (t)e−i0t dt = F (t)dt
−∞ −∞
A step function is therefore improper because Ĉ(0) is infinite.

When the pulse is displaced along the time axis such that the function is given by
F (t) = Fo t o ≤ t ≤ to + a
and zero otherwise, the transform is then
real
imag 0 to to+a
to=-a/2
to=-a/5
to=+a/10
-100. -50. 0. 50. 100.
Fig. 1.2 Continuous transform of a rectangular pulse with a = 50 µs


sin(ωa/2) −iω(to +a/2)
Ĉ(ω) = Fo a e = Ĉo (ω)e−iω(to +a/2) (1.2)
ωa/2
which has both real and imaginary parts and is not symmetric with respect to ω = 0.
On closer inspection, however, we see that the magnitudes of the two transforms
are the same; it is just that the latter is given an extra phase change of amount
ω(to + a/2). Figure 1.2 shows the transform for different amounts of shift. We
therefore associate phase changes with shifts of the signal along the time axis.
If a pulse is visualized to be at different positions relative to the time origin,
then the amplitude of the spectra will be the same, but each will have a different
phase change. That is, movement in the time domain causes phase changes in the
frequency domain. Investigating these phase changes is the fundamental application
of spectral analysis to wave propagation and is pursued later in this chapter.
For completeness, we now summarize some of the major properties of Fourier
transforms; more detailed accounts can be found in Ref. [21]. In all cases, the
results can be confirmed by taking the example of the rectangular pulse and working
through the transforms long hand. To aid in the summary, we refer to the transform
pair
F (t) ⇔ Ĉ(ω)
where the symbol ⇔ means “can be transformed into.” The double arrowheads
reinforce the idea that the transform can go in either direction and that its properties
are symmetric.
If the functions FA (t) and FB (t) have the transforms ĈA (ω) and ĈB (ω),
respectively, then the combined function [FA (t) + FB (t)] has the transform
[ĈA (ω) + ĈB (ω)]. That is,
FA (t) + FB (t) ⇔ ĈA (ω) + ĈB (ω) (1.3)
This is an essential property of the transform and it means that if a signal is

composed of the simple sum of two contributions (say an incident wave and a
reflected wave), then the transform is also composed of a simple sum of the separate
transforms. This linearity property is at the heart of superposition.
The function F (at) (where a is a nonzero constant) has the transform pair
1
F (at) ⇔ Ĉ(ω/a) (1.4)
|a|
indicating a reciprocal scaling relationship for the arguments. That is, time domain
compression corresponds to frequency domain expansion (and vice versa). For
example, if the width of a pulse is made narrower in the time domain, then its extent
in the frequency domain is made broader. It should also be noted that the amplitude
decreases because the energy is distributed over a greater range of frequencies.
If the function F (t) is time shifted by to , then it has the transform pair
F (t − to ) ⇔ Ĉ(ω)e−iωto (1.5)
This property was already seen in the last example where the rectangle was
displaced from the origin. Of course, the property refers to any change at any
position. There is a corresponding relation for frequency shifting.
The transform pairs, Eqs. (1.1), are valid for both F (t) and Ĉ(ω) being complex,
but the functions of usual interest in wave analysis are when F (t) is real. To see the
effect of this, rewrite Eq. (1.1) in terms of its real and imaginary parts as

2π FR = [ĈR cos ωt − ĈI sin ωt] dω , ĈR = [FR cos ωt + FI sin ωt]dt

2π FI = [ĈR sin ωt + ĈI cos ωt]dω , ĈI = − [FR sin ωt − FI cos ωt]dt
where the following decomposition was used
cos θ = 12 [eiθ + e−iθ ] , sin θ = −i 12 [eiθ − e−iθ ] (1.6)
From the above it is apparent that when F (t) is real-only, ĈR is even and ĈI is odd.
Mathematically, this is expressed as
ĈR (−ω) = ĈR (ω) , ĈI (−ω) = −ĈI (ω) or Ĉ(−ω) = Ĉ ∗ (ω) (1.7)
which says that the functions are symmetrical and antisymmetrical, respectively,
about the zero frequency point. This can also be expressed as saying that the negative
frequency side of the transform is the complex conjugate of the positive side.
A very interesting property arises in connection with the products of functions.
Consider the transform of two time functions

Ĉ(ω) = FA (t)FB (t)e−iωt dt
Using the inverse transform relation for FA gives

Ĉ(ω) = ĈA (ω̄)e+i ω̄t d ω̄ FB (t)e−iωt dt
This can be further rearranged as

−i(ω−ω̄)t
Ĉ(ω) = ĈA (ω̄) FB (t)e dt d ω̄ = ĈA (ω̄)ĈB (ω − ω̄) d ω̄
and is expressed as the transform pair

FA (t)FB (t) ⇔ ĈA (ω̄)ĈB (ω − ω̄) d ω̄ (1.8)
This particular form is called a convolution. We use in explaining the effects of

sampling and filtering on the computed transforms. For example, a signal truncated
in the time domain can be thought of as the product of the original signal with the
truncating function. The result in the frequency domain is then no longer a simple
representation of either.
There is a similar relation for products in the frequency domain, namely

ĈA (ω)ĈB (ω) ⇔ FA (τ )FB (t − τ ) dτ (1.9)
This shows that a time domain convolution can be performed as frequency domain
multiplication. Even though this involves both a forward and inverse transform it is
computationally faster than performing the convolution directly. All the mechanical
systems considered in the later chapters can be represented in the frequency
domain as products of the input times the system response. In fact, the reason
why the frequency domain is so useful for analyzing these systems is because the
complicated convolution relations become simple algebraic relations.
1.1.2 Discrete Fourier Transform
We discretize the CFT in two steps: first the frequency integrals are discretized
giving the Fourier series transform (FST), and then the time integrals are discretized
giving the Discrete Fourier transform (DFT).
Let the time function F (t) be known only over a period T ; to apply the
continuous Fourier transform to it, we must extend it somehow to the infinity limits.
In the Fourier series representation, the function is assumed extended to plus and
minus infinity as a periodic function with the period T . We can view this either as
separate functions of duration T placed one after the other or as the superposition
of separate functions of infinite duration but with nonzero behavior only over the
period T . We adopt this latter view as shown in Fig. 1.3.
Fig. 1.3 Periodic extended signal as a superposition of infinite signals

We saw from the rectangular pulse example that if Ĉ∞ (ω) is the transform of a
pulse, the transform of the same pulse shifted an amount T is Ĉ∞ (ω)e−iωT . It is
obvious therefore that the transform of the periodic signal can be represented as
Ĉ(ω) = Ĉ∞ (ω)[· · · + e+iω2T + e+iωT + 1 + e−iωT + e−iω2T + · · · ]
This transform shows an infinite peak whenever the frequency is one of the discrete
values ωn = 2π n/T . Under this circumstance, each of the exponential terms is
unity and there is an infinity of them. (This result should not be surprising since the
Ĉ(0) component is the area under the curve, and for the periodic rectangle of Fig. 1.3
we see that it is infinite.) At other frequencies, the exponentials are as likely to be
positive as negative and hence their sum will be relatively small. Therefore our first
conclusion about the transform of a periodic extended signal is that it shows very
sharp spectral peaks. We can go further and say that the transform is zero everywhere
except at the discrete frequency values ωn = 2π n/T where it has an infinite value.
We represent this behavior by use of the delta function, δ(x); this special function
is zero everywhere except at x = 0 where it is infinite. It has the very important
additional property that its integral over the whole domain is unity. Thus
Ĉ(ω) = Ĉ∞ (ω)A[· · · + δ(ω + 2π/T ) + δ(ω) + δ(ω − 2π/T ) + · · · ]

= Ĉ∞ (ω)A δ(ω − 2π n/T ) = Ĉ∞ (ω)A δ(ω − ωn )
n n
where A is a proportionality constant which we determine next. We reiterate that

although the transform Ĉ(ω) is a continuous function of frequency it effectively
evaluates to discrete nonzero values.
The remainder of the transform pair is given by
∞ ∞
+iωt
2π F (t) = Ĉ(ω)e dω = Ĉ∞ (ω)A δ(ω − ωn )e+iωt dω
−∞ −∞ n
Interchanging the summation and the integration, and using the properties of the
delta function, gives

