Advanced Structural Damage Detection E-Book
114,99 €
Mehr erfahren.
- Herausgeber: John Wiley & Sons
- Kategorie: Fachliteratur
- Sprache: Englisch
Structural Health Monitoring (SHM) is the interdisciplinary engineering field devoted to the monitoring and assessment of structural health and integrity. SHM technology integrates non-destructive evaluation techniques using remote sensing and smart materials to create smart self-monitoring structures characterized by increased reliability and long life. Its applications are primarily systems with critical demands concerning performance where classical onsite assessment is both difficult and expensive.
Advanced Structural Damage Detection: From Theory to Engineering Applications is written by academic experts in the field and provides students, engineers and other technical specialists with a comprehensive review of recent developments in various monitoring techniques and their applications to SHM. Contributing to an area which is the subject of intensive research and development, this book offers both theoretical principles and feasibility studies for a number of SHM techniques.
Key features:
- Takes a multidisciplinary approach and provides a comprehensive review of main SHM techniques
- Presents real case studies and practical application of techniques for damage detection in different types of structures
- Presents a number of new/novel data processing algorithms
- Demonstrates real operating prototypes
Advanced Structural Damage Detection: From Theory to Engineering Applications is a comprehensive reference for researchers and engineers and is a useful source of information for graduate students in mechanical and civil engineering
Sie lesen das E-Book in den Legimi-Apps auf:
Seitenzahl: 536
Veröffentlichungsjahr: 2013
Ähnliche
Contents
List of Contributors
Preface
Acknowledgments
1 Introduction
1.1 Introduction
1.2 Structural Damage and Structural Damage Detection
1.3 SHM as an Evolutionary Step of NDT
1.4 Interdisciplinary Nature of SHM
1.5 Structure of SHM Systems
1.6 Aspects Related to SHM Systems Design
References
2 Numerical Simulation of Elastic Wave Propagation
2.1 Introduction
2.2 Modelling Methods
2.3 Hybrid and Multiscale Modelling
2.4 The LISA Method
2.5 Coupling Scheme
2.6 Damage Modelling
2.7 Absorbing Boundary Conditions for Wave Propagation
2.8 Conclusions
References
3 Model Assisted Probability of Detection in Structural Health Monitoring
3.1 Introduction
3.2 Probability of Detection
3.3 Theoretical Aspects of POD
3.4 From POD to MAPOD
3.5 POD for SHM
3.6 MAPOD of an SHM System Considering Flaw Geometry Uncertainty
3.7 Conclusions
References
4 Nonlinear Acoustics
4.1 Introduction
4.2 Theoretical Background
4.3 Damage Detection Methods and Applications
4.4 Conclusions
References
5 Piezocomposite Transducers for Guided Waves
5.1 Introduction
5.2 Piezoelectric Transducers for Guided Waves
5.3 Novel Type of IDT-DS Based on MFC
5.4 Generation of Lamb Waves using Piezocomposite Transducers
5.5 Lamb Wave Sensing Characteristics of the IDT-DS4
5.6 Conclusions
Appendix
References
6 Electromechanical Impedance Method
6.1 Introduction
6.2 Theoretical Background
6.3 Numerical Simulations
6.4 The Developed SHM System
6.5 Laboratory Tests
6.6 Verification of the Method on Aircraft Structures
6.7 Conclusions
References
7 Beamforming of Guided Waves
7.1 Introduction
7.2 Theory
7.3 Numerical Results
7.4 Experimental Results
7.5 Discussion
7.6 Conclusions
References
8 Modal Filtering Techniques
8.1 Introduction
8.2 State of the Art
8.3 Formulation of the Method
8.4 Numerical Verification of the Method
8.5 Monitoring System Based on Modal Filtration
8.6 Laboratory Tests
8.7 Operational Tests
8.8 Summary
References
9 Vibrothermography
9.1 Introduction
9.2 State of the Art in Thermographic Nondestructive Testing
9.3 Developed Vibrothermographic Test System
9.4 Virtual Testing
9.5 Laboratory Testing
9.6 Field Measurements
9.7 Summary and Conclusions
References
10 Vision-Based Monitoring System
10.1 Introduction
10.2 State of the Art
10.3 Deflection Measurement by Means of Digital Image Correlation
10.4 Image Registration and Plane Rectification
10.5 Automatic Feature Detection and Matching
10.6 Developed Software Tool
10.7 Numerical Investigation of the Method
10.8 Laboratory Investigation of the Method
10.9 Key Studies and Evaluation of the Method
10.10 Conclusions
References
Index
© 2013 John Wiley & Sons, Ltd
Registered OfficeJohn Wiley & Sons, Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, United Kingdom
For details of our global editorial offices, for customer services and for information about how to apply for permission to reuse the copyright material in this book please see our website at www.wiley.com.
The right of the author to be identified as the author of this work has been asserted in accordance with the Copyright, Designs and Patents Act 1988.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988, without the prior permission of the publisher.
Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books.
Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The publisher is not associated with any product or vendor mentioned in this book.
Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose. It is sold on the understanding that the publisher is not engaged in rendering professional services and neither the publisher nor the author shall be liable for damages arising herefrom. If professional advice or other expert assistance is required, the services of a competent professional should be sought.
MATLAB® is a trademark of The MathWorks, Inc. and is used with permission. The MathWorks does not warrant the accuracy of the text or exercises in this book. This book’s use or discussion of MATLAB® software or related products does not constitute endorsement or sponsorship by The MathWorks of a particular pedagogical approach or particular use of the MATLAB® software.
Library of Congress Cataloging-in-Publication Data
Stepinski, Tadeusz.Advanced structural damage detection : from theory to engineering applications / Tadeusz Stepinski, Tadeusz Uhl, Wieslaw Staszewski. pages cm Includes bibliographical references and index.
ISBN 978-1-118-42298-4 (hardback)1. Structural health monitoring. 2. Structural failures. 3. Materials–Testing. I. Uhl, Tadeusz. II. Staszewski, W. J. III. Title.TA656.6.S74 2013 624.1′76–dc23
2013005209
A catalogue record for this book is available from the British Library.
List of Contributors
Łukasz Ambrozi nskiAlberto GallinaKrzysztof HolakAndrzej KlepkaPiotr KohutMichał MańkaAdam MartowiczKrzysztof MendrokPaweł PaćkoŁukasz PieczonkaMateusz RosiekMariusz SzwedoThe above authors are affiliated with the Department of Mechatronics and Robotics, Faculty of Mechanical Engineering and Robotics, AGH University of Science and Technology, Poland.
Preface
Structural Health Monitoring (SHM) is an interdisciplinary engineering field that deals with innovative methods of monitoring structural safety, integrity and performance without affecting the structure itself or particularly impairing its operation. SHM utilizes several types of sensors – embedded in or attached to – a structure to detect the presence, location, severity and consequence of damage.
SHM technology integrates nondestructive evaluation (NDE) techniques using remote sensing and smart materials to create smart self-monitoring structures characterized by an increased reliability and long life. SHM primarily applies to systems with critical demands concerning performance where classical onsite assessment is related to high costs, is difficult or even impossible.
Written by academic experts in the field, this book will provide students, engineers and other interested technical specialists with a comprehensive review of recent developments in various monitoring techniques and their applications to SHM. By providing a comprehensive review of the main SHM techniques it contributes to an area which is the subject of intensive research and development. This book offers both theoretical principles and feasibility studies for a number of SHM techniques. It also presents a number of novel data processing algorithms and demonstrates real operating prototypes.
This book reports results of the research project MONIT conducted at the AGH University of Science and Technology (AGH-UST) in Kraków and supported by The European Regional Development Fund in the framework of the programme Innovative Economy. The book includes 10 chapters written by researchers active in the research team at the Department of Robotics and Mechatronics at the AGH-UST where the Editors are faculty members.