2π F (t) = A Ĉ∞ (ωn )e+iωn t
n
Integrate both sides of this equation over a time period T , and realizing that all terms
on the right-hand side are zero except the first, gives that
T
2π F (t) dt = AĈ∞ (0)T
0
Because Ĉ∞ (0) is the area under the single pulse, we conclude that A = 2π/T .
We now have the representation of a function F (t), extended periodically, in

terms of its transform over a single period. That is,

1
n=+∞ T
F (t) = Ĉ∞ (ωn )e+iωn t , Ĉ∞ (ωn ) = F (t)e−iωn t dt (1.10)
T n=−∞ 0
Except for a normalizing constant, this is the complex Fourier series representation
of a periodic signal.
Consider the transform of the rectangular pulse of the last section. The coeffi-
cients are given by
to +a to +a
e−iωn t sin(ωn a/2) −i(to +a/2)ωn
Ĉn = Fo e−iωn t dt = Fo = Fo a e
to −iωn to ωn a/2
This result is the product of three terms. The first, Fo a, is the size of the pulse
as represented by the area. The third, the exponential, is a phase change due to the
shifting of the pulse relative to the time origin. For example, if the pulse is symmetric
about t = 0, then to = −a/2 and there is no phase change of the transform. The
second term, sinc function, is the core of the transform and its amplitude is shown
plotted in Fig. 1.4 for various periods. Notice that the spacing of the coefficients is
at every 1/T hertz in the frequency domain. For the periods T = 500, 200, 100 µs
this gives spacings of f = 2, 5, 10 kHz, respectively.
Notice that in all the cases of Fig. 1.4, the Fourier series gives the exact
values of the continuous transform, but it does so only at discrete frequencies. The
discretization of the frequencies is given by
ωn = 2π n/T
Fourier series
Continuous
-100. -50. 0. 50. 100.
Fig. 1.4 Fourier series coefficients for a rectangular pulse of nonzero duration a = 50 µs extended
with different periods T
Fig. 1.5 Discretization scheme for an arbitrary time function
Therefore, for a given pulse, a larger period gives a more dense distribution of
coefficients, approaching a continuous distribution in the limit of an infinite period.
This, of course, is the continuous transform limit. The finite time integral causes the
transform to be discretized in the frequency domain.
The discrete coefficients in the Fourier series are obtained by performing
continuous integrations over the time period. These integrations are now replaced
by summations as a further step in the numerical implementation of the continuous
transform.
In reference to Fig. 1.5, let the function F (t) be divided into M, piece-wise
constant, segments whose heights are Fm and base T = T /M. The coefficients
are now obtained from

M−1 tm + T /2
Ĉn ≈ D̂n = Fm e−iωn t dt
m=0 tm − T /2

sin ωn T /2
= T Fm e−iωn tm
ωn T /2 m
We see that this is the sum of the transforms of a series of rectangles each shifted in
time by t = tm + T /2. The contribution of each of these is now examined more
closely.
First look at the summation term. If n > M, that is, if n = M + n∗ , then the
exponential term becomes
∗ω t ∗ω t ∗ω t
e−iωn tm = e−inωo tm = e−iMωo tm e−in o m = e−i2π m e−in o m = e−in o m
Hence the summation contribution simply becomes

M−1
∗ω t
Fm e−in o m
m=0
showing that it evaluates the same as when n = n∗ . More specifically, if M = 8

say, then n = −5, 9, 11, 17 evaluates the same as n = 3, 1, 3, 1, respectively. The
discretization process has forced a periodicity into the frequency description.
Now look at the other contribution; we see that the sinc function term does
depend on the value of n and is given by
sin(x) ωn T n T n
sinc(x) ≡ , x= =π =π
x 2 T M M
The sinc function is such that it decreases rapidly with increasing argument and is
small beyond its first zero. The first zero occurs where x = π or n = M; if M
is made very large, that is, the integration segments are made very small, then it is
the higher order coefficients (i.e., large n) that are in the vicinity of the first zero.
Let it be further assumed that the magnitude of these higher order coefficients are
negligibly small. Then an approximation for the coefficients is

M
D̂n ≈ T {1} Fm e−iωn tm
m
on the assumption that it is good for n < M and that Ĉn ≈ 0 for n ≥ M.
Because there is no point in evaluating the coefficients for n > M − 1, the
approximation for the Fourier series coefficients is now taken as
N −1 N −1
1 +iωn tm 1
Fm = F (tm ) ≈ D̂n e = D̂n e+i2π nm/N
T T
n=0 n=0

N −1
N −1
D̂n = D̂(ωn ) ≈ T Fm e−iωn tm = T Fm e−i2π nm/N (1.11)
m=0 m=0
where both m and n range from 0 to N − 1. These are the definition of what
is called the discrete Fourier transform (DFT). It is interesting to note that the
exponentials do not contain dimensional quantities; only the integers n, m, N
appear. In this transform, both the time and frequency domains are discretized,
and as a consequence, the transform behaves periodically in both domains. The
dimensional scale factors T , 1/T have been retained so that the discrete transform
gives the same numerical values as the continuous transform. There are other
possibilities for these scales found in the literature.
The discrete transform enjoys all the same properties as the continuous trans-
form; the only significant difference is that both the time domain and frequency
domain functions are now periodic. To put this point into perspective consider the
following: A discrete Fourier transform seeks to represent a signal (known over a
finite time T ) by a finite number of frequencies. Thus it is the continuous Fourier
transform of a periodic signal. Alternatively, the continuous Fourier transform itself
can be viewed as a discrete Fourier transform of a signal but with an infinite period.
The lesson is that by choosing a large signal sample length, the effect due to
the periodicity assumption can be minimized and the discrete Fourier transform
approaches the continuous Fourier transform.
To help better illustrate the properties of the discrete transform, we consider an
eight-point sampled signal. This can also serve as the test case for any numerical
implementation of the DFT. Let the real-only function be given by the following
sampled values:
F1 = F2 = 1, F0 = F3 = F4 = F5 = F6 = F7 = 0
with T = 1, N = 8. This has the shape of a rectangular pulse if the points

are connected by straight lines—remember, however, that the data is sampled and
hence no information is actually known between the sampling points. Eight points
are given, thus it is implicit that the function repeats itself beyond that. That is, the
next few values are 0, 1, 1, 0, and so on. The transform becomes

7
D̂n = Fm e−i2π nm/8 = F1 e−iπ n/4 + F2 e−iπ n/2
m=0
The first ten transform points, in explicit form, are
D̂0 = 2.0
D̂1 = 0.707 − 1.707i
D̂2 = −1.0 − 1.0i
D̂3 = −0.707 + 0.293i
D̂4 = 0.0
D̂5 = −0.707 − 0.293i
D̂6 = −1.0 + 1.0i
D̂7 = 0.707 + 1.707i
D̂8 = 2.0
D̂9 = 0.707 − 1.707i
The obvious features of the transform are that it is complex and that it begins to
repeat itself beyond n = 7. Note also that D̂4 [the ( 12 N + 1)th value] is the Nyquist
value. The real part of the transform is symmetric about the Nyquist frequency,
while the imaginary part is antisymmetric. It follows from this that the sum 12 [D̂n +
D̂N −n ] gives only the real part, that is,
2.0, 0.707, −1.0, −0.707, 0.0, −0.707, ···
while the difference 12 [D̂n − D̂N −n ] gives the imaginary part
0, −1.707i, −i, +0.293i, 0, −0.293i, ···
These two functions are the even and odd decompositions, respectively, of the
transform. Also note that D̂0 is the area under the function.
1.1.3 Fast Fourier Transform Algorithm
The final step in the numerical implementation of the CFT is the development
of an efficient algorithm for performing the summations of the discrete Fourier
transform on a computer. The fast Fourier transform (FFT) is simply a very efficient
numerical scheme for computing the discrete Fourier transform. This is not a
different transform—the numbers obtained from the FFT are exactly the same in
every respect as those obtained from the DFT. The intention of this section is to just
survey the major features of the FFT algorithm and to point out how the great speed
increase is achieved. More detailed accounts can be found in the Refs. [4, 7] and
FORTRAN code is given in Ref. [19].
Consider the generic forward transform written as
N −1
Sn = Fm e−i2π nm/N , n = 0, 1, . . . , N − 1
m=0
We write this in the expanded form
S0 = {F0 + F1 + F2 + · · · }
S1 = {F0 + F1 e−i2π 1/N + F2 e−i2π 2/N + · · · }
S2 = {F0 + F1 e−i2π 2/N + F2 e−i2π 4/N + · · · }
..
.
Sn = {F0 + F1 e−i2π n/N + F2 e−i2π n2/N + · · · }
and so on. For each sum Sn , there are (N − 1) complex products and (N − 1)
complex sums. Consequently, the total number of computations (in round terms) is
on the order of 2N 2 . The purpose of the FFT is to take advantage of the special form
of the exponential terms to reduce the number of computations to less than N 2 .
The key to understanding the FFT algorithm lies in seeing the repeated forms
of numbers. This is motivated by considering the special case of N being 8. First
consider the matrix of the exponents −i2π( mn N ):
⎡ ⎤
0 0 0 0 ··· 0
⎢0 ··· (N − 1) ⎥
⎢ 1 2 3 ⎥
⎢ ⎥
−i2π ⎢ 0 2 4 6 ··· 2(N − 1) ⎥
⎢ ⎥
N ⎢0⎢ 3 6 9 ··· 3(N − 1) ⎥
⎥
⎢ .. .. .. .. .. .. ⎥
⎣. . . . . . ⎦
0 (N − 1) 2(N − 1) 3(N − 1) · · · (N − 1)(N − 1)
It is apparent that for an arbitrary value of N, 2π is not, in general, multiplied by an

integer number. These exponents, however, can be made quite regular if N is highly
composite. For example, if N is one of the following
N = 2γ = 2, 4, 8, 16, 32, 64, 112, 256, 512, 1024, . . .
then the effective number of different integers in the matrix is decreased. Thus if
N = 8 we get
⎡ ⎤
0 0 0 0 0 0 0 0
⎢0 1 2 3 4 5 6 7 ⎥
⎢0 8+2 8+4 8+6 ⎥
⎢ 2 4 6 0 ⎥
−i2π ⎢
⎢0 3 6 8+1 8+4 8+7 16 + 2 16 + 5 ⎥
⎥
8 ⎢ ⎢0 4 8+0 8+4 16 + 0 16 + 4 24 + 0 24 + 0 ⎥
⎥
⎢0 5 8+2 8+7 16 + 4 24 + 1 24 + 6 32 + 3 ⎥
⎣ ⎦
0 6 8+4 16 + 2 24 + 0 24 + 6 32 + 4 40 + 2
0 7 8+6 16 + 5 24 + 4 32 + 3 40 + 2 48 + 1
which effectively evaluate to

⎡ ⎤
0 0 0 0 0 0 0 0
⎢0 1 2 3 4 5 6 7⎥
⎢0 6⎥
⎢ 2 4 6 0 2 4 ⎥
−i2π ⎢
⎢0 3 6 1 4 7 2 5⎥⎥
8 ⎢ ⎢0 4 0 4 0 4 0 0⎥⎥
⎢0 5 2 7 4 1 6 3⎥
⎣ ⎦
0 6 4 2 0 6 4 2
0 7 6 5 4 3 2 1
This comes about because the exponentials take on the simple forms
e−i2π [0] = e−i2π [1] = e−i2π [2] = e−i2π [3] = · · · = 1
The regularity is enhanced even more if (N/2 = 4) is added to the latter part of the
odd rows, that is, if it is written as
⎡ ⎤
0 0 0 0 (0 0 0 0) + 0
⎢0 1 2 3 (0 1 2 3) + 4 ⎥
⎢0 6) + 0 ⎥
⎢ 2 4 6 (0 2 4 ⎥
−i2π ⎢
⎢0 3 6 1 (0 3 6 1) + 4 ⎥
⎥
8 ⎢ ⎢0 4 0 4 (0 4 0 4) + 0 ⎥
⎥
⎢0 5 2 7 (0 5 2 7) + 4 ⎥
⎣ ⎦
0 6 4 2 (0 6 4 2) + 0
0 7 6 5 (0 7 6 5) + 4
We see that many of the computations used in forming one of the summations
is also used in the others. For example, S0 , S2 , S4 , S6 all use the sum (F0 + F4 ).
Realizing that e−i2π 4/8 = −1, then we also see that all the odd summations contain
common terms such as (F0 −F4 ). This re-use of the same computations is the reason
a great reduction of computational effort is afforded by the FFT. The algorithm sets
up the bookkeeping so that this is done in a systematic way.
The number of computations with and without the FFT algorithm is given by
3
2N log2 N versus 2N 2
When N = 8, this gives a speed factor of only 3.5:1, but when N = 1024, this jumps
to over 100:1. It is this excellent performance at large N that makes the application
of Fourier analysis feasible for practical problems. On a benchmark machine of 1
GFlops, say, a 1024-point transform takes less than one millisecond.
The fast Fourier transform algorithm is so efficient that it has revolutionized the
whole area of spectral analysis. It can be shown quite simply that it enjoys all the
same properties of the continuous transform. Therefore, in the subsequent analyses,
we assume that any time input or response can be represented in the spectral form

F (t) = Ĉn e+iωn t
and the tasks of forward and inverse transforms are accomplished with a computer
program.
1.1.4 Space Distributions Using Fourier Series
In some of the chapters we have occasion to represent space distributions using

Fourier transforms. Because, typically, only a limited number of points are of
interest (think of strain gage or accelerometer locations) it is more efficient to use
direct summations of Fourier series rather than the FFT algorithm. For this reason,
we now summarize our use of the Fourier series transform.
For concreteness, let the function be symmetric in the space window − 12 W <
x < 12 W and represent it using a cosine series. That is,
1 M 1 M
f (x) = a0 + am cos(2π mx/W ) = a0 + am cos(ξm x)
W 1 W 1
Multiply both sides by cos(ξn x) and integrate over the complete window. Because
the cosine functions are orthogonal on the window, and the integrals of the squares
are 12 W , the right-hand side is nonzero only if n = m and we get
+W/2
f (x) cos(ξm ) dx = {a0 , 12 am }
−W/2
We thus represent it by the transform pair

1 M +W/2
f (x) = a0 + am cos(ξm x) , {a0 , 2 am }
1
= f (x) cos(ξm ) dx
W 1 −W/2
If the function is antisymmetric on the window, then

1 M +W/2
f (x) = bm sin(ξm x) , 1
2 bm = f (x) sin(ξm ) dx
W 1 −W/2
If the integrals are computed numerically (through quadratures), then for efficiency,
both need only be computed on the half window and subsequently multiplied
by two. A general function is represented as the superposition of these two
representations.
As an example, consider a rectangular function of width a symmetrically
positioned at x = 0 as shown in Fig. 1.6. The coefficients evaluate to
-0.5 0.0 0.5 -10. -5. 0. 5. 10.
Fig. 1.6 Fourier series representation of space distributions. (a) Rectangular distribution. Effect of
window size and number of terms. (b) Representing a concentrated load on a large space window
W = 800a

2
{a0 , 2 am }
1
= a, sin(ξm a/2) fo
ξm
The figure shows the reconstructions. Doubling the number of terms improves the
representation; the same effect is achieved by halving the window size. In our later
developments, the window size is chosen, thus if the window is too large for a given
number of terms accuracy suffers, if it is too small, we get the equivalent of wrap-
around problems which we interpret as reflections or the effect of image loads.
To elaborate on this last point, in some of our later analyses we have a need
to model a large space subjected to a relatively narrow distributed load (think of
an infinite beam subjected to a point load). Figure 1.6a shows that a rectangular
distribution is represented reasonably well with M = 40 when the window is W =
5a which implies that a window size of 800a requires M = 6400 which is a large
number. Therefore, ways of doing this efficiently and rationally are useful.
A Gaussian distribution has the representation
M 2 −(αmπ/W )2
e−(x/α) = 12 a0 +
2
am cos(ξ x) , am = e
1 W
A function of somewhat similar shape is the sine-squared function of extent a and
a reasonable connection sets a = 3.32α. The heavy and thin lines in Fig. 1.6b
for M = 2048 compare the Gaussian and sine-squared functions, they differ only
at their base where the latter has a definite delimination of its compact support.
Staying with the example of W = 800a, Fig. 1.6b shows the effect of the number
of terms in the summation. Superficially, there is a significant deterioration as the
number of terms is decreased. What is surprising is that in each case shown the area
under the curve is the same. Thus, if these plots represented a traction distribution,
then the resultant force would be the same in each case. Consequently, especially if
relatively remote responses are of interest, the resultant load is the significant factor.
In other words, while the M = 256 case does not represent the actual distribution,
it does represent the resultant effect and therefore could accurately predict remote
behaviors. The example problems in Chaps. 6 and 7 use this extensively and we
refer to the approach as periodically extended loads (pEL).
1.2 Applications Using the FFT Algorithm
The following examples serve to show the basic procedures used in applying the
FFT to transient signals. The FFT is a transform for which no information is gained
or lost relative to the original signal. However, the information is presented in a
different format that often enriches the understanding of the information. In the
examples to follow, we try to present that duality of the given information.
1.2 Applications Using the FFT Algorithm 23
The examples are divided into two groups: one group explores the properties of
the FFT, the other shows applications to experimental data.
1.2.1 Explorations of the Properties
Reference [12] describes a few explorations of the frequency analysis of signals; it