Chapter 1 contains an introduction that briefly presents SHM philosophy and explains its relationship to the traditional NDE techniques. The interdisciplinary character of SHM is outlined in the context of NDE and condition monitoring (CM). Structural damage and structural damage detection aspects are discussed and the main levels of SHM procedures are outlined. The structure of the SHM systems using global and local SHM methods is presented. Aspects related to the design process of SHM systems are discussed in the final part of the chapter.
Chapter 2 is devoted to the numerical simulation of elastic wave propagation in planar structures. A brief overview of the available numerical methods for elastic wave propagation is presented, which include: finite element methods (both implicit and explicit formulations), spectral element method, finite volume method and finite difference methods. Both theoretical background and practical aspects are considered in the overview. The main focus in this chapter is on the new implementations of LISA (Local Interaction Simulation Approach) which belongs to the finite difference methods. Recently, LISA has been rediscovered and found attractive on account of rapidly evolving techniques based on graphical processing units (GPUs). The GPU implementation of LISA developed at AGH-UST is presented and its performance is illustrated using benchmark modelling cases.
Chapter 3 presents an application of the model assisted probability of detection (MAPOD) for SHM systems under uncertain crack configuration and system variability. The phenomena related to a SHM system are studied using numerical simulations of planar structures monitored by transducer arrays. The monitoring reliability is evaluated using a three-dimensional model including both Lamb wave propagation in a plate-like structure and scattering from cracks with different configurations. Computer models are implemented using parallel processing technology, presented in Chapter 2, which significantly speeds up the simulation time. The configurations considered in the chapter account for the variations of the relative crack position, orientation and size.
Chapter 4 provides the state of the art in the area of applications of nonlinear acoustics to SHM. An extensive literature review dealing with physical mechanisms related to nonlinearities encountered in elastic materials in the presence of damage is provided as an introduction. Different stress–strain characteristics are considered and two particular effects, contact acoustics nonlinearity and nonlinear resonance, are presented in some detail. The principles of frequency mixing that occurs if two waves with different amplitude and frequency are introduced into a damaged structure are explained. The main part of the chapter contains a review of damage detection methods and their applications to metallic and composite structures as well as glass, plexiglas, concrete and rocks.
Chapter 5 discusses piezoelectric transducers used for generation and sensing surface and Lamb waves. After a short overview of the transducer designs used for that purpose, the focus is on the flexible transducers made of piezocomposite materials. A novel type of piezoelectric transducer, based on a macro-fibre composite (MFC), is presented. Contrary to the commonly used conventional MFC transducers with dense electrodes, the transducers presented here are provided with sparse interdigital electrodes matched to a certain wavelength. Two different designs of interdigital transducers (IDTs) made of MFC substrate are presented and compared with the classical MFC transducers with dense electrodes. The comparison includes the results of numerical simulations and experimental tests performed using a scanning laser vibrometer for the conventional MFC and the proposed IDT attached to an aluminium plate.
The electromechanical impedance (EMI) method is presented in Chapter 6. After providing a theoretical background, the measurement setups and signal processing algorithms used for damage detection are introduced. The EMI is illustrated with the results of numerical finite element simulations performed for simple mechanical structures, such as beams and plates. Experimental results obtained in the laboratory conditions for simple structures (e.g., aluminium plate and pipeline section) are also presented. Finally, results of the EMI measurements performed on two aircraft structures (bolted joint in the main undercarriage bay and riveted fuselage panel) are presented and discussed in terms of EMI feasibility for SHM.