uses the program DiSPtool which is part of the QED package to accomplish the
tasks. The following are abbreviated results.
Consider the rectangular pulse already treated and that has the transform given
by Eq. (1.2). The effect of different sample lengths and number of points used are
shown in Fig. 1.7. Because of the discrete sampling, note that the vertical sides of
the rectangle always have a rise time of T . In the examples, the jump is treated by
using its half value as shown in the time plots.
First, it is noticed that the transform is symmetric about the middle or Nyquist
frequency. This is a consequence of the input signal being real-only—if it were
complex, then the transform would fill the complete range. What this means is that
N real points are transformed into N/2 complex points and no information is gained
or lost. Therefore, the useful frequency range extends only up to the Nyquist, given
by
1
fNyquist =
2 T
FFT
CFT
0. 160. 320. 480. 0. 50. 100. 150.
Fig. 1.7 FFT transform of a rectangular pulse of width a = 50 µs. (a) Time domain signals.
Circles are sampled data. (b) Frequency domain transforms
This range is increased only by decreasing T . Thus, for a fixed number of points,
fine resolution in the time domain (small T ) means course resolution in the
frequency domain. Finer resolution in the frequency domain is achieved only by
increasing the sample length T .
It is also seen from the figure that the match between the FFT amplitude and the
continuous Fourier transform is very good at low frequencies but gets worse at the
higher frequencies. As mentioned before, the discrete and the continuous transforms
match closely only if the highest significant frequency in the signal is less than
the Nyquist. The rectangular pulse can be represented exactly only by using an
infinite number of sinusoids but the contributing amplitudes get smaller as frequency
increases and therefore a finite number of sinusoids can suffice.
For a given sample rate T , the number of samples only determines the density
of transform points. Thus the top two FFTs in Fig. 1.7 are numerically identical at
the common frequencies. Increasing the sample rate for a given number of samples
increases the Nyquist frequency, and therefore the range of comparison between the
discrete and continuous transforms. It is seen that by the fourth plot, the amplitudes
in the vicinity of the Nyquist are negligible. Using the discrete transform puts
an upper limit on the maximum frequency available to characterize the signal. If
the signal is not smooth, the amplitudes of the high-frequency sinusoids used to
describe the signal are high. Any attempt at capping the high-frequency sinusoids
introduces a distortion in the amplitudes of the lower-frequency sinusoids. This is
called aliasing and is discussed in more detail presently.
For future reference, we now summarize some of the inter-relationships among
the various parameters. Consider a signal of duration T sampled as N points. The
discretization rates in the two domains are
T 1
T = , f =
N T
Various forms for the Nyquist frequency are
N 1 N f
fNyquist = = =
2T 2 T 2
The complementarity of information between the time and frequency domains
is illustrated in Fig. 1.8. On comparing the top plots, we notice that the longer
duration pulse has a shorter main frequency range; however, both exhibit side lobes
which extend significantly along the frequency range. The second trace shows a
smoothed version of the second triangle compared to a rectangular pulse of the same
duration; in the frequency domain it is almost identical to the shorter triangle, the
only difference being the reduction in the high-frequency side lobes. Interestingly,
however, the rectangular pulse has a shorter main significant frequency range but the
side lobes are very significant extending beyond what is shown. Thus time domain
smoothing (using moving averages, say) acts as a high band filter in the frequency
0. 100. 200.
. 300. 400. 0. 10. 20. 30. 40. 50.
Fig. 1.8 Comparison of some pulse-type signals. (a) Time domain signals. (b) Frequency domain
transforms
domain. In the wave analyses of the later chapters, we need to control the frequency
range of the inputs and we do this by using signals without sharp edges.
As a rule of thumb, a triangular pulse has a frequency content of about 2/Tp
where Tp is the duration of the pulse. This is confirmed in the figure for the
smoothed triangular pulse.
The remaining two plots are for modulated signals—in this case a sinusoid
modulated (multiplied) by a triangle. The sinusoid function on its own would give
a peak at 20 kHz, whereas the triangular modulations on their own are as shown for
the top traces. The product of the two in the time domain is a convolution in the
frequency domain. The convolution tends to distribute the effect of the pulse but it
is worth noting that while both waveforms have the same time domain amplitudes,
their transforms have different amplitudes indicating different energy levels. A point
of interest here is that if a very narrow-banded signal in the frequency domain is
desired, then it is necessary to extend the time domain signal as shown in the bottom
example.
The modulated waves are called wave packets or wave groups because they
contain a narrow collection (group) of frequency components. It might have been
thought that the triangular pulse is a more natural group because in the time domain
it is highly localized, but what is also important to us is the frequency content of the
wave and we see that the modulated wave is localized in both the space and time
domains. We use this wave packet idea regularly throughout the remaining chapters
to interrogate complex systems.
The signals we deal with often have multiple reflections. A sense of their effect
can be gaged from Fig. 1.9 where the traces are arranged so as to have a different
number of reflections.
In the time domain, we recognize the multiple reflections as localized distur-
bances in time; in fact, in this simple example we can cut the trace so as to isolate
each of the reflections separately. No such separation is possible in the frequency
domain because any small time-localized portion of signal contributes over the
total frequency range. Consequently, we get an interference effect—the adding and
0.0 0.5 1.0 1.5 2.0 0. 5. 10. 15. 20.
Fig. 1.9 Effect of the superposition of multiple reflections. (a) Time domain signals. (b) Fre-
quency domain transforms
subtracting of phases result in amplifying certain components. We can see from the
figure the evolution of spectral peaks as the number of reflections is increased. In
vibration analyses, which would correspond to an infinite number of reflections, the
spectral peaks become very sharp and this situation would then be called resonance.
1.2.2 Experimental Aspects of Wave Signals
It seems appropriate at this stage to summarize some of the experimental aspects

of recording the wave signals in structures. By necessity, it is only cursory, but it is
hoped that it gives a feel for the type and quality of data recorded. It is necessary to
record an infinitely long trace in order to exactly characterize a general signal. This
is obviously impractical, so the problem reduces to one of obtaining best estimates
for the spectra from a finite trace. This section summarizes some of the necessary
procedures used for spectral estimation, more details and citations are given in
Ref. [11]. Some of these procedures are similar to those used for vibration signals.
Some of the characteristics of the testing to record transient signals associated
with impact and stress wave propagation in structures are:
• High-frequency content (1 kHz–1 MHz),
• Only initial portion (∼ 5000 µs) of the signal is analyzed,
• Only a small number of “runs” are performed,
• The number of data recording channels is usually limited (2–8).
There is quite a range of measurement techniques available for studying wave
propagation problems. Of course, some are more suitable for certain situations
than others, but because the emphasis in this book is on structures, then only those
techniques most commonly used for structural analysis are surveyed. A typical setup
is shown in Fig. 1.10.
Force Force
Transducer Transducer
PCB 303A03
accelerometers Power
Supplies
Tek AM502
computer waveform recorder pre-amps
Fig. 1.10 Typical experimental setup for a plate (left) or frame (right)
Most types of electrical resistance strain gages are usually adequate and because
the events are of short duration, temperature compensation is usually not a serious
concern. The largest gage size allowable is dependent on the highest significant
frequency (or shortest wavelength) of the signal, but generally gages equal to or less
than 3 mm (0.125 in.) are good for most situations. A rule of thumb for estimating
the maximum length is

1 E
L= (1.12)
10f ρ
where L is the gage length, f is the highest significant frequency, and E and ρ are
the Young’s modulus and density, respectively, of the structural material. A typical
gage of length 3 mm (0.125 in.) is good to about 160 kHz. Other aspects of the use
of strain gages in dynamic situations are covered in Refs. [2, 8, 18]. Either constant-
current, potentiometer or Wheatstone bridge gage circuits can be used. Nulling of
the Wheatstone bridge is not necessary for dynamic problems because only the ac
portion of the signal is recorded.
A wide variety of small accelerometers are available. There is a problem with
accelerometers, however, in that if the frequency is high enough, ringing is observed
in the signal. This can often occur for frequencies as low as 20 kHz. Generally
speaking, accelerometers do not have the necessary frequency response of strain
gages, but they are movable, thus making them more suitable for larger, complex
structures. Also, note that if the stress is of interest, then integration of the signal is
required.
It is impossible to be categorical about the minimum setup necessary for
waveform recording but the following pieces of equipment definitely seem essential.
Good preamplifiers are essential for dynamic work; they not only provide the gain
for the low voltages from the bridge but also perform signal conditioning. The
former gives the flexibility in choice of gage circuit as well as reducing the burden
on the recorder for providing amplification. A separate preamplifier is needed for
each strain gage circuit and a typical one should have up to 100K gain, dc to 1
MHz frequency response, and selectable band pass filtering. Generally, the analog
filtering (which is always necessary) is performed at or below the Nyquist frequency
associated with the digitizing.
The most significant advance in recent years is the availability of digital
recorders. A typical digital waveform recorder would have 4 channels, 2048 8-
bit words of memory, and a sample rate of 0.2 µs to 0.1 s. It is common for
high-speed recorders to digitize to 8-bit words (although newer ones use 12-bit
words). This means it has a resolution of 1/256 of the full scale. If the signal
can be made to fill at least half full scale, then this resolution of about 1% of
the maximum signal is adequate. This situation is best achieved by having good
preamplifiers. Many of the newer digital recorders combine the features of an
oscilloscope with those of a small stand-alone computer. The DASH-18 [1] is an
example of the newer type of recorders; it is like a digital strip chart recorder and
notebook computer rolled into one. The DASH-18 can record up to 18 channels
and has a very flexible interface for manipulating the recorded data. There are also
relatively inexpensive recorders available that can be installed in desktop computers.
The Omega Instruments DAS-58 data acquisition card is an example; this is a 12-bit
card capable of 1 Mhz sampling rate and storing up to 1M data points in the on-
board memory. A total of 8 multiplexed channels can be sampled giving the fastest
possible sampling rate of 1M/8 = 125 K samples per second.
The measured signal can contain unwanted contributions from at least two
sources. The first is the ever present electrical noise. Analog filters can be used to
remove much of this, although care must be taken not to remove some of the signal
itself. The second may arise from unwanted reflections. This is especially prevalent
for dispersive signals as the recording is usually extended to capture as much as
possible of the initial passage of the wave. There are many ways of smoothing
digital data but the simplest is to use moving averages of various amounts and points
of application. Figure 1.11 shows an example of smoothing a signal that has an
exaggerated amount of noise and reflections present. The following sequence was
used on data values sampled at every 1 μs:
Another random document with
no related content on Scribd:
exposure to the wildest winter night. For two-and-fifty years he
was the subject of a supernatural palpitation, which kept his bed
and chair, and everything moveable about him, in a perpetual
tremble. For that space of time his breast was miraculously
swollen to the thickness of a fist above his heart. On a post-
mortem examination of the holy corpse, it was found that two of
the ribs had been broken, to allow the sacred ardour of his heart
more room to play! The doctors swore solemnly that the
phenomenon could be nothing less than a miracle. A divine hand
had thus literally ‘enlarged the heart’ of the devotee.[320] St. Philip
enjoyed, with many other saints, the privilege of being
miraculously elevated into the air by the fervour of his
heavenward aspirations. The Acta Sanctorum relates how Ida of
Louvain—seized with an overwhelming desire to present her
gifts with the Wise Men to the child Jesus—received, on the eve
of the Three Kings, the distinguished favour of being permitted to
swell to a terrific size, and then gradually to return to her original
dimensions. On another occasion, she was gratified by being
thrown down in the street in an ecstacy, and enlarging so that her
horror-stricken attendant had to embrace her with all her might to
keep her from bursting. The noses of eminent saints have been
endowed with so subtile a sense that they have detected the
stench of concealed sins, and enjoyed, as a literal fragrance, the
well-known odour of sanctity. St. Philip Neri was frequently
obliged to hold his nose and turn away his head when
confessing very wicked people. In walking the streets of some
depraved Italian town, the poor man must have endured all the
pains of Coleridge in Cologne, where, he says,
‘I counted two-and-seventy stenches,