Chapter 7 reviews methods used for selective focusing of Lamb waves using transducer arrays. After a brief introduction to the field of phased arrays, conventional beamforming techniques based on delay and sum (DAS) operation are discussed. First, a few example two-dimensional (2D) array topologies are compared in terms of spatial resolution based on the simulation results performed using a simplified model with point-like transducers and dispersion-free medium. In the second part, a new beamforming technique is presented, which is an extension of the DORT method (a French acronym for the decomposition of time reversal operator) where the continuous wavelet transform (CWT) is used for the time–frequency representation (TFR) of nonstationary snapshots. An application of the proposed technique to self focusing of Lamb waves in an aluminium plate is demonstrated both for a linear- and a 2D star-shaped array. It is shown that the decomposition of the time reversal operator obtained with the proposed method enables separation of point-like scatterers in the aluminium plate and allows the focusing of Lamb waves on a number of individual damages present in the plate.
Chapter 8 presents the theory and applications of modal filters with focus on their potential in SHM systems. Modal filters can be successfully used for global monitoring of large engineering structures. Structural modification of a mechanical system (e.g. drop in stiffness due to a crack) results in the appearance of peaks at the output of the modal filter. These peaks result from the imperfect modal filtration due to the system’s local structural changes. An SHM system based on modal filters is theoretically insensitive to environmental changes, such as temperature or humidity variation (global structural changes do not cause a drop in modal filtering accuracy). A number of practical implementations of the presented technique are provided including a description of the developed SHM system as well as the results of its extensive simulation, laboratory and operational testing. It is demonstrated that modal filters can create a valuable damage detection indicator characterized by low computational effort (due to the data reduction ability); moreover, an SHM system utilizing this concept is easy to automate.
Different ways of using thermographic measurements in NDE and SHM are discussed in Chapter 9. First, classification of the thermographic techniques focused on active thermography with internal excitation, including vibrothermography (VT), is presented. An overview of the measurement equipment including thermographic cameras and excitation sources is also provided. Numerical simulations of coupled thermo mechanical phenomena using explicit finite elements are presented. Hardware and software components of the VT measurement system developed at AGH-UST are provided and illustrated by the description of laboratory measurements performed using this system on composite and metallic samples. A parametric study of the influence of measurement parameters on the resultant thermal response is also provided. Finally, measurements performed on a military aircraft fuselage and wing panels are reported and the practical issues related to field measurements are discussed.
Chapter 10 is devoted to vision based monitoring systems. It starts from an overview of the background and related work in the field of vision based measurement methods; the systems available in the market are presented and their advantages and shortcomings are discussed. The following section focuses on the main steps of the developed method, such as camera calibration and scale coefficient calculation algorithms as well as image rectification using homographic mapping. Simulation tests of the developed method carried out using the developed programming tool as well as using the virtual model of the construction are reported. Results of the numerical investigations of the uncertainty propagation in the proposed algorithms are presented and discussed; the probability of damage detection is considered as well. Laboratory tests, carried out using a setup consisting of a steel frame loaded by a point force and two high resolution digital single-lens reflex (SLR) cameras, are reported. Results of the vision based method are compared with those obtained using other contact and non contact measurement techniques. Finally, an evaluation of the method on civil engineering construction is presented.
Acknowledgments
The research presented in this book has been supported by funding from the research project MONIT (No. POIG.01.01.02-00-013/08-00) co-financed by the European Regional Development Fund under the operational programme Innovative Economy and research project N N501158640 sponsored by the Polish National Science Center.
1
Introduction
Tadeusz Uhl1, Tadeusz Stepinski1,2 and Wieslaw Staszewski1
1Department of Mechatronics and Robotics, Faculty of Mechanical Engineering and Robotics, AGH University of Science and Technology, Poland2Signals and Systems, Department of Engineering Sciences, Uppsala University, Sweden
1.1 Introduction
It is widely accepted that maintenance of engineering structures is important to ensure structural integrity and safety. This is particularly relevant to civil engineering and transportation. Aerospace structures for example are inspected regularly. Airframes are monitored for possible fatigue cracks.
A variety of different Nondestructive Testing and Evaluation (NDT/E) methods have been developed for damage detection. Ultrasonic inspection and eddy current technique are good examples of mature, well-established technologies that are widely used for crack detection. NDT/E techniques are often limited to single-point measurements and require scanning when large areas need to be monitored.
In recent years there have been a range of new damage detection techniques and sensing technologies. These methods allow for global, online monitoring of large structures and fall into the area of Structural Health Monitoring (SHM). They are capable of achieving continuous monitoring for damage involving the application of new sensors. Damage monitoring systems, which often use advanced sensor technologies, are concerned with a new design philosophy. Actuators, sensors, and signal processing are integrated to offer progress in this area.
SHM involves integrating sensors and actuators, possibly smart materials, data transmission and computational power within a structure in order to detect, localize, assess and predict damage which can be a cause of structure malfunction now or in the future (Adams 2007; Balageas et al. 2006). A typical SHM system is associated with an online global damage identification in structures; such systems are most often applied in aerospace (Staszewski et al. 2004) and civil engineering (Wenzel 2005).
Although SHM systems utilize NDT/E methods as tools there are many differences in SHM and NDT/E operation principles. NDT/E techniques are commonly applied offline and locally in regions of expected damage while SHM methods should provide real time monitoring of a whole structure during its operation. SHM is a successive step in the evolution of structure diagnostics, which historically evolved from the damage detection concept implemented in the form of condition monitoring (CM) systems. CM systems should be capable of detecting damage based on a global assessment of technical structures during their operation. SHM is expected to go further by the ability of detecting damages in early stages of their development or, in an ideal case, of predicting their occurrence before they really take place (Inman et al. 2005).
1.2 Structural Damage and Structural Damage Detection
There are many different connotations of the term damage in mechanical structures which in the area of SHM damage can be understood intuitively as an imperfection, defect or failing which impairs functional and working conditions of engineering structures. A more precise definition of damage can be offered when system analysis is used. Structures can be modelled as systems with input excitations and output measurable signals. In this context damage can be considered as an additional excitation that results in energy flow and transformation, leading to modifications of output signals. Therefore damage detection is an inverse problem; measurable outputs are used to detect damage. Damage can also be regarded as a modification to material properties and/or structural physical parameters. These properties and parameters can be modified due to sedimentation and plasticity of material or fatigue and corrosion. In this context damage detection is an identification problem. Material properties and physical parameters need to be extracted to assess damage.
Many different damage detection methods have been developed in the last few decades. Altogether these methods can be classified into model based and signal based approaches. Vibration based methods often utilize physical and/or modal parameters, obtained from physical models, for damage detection. Models are also essential when loads are monitored to obtain information about structural usage. Signal based methods rely on various types of direct measurements such as noise, vibration, ultrasound or temperature.
Both, i.e. model and signal based, approaches require signal processing techniques; the former to develop appropriate models and to analyse changes in these models that are relevant to damage; and the latter to extract features and establish a relationship between these features and possible damage.
The majority of signal based methods rely on a relationship between structural condition and signal features or symptoms. This condition – symptom relationship is often not easy to analyse due to the complexity of engineering structures, sophistication of design and use of advanced materials. Different signal processing methods need to be used for the analysis. This includes signature and advanced signature analysis. The former is based on simple features such as statistical spectral moments or physical/modal parameters. The latter uses multidimensional features (e.g. vectors, matrices, images) such as spectra, signal instantaneous characteristics or time – frequency distributions.
In this context, damage detection can be regarded as a problem of pattern recognition. Pattern recognition requires feature selection procedures for training and is usually based on statistical, syntactic or neural approaches. Many recent studies in this area are based on new developments related to signal processing (Staszewski and Worden 2009) and machine learning (Worden et al. 2011). It is clear that these developments are essential for implementation of any SHM system.
Damage detection forms the primary objective of the overall problem of damage identification. SHM systems’ tasks can be classified as a process consisting of five activities that form five important elements or levels (Balageas et al. 2006; Cempel 1991; Rytter 1993), as shown in Figure 1.1.