All well-defined, and several stinks!’
Maria of Oignys received what theurgic mysticism calls the gift of

jubilation. For three days and nights upon the point of death, she
sang without remission her ecstatic swan-song, at the top of a
voice whose hoarseness was miraculously healed. She felt as
though the wing of an angel were spread upon her breast,
thrilling her heart with the rapture, and pouring from her lips the
praises, of the heavenly world. With the melodious modulation of
an inspired recitative, she descanted on the mysteries of the
Trinity and the incarnation—improvised profound expositions of
the Scripture—invoked the saints, and interceded for her friends.
[321]
A nun who visited Catharina Ricci in her ecstasy, saw with
amazement her face transformed into the likeness of the
Redeemer’s countenance. St. Hildegard, in the enjoyment and
description of her visions, and in the utterance of her prophecies,
was inspired with a complete theological terminology hitherto
unknown to mortals. A glossary of the divine tongue was long
preserved among her manuscripts at Wiesbaden.[322] It is
recorded in the life of St. Veronica of Binasco, that she received
the miraculous gift of tears in a measure so copious, that the
spot where she knelt appeared as though a jug of water had
been overset there. She was obliged to have an earthen vessel
ready in her cell to receive the supernatural efflux, which filled it
frequently to the weight of several Milan pounds! Ida of Nivelles,
when in an ecstasy one day, had it revealed to her that a dear
friend was at the same moment in the same condition. The friend
also was simultaneously made aware that Ida was immersed in
the same abyss of divine light with herself. Thenceforward they
were as one soul in the Lord, and the Virgin Mary appeared to
make a third in the saintly fellowship. Ida was frequently enabled
to communicate with spiritual personages, without words, after
the manner of angelic natures. On one occasion, when at a
distance from a priest to whom she was much attached, both she
and the holy man were entranced at the same time; and, when
wrapt to heaven, he beheld her in the presence of Christ, at
whose command she communicated to him, by a spiritual kiss, a
portion of the grace with which she herself had been so richly
endowed. To Clara of Montfaucon allusion has already been
made. In the right side of her heart was found, completely
formed, a little figure of Christ upon the cross, about the size of a
thumb. On the left, under what resembled the bloody cloth, lay
the instruments of the passion—the crown of thorns, the nails,
&c. So sharp was the miniature lance, that the Vicar-General
Berengarius, commissioned to assist at the examination by the
Bishop of Spoleto, pricked therewith his reverend finger. This
marvel was surpassed in the eighteenth century by a miracle
more piquant still. Veronica Giuliani caused a drawing to be
made of the many forms and letters which she declared had
been supernaturally modelled within her heart. To the exultation
of the faithful—and the everlasting confusion of all Jews,
Protestants, and Turks—a post-mortem examination disclosed
the accuracy of her description, to the minutest point. There were
the sacred initials in a large and distinct Roman character, the
crown of thorns, two flames, seven swords, the spear, the reed,
&c.—all arranged just as in the diagram she had furnished.[323]
The diocese of Liège was edified, in the twelfth century, by
seeing, in the person of the celebrated Christina Mirabilis, how
completely the upward tendency of protracted devotion might
vanquish the law of gravitation. So strongly was she drawn away
from this gross earth, that the difficulty was to keep her on the
ground. She was continually flying up to the tops of lonely towers
and trees, there to enjoy a rapture with the angels, and a roost
with the birds. In the frequency, the elevation, and the duration of
her ascents into the air, she surpassed even the high-flown
devotion of St. Peter of Alcantara, who was often seen
suspended high above the fig-trees which overshadowed his
hermitage at Badajos—his eyes upturned, his arms outspread—
while the servant sent to summon him to dinner, gazed with open
mouth, and sublunary cabbage cooled below. The limbs of
Christina lost the rigidity, as her body lost the grossness,
common to vulgar humanity. In her ecstasies she was contracted
into the spherical form—her head was drawn inward and
downward towards her breast, and she rolled up like a
hedgehog. When her relatives wished to take and secure her,
they had to employ a man to hunt her like a bird. Having started
his game, he had a long run across country before he brought
her down, in a very unsportsmanlike manner, by a stroke with his
bludgeon which broke her shin. When a few miracles had been
wrought to vindicate her aërostatic mission, she was allowed to
fly about in peace.[324] She has occupied, ever since, the first
place in the ornithology of Roman-catholic saintship. Such are a
few of the specimens which might be collected in multitudes from
Romanist records, showing how that communion has bestowed
its highest favour on the most coarse and materialised
apprehensions of spiritual truth. Extravagant inventions such as
these—monstrous as the adventures of Baron Münchausen,
without their wit—have been invested with the sanction and
defended by the thunder of the Papal chair. Yet this very Church
of Rome incarcerated Molinos and Madame Guyon as
dangerous enthusiasts.
VI.
Madame Guyon had still some lessons to learn. On a visit to

Paris, the glittering equipages of the park, and the gaieties of St.
Cloud, revived the old love of seeing and being seen. During a
tour in the provinces with her husband, flattering visits and
graceful compliments everywhere followed such beauty, such
accomplishments, and such virtue, with a delicate and
intoxicating applause. Vanity—dormant, but not dead—awoke
within her for the last time. She acknowledged, with bitter self-
reproach, the power of the world, the weakness of her own
resolves. In the spiritual desertion which ensued, she recognised
the displeasure of her Lord, and was wretched. She applied to
confessors—they were miserable comforters, all of them. They
praised her while she herself was filled with self-loathing. She
estimated the magnitude of her sins by the greatness of the
favour which had been shown her. The bland worldliness of her
religious advisers could not blind so true a heart, or pacify so
wakeful a conscience. She found relief only in a repentant
renewal of her self-dedication to the Saviour, in renouncing for
ever the last remnant of confidence in any strength of her own.
It was about this period that she had a remarkable conversation
with a beggar, whom she found upon a bridge, as, followed by
her footman, she was walking one day to church. This singular
mendicant refused her offered alms—spoke to her of God and
divine things—and then of her own state, her devotion, her trials,
and her faults. He declared that God required of her not merely
to labour as others did to secure their salvation, that they might
escape the pains of hell, but to aim at such perfection and purity
in this life, as to escape those of purgatory. She asked him who
he was. He replied, that he had formerly been a beggar, but now
was such no more;—mingled with the stream of people, and she
never saw him afterwards.[325]
The beauty of Madame Guyon had cost her tender conscience
many a pang. She had wept and prayed over that secret love of
display which had repeatedly induced her to mingle with the
thoughtless amusements of the world. At four-and-twenty the
virulence of the small-pox released her from that snare. M.
Guyon was laid up with the gout. She was left, when the disorder
seized her, to the tender mercies of her mother-in-law. That
inhuman woman refused to allow any but her own physician to
attend her, yet for him she would not send. The disease,
unchecked, had reached its height, when a medical man,
passing that way, happened to call at the house. Shocked at the
spectacle Madame Guyon presented, he was proceeding at
once to bleed her, expressing, in no measured terms, his
indignation at the barbarity of such neglect. The mother-in-law
would not hear of such a thing. He performed the operation in
spite of her threats and invectives, leaving her almost beside
herself with rage. That lancet saved the life of Madame Guyon,
and disappointed the relative who had hoped to see her die.
When at length she recovered, she refused to avail herself of the
cosmetics generally used to conceal the ravages of the disorder.
Throughout her suffering she had never uttered a murmur, or felt
a fear. She had even concealed the cruelty of her mother-in-law.
She said, that if God had designed her to retain her beauty, He
would not have sent the scourge to remove it. Her friends
expected to find her inconsolable—they heard her speak only of
thankfulness and joy. Her confessor reproached her with spiritual
pride. The affection of her husband was visibly diminished; yet
the heart of Madame Guyon overflowed with joy. It appeared to
her, that the God to whom she longed to be wholly given up had
accepted her surrender, and was removing everything that might
interpose between Himself and her.[326]
VII.
The experience of Madame Guyon, hitherto, had been such as