Figure 1.1 Main levels of SHM procedure
These are: I, damage detection; II, damage localization; III, assessment of damage size; IV, remaining life prediction; and V, smart structures with self-evaluating, self-healing or control capabilities. In this context detection gives a qualitative indication that damage might be present, localization gives information about the probable position of damage, assessment estimates its severity by providing information about damage type and size and finally, prognosis estimates the residual structural life and predicts possible breakdown or failure. The first three levels (i.e. detection, localization and assessment) are mostly related to system identification, modelling and signal processing aspects. The level of prognosis falls into the field of fatigue analysis, fracture mechanics, design assessment, reliability and statistical analysis. This level is very intensively investigated in many laboratories but there are currently no commercially available solutions. All these levels require various elements of data, signal and/or information processing.
1.3 SHM as an Evolutionary Step of NDT
Damage detection/monitoring, NDT/E and SHM are often misunderstood as synonyms and may have the same meaning in many engineering areas. Damage, health and monitoring of structures can be described using various definitions. In general, health is the ability to function/perform and maintain structural integrity throughout the entire lifetime of the structure; monitoring is the process of diagnosis and prognosis and damage is a material, structural or functional failure. Also, in this context, structural integrity is the boundary condition between safety and failure of engineering components and structures. In aircraft maintenance, damage detection and direct monitoring of damage accumulation offers an alternative approach to loads monitoring.
Recent developments in SHM are related either to modifications of well-established techniques, new equipment and sensor technologies or new monitoring principles. This can be illustrated using three examples. First, Acoustic Emission (AE) is a well-established NDT technique used for damage detection for many years. However, when optical fibre sensors – that can be integrated with monitoring structures – are used, AE (passive NDT approach) can be combined with Lamb wave based damage detection (active SHM approach). Secondly, the first NDT/E application of Lamb waves goes back to the 1950s although significant progress was achieved when low profile, smart transducers (e.g. piezoceramic, polymer, discs, paints, fibres) were introduced in the early 1990s allowing a real SHM approach. Thirdly, new damage detection methods based on a nonclassical approach to nonlinear acoustics have been proposed recently offering good damage detection sensitivity.
NDT techniques are often limited to single point measurements and require scanning when large areas need to be monitored. There have been a range of new damage detection techniques and sensing technologies in recent years. SHM methods allow for global, online monitoring of large structures and also offer damage localization. These methods are capable of achieving continuous monitoring for damage with the application of new sensors. Damage monitoring systems that use smart sensor technologies are concerned with a design philosophy directed to the integration of actuators, sensors, and signal processing. There has been an enormous research effort in this area in the last 20–30 years.
SHM, damage detection/monitoring and NDT are often used replaceably to describe the process of nondestructively evaluating structural condition. However, only SHM defines the entire process of implementing a strategy that includes five important identification elements (or levels), as discussed above. What distinguishes SHM from NDT is the global and online implementation of various damage detection technologies which require periodically spaced measurements (or observations), as accurately pointed out in Adams (2007). This process of online implementation needs more advanced signal processing for reliable damage detection than classical NDT techniques.
NDT involves comparing the known input of a measured signal with a known model – it does not require sacrificing the physical system, as disassembly or failure testing would. NDT is usually carried out offline in a local manner, after the damage has been located, or periodically, to improve performance of a structure. NDT techniques are mainly used to characterize damages and assess their severity, if their location is known.
1.4 Interdisciplinary Nature of SHM
SHM is an interdisciplinary area of research that integrates such basic sciences as materials science, mechanics, electronics and computer science, and which is strongly related to structures and their life cycle. This is illustrated schematically in Figure 1.2.
Figure 1.2 Interdisciplinary nature of SHM
SHM approaches require three important elements for damage detection implementation. These are: (1) knowledge of monitored structures and possible damage scenarios; (2) sensors and instrumentation used to obtain signals/data that can be used for damage detection; and (3) relevant analysis which can extract information about possible damage and overall structural integrity. Modelling, numerical simulations and signal processing are important activities of all these three elements.