to teach her the surrender of every earthly source of gratification
or ground of confidence. Yet one more painful stage on the road
to self-annihilation remained to be traversed. She must learn to
give up cheerfully even spiritual pleasures. In the year 1674,
according to the probable calculation of Mr. Upham, she was
made to enter what she terms a state of desolation, which lasted,
with little intermission, for nearly seven years.[327] All was
emptiness, darkness, sorrow. She describes herself as cast
down, like Nebuchadnezzar, from a throne of enjoyment, to live
among the beasts. ‘Alas!’ she exclaimed, ‘is it possible that this
heart, formerly all on fire, should now become like ice?’ The
heavens were as brass, and shut out her prayers; horror and
trembling took the place of tranquillity; hopelessly oppressed with
guilt, she saw herself a victim destined for hell. In vain for her did
the church doors open, the holy bells ring, the deep-voiced
intonations of the priest arise and fall, the chanted psalm ascend
through clouds of azure wandering incense. The power and the
charm of the service had departed. Of what avail was music to a
burning wilderness athirst for rain? Gladly would she have had
recourse to the vow, to the pilgrimage, to the penance, to any
extremity of self-torture. She felt the impotence of such remedies
for such anguish. She had no ear for comfort, no eye for hope,
not even a voice for complaint.
During this period the emotional element of religion in her mind
appears to have suffered an almost entire suspension.
Regarding the loss of certain feelings of delight as the loss of the
divine favour, she naturally sank deeper and deeper in
despondency. A condition by no means uncommon in ordinary
Christian experience assumed, in her case, a morbid character.
Our emotions may be chilled, or kindled, in ever-varying
degrees, from innumerable causes. We must accustom
ourselves to the habitual performance of duty, whether attended
or not with feelings of a pleasurable nature. It is generally found
that those powerful emotions of joy which attend, at first, the new
and exalting consciousness of peace with God, subside after
awhile. As we grow in religious strength and knowledge, a
steady principle supplies their place. We are refreshed, from time
to time, by seasons of heightened joy and confidence, but we
cease to be dependent upon feeling. At the same time, there is
nothing in Scripture to check our desire for retaining as
constantly as possible a sober gladness, for finding duty
delightful, and the ‘joy of the Lord’ our strength. These are the
truths which the one-sided and unqualified expressions of
Madame Guyon at once exaggerate and obscure.
During this dark interval M. Guyon died. His widow undertook the
formidable task of settling his disordered affairs. Her brother
gave her no assistance; her mother-in-law harassed and
hindered to her utmost; yet Madame Guyon succeeded in
arranging a chaos of papers, and bringing a hopeless imbroglio
of business matters into order, with an integrity and a skill which
excited universal admiration. She felt it was her duty; she
believed that Divine assistance was vouchsafed for its discharge.
Of business, she says, she knew as little as of Arabic; but she
knew not what she could accomplish till she tried. Minds far more
visionary than hers have evinced a still greater aptitude for
practical affairs.
The 22nd of July, 1680, is celebrated by Madame Guyon as the
happy era of her deliverance. A letter from La Combe was the
instrument of a restoration as wonderful, in her eyes, as the
bondage. This ecclesiastic had been first introduced by Madame
Guyon into the path of mystical perfection. His name is
associated with her own in the early history of the Quietist
movement. He subsequently became her Director, but was
always more her disciple than her guide. His admiration for her
amounted to a passion. Incessant persecution and long solitary
imprisonment combined, with devotional extravagance, to cloud
with insanity at last an intellect never powerful. This feeble and
affectionate soul perished, the victim of Quietism, and perhaps of
love. It should not be forgotten, that before the inward condition
of Madame Guyon changed thus remarkably for the better, her
outward circumstances had undergone a similar improvement.
She lived now in her own house, with her children about her.
That Sycorax, her mother-in-law, dropped gall no longer into her
daily cup of life. Domestic tormentors, worse than the goblins
which buffeted St. Antony, assailed her peace no more. An outer
sky grown thus serene, an air thus purified, may well have
contributed to chase away the night of the soul, and to give to a
few words of kindly counsel from La Combe the brightness of the
day-star. Our simple-hearted enthusiast was not so absolutely
indifferent as she thought herself to the changes of this transitory
world.
VIII.
Madame Guyon had now triumphantly sustained the last of those

trials, which, like the probation of the ancient mysteries, made
the porch of mystical initiation a passage terrible with pain and
peril. Henceforward, she is the finished Quietist: henceforward,
when she relates her own experience, she describes Quietism.
At times, when the children did not require her care, she would
walk out into a neighbouring wood, and there, under the shade of
the trees, amidst the singing of the birds, she now passed as
many happy hours as she had known months of sorrow. Her own
language will best indicate the thoughts which occupied this
peaceful retirement, and exhibit the principle there deepened
and matured. She says here in her Autobiography—
‘When I had lost all created supports, and even divine ones, I
then found myself happily necessitated to fall into the pure
divine, and to fall into it through all which seemed to remove me
farther from it. In losing all the gifts, with all their supports, I
found the Giver. Oh, poor creatures, who pass along all your
time in feeding on the gifts of God, and think therein to be most
favoured and happy, how I pity you if ye stop here, short of the
true rest, and cease to go forward to God, through resignation of
the same gifts! How many pass all their lives this way, and think
highly of themselves therein! There are others who, being
designed of God to die to themselves, yet pass all their time in a
dying life, and in inward agonies, without ever entering into God
through death and total loss, because they are always willing to
retain something under plausible pretexts, and so never lose self
to the whole extent of the designs of God. Wherefore, they never
enjoy God in his fulness,—a loss that will not perfectly be known
until another life.’[328]
She describes herself as having ceased from all self-originated
action and choice. To her amazement and unspeakable
happiness, it appeared as though all such natural movement
existed no longer,—a higher power had displaced and occupied
its room. ‘I even perceived no more (she continues) the soul
which He had formerly conducted by His rod and His staff,
because now He alone appeared to me, my soul having given up
its place to Him. It seemed to me as if it was wholly and
altogether passed into its God, to make but one and the same
thing with Him; even as a little drop of water cast into the sea
receives the qualities of the sea.’ She speaks of herself as now
practising the virtues no longer as virtues—that is, not by
separate and constrained efforts. It would have required effort
not to practise them.[329]
Somewhat later she expresses herself as follows:—
‘The soul passing out of itself by dying to itself necessarily
passes into its divine object. This is the law of its transition.
When it passes out of self, which is limited, and therefore is not
God, and consequently is evil, it necessarily passes into the
unlimited and universal, which is God, and therefore is the true
good. My own experience seemed to me to be a verification of
this. My spirit, disenthralled from selfishness, became united with
and lost in God, its Sovereign, who attracted it more and more to
Himself. And this was so much the case, that I could seem to
see and know God only, and not myself.... It was thus that my
soul was lost in God, who communicated to it His qualities,
having drawn it out of all that it had of its own.... O happy
poverty, happy loss, happy nothing, which gives no less than
God Himself in his own immensity,—no more circumscribed to
the limited manner of the creation, but always drawing it out of
that to plunge it wholly into His divine Essence. Then the soul
knows that all the states of self-pleasing visions, of intellectual
illuminations, of ecstasies and raptures, of whatever value they
might once have been, are now rather obstacles than
advancements; and that they are not of service in the state of
experience which is far above them; because the state which
has props or supports, which is the case with the merely
illuminated and ecstatic state, rests in them in some degree, and
has pain to lose them. But the soul cannot arrive at the state of
which I am now speaking, without the loss of all such supports
and helps.... The soul is then so submissive, and perhaps we
may say so passive,—that is to say, is so disposed equally to
receive from the hand of God either good or evil,—as is truly
astonishing. It receives both the one and the other without any
selfish emotions, letting them flow and be lost as they came.’[330]
These passages convey the substance of the doctrine which,
illustrated and expressed in various ways, pervades all the
writings of Madame Guyon. This is the principle adorned by the
fancy of her Torrents and inculcated in the practical directions of
her Short Method of Prayer. Such is the state to which Quietism
proposes to conduct its votaries. In some places, she qualifies
the strength of her expressions,—she admits that we are not at
all times equally conscious of this absolute union of the soul with
its centre,—the lower nature may not be always insensible to
distress. But the higher, the inmost element of the soul is all the
while profoundly calm, and recollection presently imparts a
similar repose to the inferior nature. When the soul has thus
passed, as she phrases it, out of the Nothing into the All, when
its feet are set in ‘a large room’ (nothing less, according to her
interpretation, than the compass of Infinity), ‘a substantial or
essential word’ is spoken there. It is a continuous word—potent,
ineffable, ever uttered without language. It is the immediate
unchecked operation of resident Deity. What it speaks, it effects.
It is blissful and mysterious as the language of heaven. With
Madame Guyon, the events of Providence are God, and the
decisions of the sanctified judgment respecting them are nothing
less than the immediate voice of God in the soul. She compares
the nature thus at rest in God to a tablet on which the divine
hand writes,—it must be held perfectly still, else the characters
traced there will be distorted or incomplete. In her very humility
she verges on the audacity which arrogates inspiration. If she,
passive and helpless, really acts no more, the impulses she
feels, her words, her actions, must all bear the impress of an
infallible divine sanction. It is easy to see that her speech and
action—always well-meant, but frequently ill-judged,—were her
own after all, though nothing of her own seemed left. She
acknowledges that she was sometimes at a loss as to the course
of duty. She was guided more than once by random passages of
the Bible, and the casual expressions of others, somewhat after
the fashion of the Sortes Virgilianæ and the omens of ancient
Rome. Her knowledge of Scripture, the native power of her
intellect, and the tenderness of her conscience, preserved her
from pushing such a view of the inward light to its worst extreme.
IX.
The admixture of error in the doctrine which Madame Guyon was