The interdisciplinary nature of SHM requires a dedicated approach during design, manufacturing and operation, therefore SHM systems are mostly installed on new structures and rarely on old ones that have been operating for a long time. Installation of SHM systems on structures with unknown history of operation is very difficult and the probability of correct damage assessment is much lower than for the new structures.
The design of an SHM system depends on the type of damages which can occur, type of materials applied for the design and physical phenomena employed for damage detection. The complexity of an SHM system design depends on the local nature of material damages that are most likely to occur in the assessed structure and that may not significantly influence the structure’s response measured normally during its operation, e.g. its low frequency vibration spectrum.
Another factor that makes SHM data from a damaged structure difficult to acquire is limited accessibility of its particular components during operation. This often requires an in-depth study of local structure behaviour with the application of analytical and simulation tools that are widely used for understanding damaged structure behaviour and characteristics of the related signals. Due to the relatively high cost of an SHM system implementation of a very careful design and optimization process is recommended, taking into account minimization of the influence of the system on the structure’s performance, minimization of the hardware costs (e.g. number of sensors and actuators), and maximization of the correctness of damage detection and assessment.
Nowadays, multi-physics and multiscale simulation are extremely valuable tools in designing SHM systems. The design process consists of several steps, the most challenging of which are: (1) selecting a phenomenon which is sensitive enough to the damages that are to be detected; (2) defining the required sensing system with self-validation capability; (3) selecting data acquisition and processing architecture; (4) defining feature extraction and information reduction procedures; (5) formulating and implementing procedure of damage detection; and last but not least, (6) damage localization and its size assessment.
There are no general rules on how to solve all these design and implementation problems for any structure. The design methods are dedicated to a given structure, used materials and chosen physical phenomena employed for health monitoring. SHM technology helps to achieve better operational safety and has an economic impact on decreasing maintenance and operating costs because it allows the prediction of possible damage a long time before its appearance and, consequently, gives operators enough time to plan a proactive service and maintenance action.
There are several disciplines that are very closely related to SHM:
CM – condition monitoring (Inman
et al.
2005);
NDT/E – Nondestructive Testing/Evaluation (Staszewski
et al.
2004);
SPC – statistical process control (Inman
et al.
2005);
DP – damage prognosis (Inman
et al.
2005);
MP – Maintenance Planning (Pietrzyk and Uhl 2005), for instance, RCM (Reliability Centred Maintenance).
CM is in many aspects very similar to SHM but in practical use it is dedicated to rotating and reciprocating machinery. The main features of the CM approach are: damage localization is approximately known as well as the type of damage – the number of possible damages is limited, databases with damage symptoms are available, the influence of environmental conditions on the measurement results is very slight, and the economic benefits from employment of CM procedures are well defined. An essential advantage of CM over SHM methods is the well defined economic benefits from the use of CM procedures as well as the fact that many standards used worldwide require the monitoring of rotating machinery.
On the other hand, SHM also has disadvantages: localization of damage is not known, there are difficulties in measurements, due to the limited admittance to the monitored structural components, the type of damage is often difficult to identify, the influence of environmental conditions on measurement results is significant, and the cost of SHM systems is relatively high, which is the reason for their application only on critical structures.
CM systems rely on measurements of structural responses during operation, but they do not use dedicated actuators to excite or trigger effects which can help to detect damage. The differences between CM, NDT and SHM systems in terms of integration of hardware with a structure are illustrated in Figure 1.3.
Figure 1.3 Schematic diagram of a typical (a) CM system, (0b) SHM system and (c) NDT system
The main difference between NDT and SHM systems can be noticed in the hardware architecture. In the case of a SHM system, sensors and actuators are built into (or integrated with) the structure, while NDT is an external system with an independent (not integrated with the structure) set of sensors and actuators [cf. Figure 1.3(c)]. Integration of SHM and NDT systems with the other tools is shown schematically in Figure 1.4.