henceforward to preach with so much self-denying love, so much
intrepid constancy, appears to us to lie upon the surface. The
passages we have given convey, unquestionably, the idea of a
practical substitution of God for the soul in the case of the
perfectly sanctified. The soul within the soul is Deity. When all is
desolate, silent, the divine Majesty arises, thinks, feels, and acts,
within the transformed humanity. It is quite true that, as
sanctification progresses, Christian virtue becomes more easy
as the new habit gains strength. In many respects it is true, as
Madame Guyon says, that effort would be requisite to neglect or
violate certain duties or commands rather than to perform them.
But this facility results from the constitution of our nature. We
carry on the new economy within with less outcry, less labour,
less confusion and resistance than we did when the revolution
was recent, but we carry it on still—working with divine
assistance. God works in man, but not instead of man. It is one
thing to harmonize, in some measure, the human will with the
divine, another to substitute divine volitions for the human. Every
man has within him Conscience—the judge often bribed or
clamoured down; Will—the marshal; Imagination—the poet;
Understanding—the student; Desire—the merchant, venturing its
store of affection, and gazing out on the future in search of some
home-bound argosy of happiness. But all these powers are
found untrue to their allegiance. The ermine—the baton—the
song—the books—the merchandize, are at the service of a
usurper—Sin. When the Spirit renews the mind, there is no
massacre—no slaughterous sword filling with death the streets
of the soul’s city, and making man the ruin of his former self.
These faculties are restored to loyalty, and reinstated under God.
Then Conscience gives verdict, for the most part, according to
the divine statute-book, and is habitually obeyed. Then the lordly
Will assumes again a lowly yet noble vassalage. Then the dream
of Imagination is a dream no longer, for the reality of heaven
transcends it. Then the Understanding burns the magic books in
the market-place, and breaks the wand of its curious arts—but
studies still, for eternity as well as time. The activity of Desire
amasses still, according to its nature,—for some treasure man
must have. But the treasure is on earth no longer. It is the
advantage of such a religion that the very same laws of our
being guide our spiritual and our natural life. The same self-
controul and watchful diligence which built up the worldly habits
towards the summits of success, may be applied at once to
those habits which ripen us for heaven. The old experience will
serve. But the mystic can find no common point between himself
and other men. He is cut off from them, for he believes he has
another constitution of being, inconceivable by them—not merely
other tastes and a higher aim. The object of Christian love may
be incomprehensible, but the affection itself is not so. It is
dangerous to represent it as a mysterious and almost
unaccountable sentiment, which finds no parallel in our
experience elsewhere. Our faith in Christ, as well as our love to
Christ, are similar to our faith and love as exercised towards our
fellow-creatures. Regeneration imparts no new faculty, it gives
only a new direction to the old.
X.
Quietism opposed to the mercenary religion of the common and

consistent Romanism around it, the doctrine of disinterested
love. Revolting from the coarse machinery of a corrupt system, it
took refuge in an unnatural refinement. The love inculcated in
Scripture is equally remote from the impracticable indifference of
Quietism and the commercial principle of Superstition. Long ago,
at Alexandria, Philo endeavoured to escape from an effete and
carnal Judaism to a similar elevation. The Persian Sufis were
animated with the same ambition in reaction against the frigid
legalism of the creed of Islam. Extreme was opposed to extreme,
in like manner, when Quietism, disgusted with the unblushing
inconsistencies of nominal Christianity, proclaimed its doctrine of
perfection—of complete sanctification by faith. This is not a
principle peculiar to mysticism. It is of little practical importance.
It is difficult to see how it can be applied to individual experience.
The man who has reached such a state of purity must be the last
to know it. If we do not, by some strange confusion of thought,
identify ourselves with God, the nearer we approach Him the
more profoundly must we be conscious of our distance. As, in a
still water, we may see reflected the bird that sings in an
overhanging tree, and the bird that soars towards the zenith—the
image deepest as the ascent is highest—so it is with our
approximation to the Infinite Holiness. Madame Guyon admits
that she found it necessary jealously to guard humility, to watch
and pray—that her state was only of ‘comparative immutability.’ It
appears to us that perfection is prescribed as a goal ever to be
approached, but ever practically inaccessible. Whatever degree
of sanctification any one may have attained, it must always be
possible to conceive of a state yet more advanced,—it must
always be a duty diligently to labour towards it.
Quietist as she was, few lives have been more busy than that of
Madame Guyon with the activities of an indefatigable
benevolence. It was only self-originated action which she strove
to annihilate. In her case, especially, Quietism contained a
reformatory principle. Genuflexions and crossings were of little
value in comparison with inward abasement and crucifixion. The
prayers repeated by rote in the oratory, were immeasurably
inferior to that Prayer of Silence she so strongly commends—
that prayer which, unlimited to times and seasons, unhindered by
words, is a state rather than an act, a sentiment rather than a
request,—a continuous sense of submission, which breathes,
moment by moment, from the serene depth of the soul, ‘Thy will
be done.’[331]
As contrasted with the mysticism of St. Theresa, that of Madame
Guyon appears to great advantage. She guards her readers
against attempting to form any image of God. She aspires to an
intellectual elevation—a spiritual intuition, above the sensuous
region of theurgy, of visions, and of dreams. She saw no Jesuits
in heaven bearing white banners among the heavenly throng of
the redeemed. She beheld no Devil, ‘like a little negro,’ sitting on
her breviary. She did not see the Saviour in an ecstasy, drawing
the nail out of His hand. She felt no large white dove fluttering
above her head. But she did not spend her days in founding
convents—a slave to the interests of the clergy. So they made a
saint of Theresa, and a confessor of Madame Guyon.
XI.
In the summer of 1681, Madame Guyon, now thirty-four years of
age, quitted Paris for Gex, a town lying at the foot of the Jura,
about twelve miles from Geneva. It was arranged that she should
take some part in the foundation and management of a new
religious and charitable institution there. A period of five years
was destined to elapse before her return to the capital. During
this interval, she resided successively at Gex, Thonon, Turin,
and Grenoble. Wherever she went, she was indefatigable in
works of charity, and also in the diffusion of her peculiar
doctrines concerning self-abandonment and disinterested love.
Strong in the persuasion of her mission, she could not rest
without endeavouring to influence the minds around her. The
singular charm of her conversation won a speedy ascendency
over nearly all with whom she came in contact. It is easy to see
how a remarkable natural gift in this direction contributed both to
the attempt and the success. But the Quietest had buried nature,
and to nature she would owe nothing,—these conversational
powers could be, in her eyes, only a special gift of utterance from
above. This mistake reminds us of the story of certain monks
upon whose cloister garden the snow never lay, though all the
country round was buried in the rigour of a northern winter. The
marvellous exemption, long attributed by superstition to miracle,
was discovered to arise simply from certain thermal springs
which had their source within the sacred inclosure. It is thus that
the warmth and vivacity of natural temperament has been
commonly regarded by the mystic, as nothing less than a fiery
impartation from the altar of the celestial temple.
At Thonon her apartment was visited by a succession of
applicants from every class, who laid bare their hearts before
her, and sought from her lips spiritual guidance or consolation.
She met them separately and in groups, for conference and for
prayer. At Grenoble, she says she was for some time engaged
from six o’clock in the morning till eight at evening in speaking of
God to all sorts of persons,—‘friars, priests, men of the world,
maids, wives, widows, all came, one after another, to hear what I
had to say.’[332] Her efforts among the members of the House of
the Novitiates in that city, were eminently successful, and she
appears to have been of real service to many who had sought
peace in vain, by the austerities and the routine of monastic
seclusion. Meanwhile, she was active, both at Thonon and
Grenoble, in the establishment of hospitals. She carried on a
large and continually increasing correspondence. In the former
place she wrote her Torrents, in the latter, she published her
Short Method of Prayer, and commenced her Commentaries on
the Bible.[333]
But alas! all this earnest, tireless toil is unauthorized. Bigotry
takes the alarm, and cries the Church is in danger. Priests who
were asleep—priests who were place-hunting—priests who were
pleasure-hunting, awoke from their doze, or drew breath in their
chase, to observe this woman whose life rebuked them—to
observe and to assail her; for rebuke, in their terminology, was
scandal. Persecution hemmed her in on every side; no
annoyance was too petty, no calumny too gross, for priestly
jealousy. The inmates of the religious community she had
enriched were taught to insult her—tricks were devised to
frighten her by horrible appearances and unearthly noises—her
windows were broken—her letters were intercepted. Thus,
before a year had elapsed, she was driven from Gex. Some
called her a sorceress; others, more malignant yet, stigmatized
her as half a Protestant. She had indeed recommended the
reading of the Scriptures to all, and spoken slightingly of mere
bowing and bead-counting. Monstrous contumacy—said, with
one voice, spiritual slaves and spiritual slave-owners—that a
woman desired by her bishop to do one thing, should discover
an inward call to do another. At Thonon the priests burnt in the
public square all the books they could find treating of the inner
life, and went home elated with their performance. One thought
may have embittered their triumph—had it only been living flesh
instead of mere paper! She inhabited a poor cottage that stood
by itself in the fields, at some distance from Thonon. Attached to
it was a little garden, in the management of which she took
pleasure. One night a rabble from the town were incited to terrify
her with their drunken riot,—they trampled down and laid waste
the garden, hurled stones in at the windows, and shouted their
threats, insults, and curses, round the house the whole night.
Then came an episcopal order to quit the diocese. When
compelled subsequently, by the opposition she encountered, to
withdraw secretly from Grenoble, she was advised to take refuge
at Marseilles. She arrived in that city at ten o’clock in the
morning, but that very afternoon all was in uproar against her, so
vigilant and implacable were her enemies.
Note to page 214.
Autobiography, chapp. viii. and x. In describing her state of mind at

this time, she says,—‘This immersion in God immerged all things. I
could no more see the saints, nor even the blessed Virgin, out of
God; but I beheld them all in Him. And though I tenderly loved
certain saints, as St. Peter, St. Paul, St. Mary Magdalen, St.
Theresa, with all those who were spiritual, yet I could not form to
myself images of them, nor invoke any of them out of God.’ Here a
genuine religious fervour, described in the language of mystical
theology, has overcome superstition, and placed her, unconsciously,
in a position similar to that of Molinos with regard to these
professedly subordinate objects of Romanist worship. It may be
observed, in passing, that while Rome pretends to subordinate saint-
worship, she denounces those of her children who really do so, as
heretical, i.e., reformatory, in their tendency.
Madame Guyon was enabled at this period to enjoy a habitual
inward prayer,—‘a prayer of rejoicing and possession, wherein the
taste of God was so great, so pure, unblended, and uninterrupted,
that it drew and absorbed the powers of the soul into a profound
recollection, without act or discourse. For I had now no sight but of
Jesus Christ alone. All else was excluded, in order to love with the
greater extent, without any selfish motives or reasons for loving.’
With much good sense, she declares this continual and immediate
sense of the Divine presence far safer and higher than the sensible
relish of ecstasies and ravishments,—than distinct interior words or
revelations of things to come,—so often imaginary, so apt to divert
our desires from the Giver to the gifts;—this is the revelation of
Jesus Christ, which makes us new creatures, the manifestation of
the Word within us, who cannot deceive,—the life of true and naked
faith, which darkens all self-pleasing lights, and reveals the minutest
faults, that pure love may reign in the centre of the soul. Thus, while
inheriting the phraseology of the mystics (and we discern in these
accounts of her early experience the influence of her later readings
in mystical theology), she is less sensuous than Theresa, less
artificial than John. Like the latter, she assigns to love the office of
annihilating the will, to faith that of absorbing the understanding, ‘so
as to make it decline all reasonings, all particular brightnesses and
illustrations.’ The Annihilation of the Will, or the Union in the Will of
God, consists, with her, simply in a state of complete docility, the
soul yielding itself up to be emptied of all which is its own, till it finds
itself by little and little detached from every self-originated motion,
and placed ‘in a holy indifference for willing;—wishing nothing but
what God does and wills.’—P. 70.
Note to page 218.
She describes herself, when at Thonon, as causing sundry devils to

withdraw with a word. But the said devils, like some other sights and
sounds which terrified her there, were probably the contrivance of
the monks who persecuted her, with whom expertness in such tricks
was doubtless reckoned among the accomplishments of sanctity.
When at the same place (she was then a little past thirty), Madame
Guyon believed that a certain virtue was vouchsafed her—a gift of
spiritual and sometimes of bodily healing, dependent, however, for
its successful operation, on the degree of susceptibility in the
recipients.—Autobiography, part II. c. xii.
There also she underwent some of her most painful and mysterious
experiences with regard to Father La Combe. She says,—‘Our Lord
gave me, with the weaknesses of a child, such a power over souls,
that with a word I put them in pain or in peace, as was necessary for
their good. I saw that God made Himself to be obeyed, in and
through me, like an absolute Sovereign. I neither resisted Him nor
took part in anything.... Our Lord had given us both (herself and La
Combe) to understand that He would unite us by faith and by the
cross. Ours, then, has been a union of the cross in every respect, as
well as by what I have made him suffer, as by what I have suffered
for him.... The sufferings which I have had on his account were such
as to reduce me sometimes to extremity, which continued for several
years. For though I have been much more of my time far from him
than near him, that did not relieve my suffering, which continued till
he was perfectly emptied of himself, and to the very point of
submission which God required of him.... He hath occasioned me
cruel pains when I was near a hundred leagues from him. I felt his
disposition. If he was faithful in letting Self be destroyed, I was in a
state of peace and enlargement. If he was unfaithful in reflection or
hesitation, I suffered till that was passed over. He had no need to
write me an account of his condition, for I knew it; but when he did
write, it proved to be such as I had felt it.’—Ibid. p. 51.
She says that frequently, when Father La Combe came to confess
her, she could not speak a word to him; she felt take place within her
the same silence toward him, which she had experienced in regard
to God. I understood, she adds, that God wished to teach me that
the language of angels might be learnt by men on earth,—that is,
converse without words. She was gradually reduced to this wordless
communication alone, in her interviews with La Combe; and they
imagined that they understood each other, ‘in a manner ineffable and
divine.’ She regarded the use of speech, or of the pen, as a kind of
accommodation on her part to the weakness of souls not sufficiently
advanced for these internal communications.
Here Madame Guyon anticipates the Quakers. Compare Barclay’s
Apology, Prop. xi. §§ 6, 7.
Shortly after her arrival in Paris, she describes herself as favoured,
from the plenitude which filled her soul, with ‘a discharge on her
best-disposed children to their mutual joy and comfort, and not only
when present, but sometimes when absent.’ ‘I even felt it,’ she adds,
‘to flow from me into their souls. When they wrote to me, they
informed me that at such times they had received abundant infusions
of divine grace.’—Ibid. part III. c. i.
Note to page 223.
Autobiography, part I. c. xiii. Here Madame Guyon has found

confessors blind guides, and confessions profitless; and furthermore,
she is encouraged and instructed in the inward life by a despised
layman. There is every reason to believe that the experience of
Madame Guyon, and the doctrines of the beggar, were shared to
some extent by many more. Madame Guyon speaks as Theresa
does of the internal pains of the soul as equivalent to those of
purgatory. (c. xi.) The teaching of the quondam mendicant
concerning an internal and present instead of a future purgatory, was
not in itself contrary to the declarations of orthodox mysticism. But
many were beginning to seek in this perfectionist doctrine a refuge
from the exactions of the priesthood. With creatures of the clergy like
Theresa, or with monks like John of the Cross, such a tenet would
be retained within the limits required by the ecclesiastical interest. It
might stimulate religious zeal—it would never intercept religious
obedience. But it was not always so among the people—it was not
so with many of the followers of Molinos. The jealous vigilance of
priestcraft saw that it had everything to fear from a current belief
among the laity, that a state of spiritual perfection, rendering
purgatory needless, was of possible attainment—might be reached
by secret self-sacrifice, in the use of very simple means. If such a
notion prevailed, the lucrative traffic of indulgences might totter on
the verge of bankruptcy. No devotee would impoverish himself to buy
exemption hereafter from a purifying process which he believed
himself now experiencing, in the hourly sorrows he patiently
endured. It was at least possible—it had been known to happen, that
the soul which struggled to escape itself—to rise beyond the gifts of
God, to God—to ascend, beyond words and means, to repose in
Him,—which desired only the Divine will, feared only the Divine
displeasure,—which sought to ignore so utterly its own capacity and
power, might come to attach paramount importance no longer to the

Wave Propagation in Structures Mechanical Engineering Series James F Doyle Online Ebook Texxtbook Full Chapter PDF

Uploaded by

Copyright:

Available Formats

You might also like

Wave Propagation in Structures Mechanical Engineering Series James F Doyle Online Ebook Texxtbook Full Chapter PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Wave Propagation in Structures Mechanical Engineering Series James F Doyle Online Ebook Texxtbook Full Chapter PDF

Uploaded by

Copyright:

Available Formats

Wave Propagation in Structures

(Mechanical Engineering Series) James

Radio Wave Propagation Fundamentals, Second Edition

Advanced Electromagnetic Wave Propagation Methods 1st

Wave Fields in Real Media: Wave Propagation in

Nonlinear Wave and Plasma Structures in the Auroral and

Aircraft Structures for Engineering Students (Aerospace

Fear No Evil Alex Cross 29 1st Edition James Patterson

Design of Steel-Concrete Composite Structures Using

Mastering Uncertainty in Mechanical Engineering 1st

More information about this series at http://www.springer.com/series/1161

ISSN 0941-5122 ISSN 2192-063X (electronic)

Had I the heavens’ embroidered cloths,

straight segments connected end to end, or directly in terms of coupled differential

West Lafayette, IN, USA James F. Doyle

2.3.2 The Spectrum Relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

4.4 Refined Beam Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155

6 Waves in Plates and Cylinders. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237

8.2 Panel Excitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 336

L Length of element, distance to boundary

BC, pBC Boundary condition, periodic BC

This book is an introduction to the spectral analysis method as a means of solving

What Is Spectral Analysis?

Structures and Waveguides

A structure can be as simple as a cantilevered diving board or as complicated

Wave Propagation and Vibrations

Themes and Threads

Property Common (US) SI

When a natural nondimensionalizing factor, such as co or the acceleration g,

1.1 Fourier Transforms

The continuous transform is a convenient starting point for discussing spectral

1.1.1 Continuous Fourier Transforms

where Ĉ(ω) is the continuous

F (t) = Fo − a/2 ≤ t ≤ a/2

In this particular case the transform is real-only and symmetric about ω = 0 as

A step function is therefore improper because Ĉ(0) is infinite.

and zero otherwise, the transform is then

-100. -50. 0. 50. 100.

Fig. 1.2 Continuous transform of a rectangular pulse with a = 50 µs

FA (t) + FB (t) ⇔ ĈA (ω) + ĈB (ω) (1.3)

This is an essential property of the transform and it means that if a signal is

where the following decomposition was used

cos θ = 12 [eiθ + e−iθ ] , sin θ = −i 12 [eiθ − e−iθ ] (1.6)

Using the inverse transform relation for FA gives

This can be further rearranged as

and is expressed as the transform pair

This particular form is called a convolution. We use in explaining the effects of

1.1.2 Discrete Fourier Transform

Fig. 1.3 Periodic extended signal as a superposition of infinite signals

Ĉ(ω) = Ĉ∞ (ω)[· · · + e+iω2T + e+iωT + 1 + e−iωT + e−iω2T + · · · ]

Ĉ(ω) = Ĉ∞ (ω)A[· · · + δ(ω + 2π/T ) + δ(ω) + δ(ω − 2π/T ) + · · · ]

where A is a proportionality constant which we determine next. We reiterate that

We now have the representation of a function F (t), extended periodically, in

-100. -50. 0. 50. 100.

Fig. 1.5 Discretization scheme for an arbitrary time function

Hence the summation contribution simply becomes

showing that it evaluates the same as when n = n∗ . More specifically, if M = 8

with T = 1, N = 8. This has the shape of a rectangular pulse if the points