Figure 1.4 Main components of an SHM system
The main difference between NDT/E and SHM is their implementation – NDT/E techniques are implemented offline while SHM ones are implemented online, which makes SHM tasks much more complex than the autonomous NDT/E applications.
SPC has similar aims to SHM but the final aim of SPC systems is not only detecting structural damages but process diagnostics – they use a variety of sensors to monitor changes in the process parameters. The process parameters can change due to structural failure and in this respect SHM and SPC are comparable.
DP is used to predict the remaining lifetime of operating structures during which their performance will remain above a given threshold. DP systems use the knowledge about damage size and location as well as expected operational loads. The remaining life prediction is based on a predictive model that acquires information from a usage monitoring system (the system that monitors loading cycle during the structure’s operation), the SHM system as well as the past, current and future environmental conditions and expected load levels.
Today’s DP systems give only a very rough estimation of remaining life prognosis, owing to the very complex physics of structure destruction if material level is to be considered. Multiscale simulation including molecular dynamics (Packo and Uhl 2011) methods can be helpful to solve this problem in the future. Chapter 2 contains recent results concerning modern numerical simulation methods.
The interaction between different types of monitoring systems in DP is shown diagrammatically in Figure 1.5 (Farrar et al. 2005). The most general is the MP system which defines requirements and tasks that are to be accomplished for achieving, restoring and maintaining the operational capability for the whole life of a structure.
Figure 1.5 Schematic diagram of a typical DP procedure. Adapted from Inman D, Farrar C, Lopes V and Steffen J. Damage Prognosis for Aerospace, Civil and Mechanical Systems, © 2005 John Wiley and Sons Ltd.
MP systems use data from the installed SHM system but also help to analyse historical data in order to detect events that could have been the reason for performance loss. This approach enables preventive service action before damage occurs. Several approaches can be distinguished within this discipline, one of the most useful for mechanical structures is RCM that helps minimize maintenance costs and minimize the risk of structural failure (Pietrzyk and Uhl 2005).
1.5 Structure of SHM Systems
A typical SHM system includes two main parts – a hardware section and an algorithmic section with software. The hardware part includes sensors and optional actuators and the units performing signal conditioning and acquisition, communication, and power supply. These components work autonomously and very often are embedded into the structure. Communication and power supply problems can be often encountered in this type of architecture, which calls for miniaturization of the applied hardware components. The problem with power supply can be solved by energy harvesting units that are currently the subject of many research projects.
Examples of already available technologies are shown in (Priya and Inman 2010). A feasible solution to the miniaturization of SHM hardware is designing and manufacturing MEMS chips dedicated to the SHM purpose, which include sensors, actuators, communication units and processors integrated on board (Staszewski et al. 2004). Wireless communication is one of the most commonly used data transfer type in SHM systems, there are dedicated solutions but some of the commercially available, such as zigbee or radio-frequency identification (RFID), can also be applied (Dargie and Poellabauer 2010; Lynch and Loh 2006; Uhl et al. 2007).
The software part contains basic procedures for signal processing, signal fusion, hardware control, structure health detection and remaining life prognosis. More advanced systems contain also some procedures related to structural health management.
1.5.1 Local SHM Methods
Modern SHM approaches can be classified into two main groups, the global methods (Adams and Farrar 2002; Doebling et al. 1996; Uhl 2004) and local methods (Grimberg et al. 2001; Maldague 2007; Raghavan and Cesnik 2007). Local methods monitor a small area of structure surrounding the sensor (sensors) using measurements of a structural response to certain applied excitation. Ultrasonic waves (Raghavan and Cesnik 2007), eddy currents (Grimberg et al. 2001), thermal field (Maldague 2007; Uhl et al. 2008) and acoustic emission (Pao 1978) are examples of phenomena that are most commonly employed for local SHM. The methods that are most often used for the design of SHM systems are guided waves (GWs; Raghavan and Cesnik 2007), those based on FBG sensors (strain, temperature measurements and ultrasound sensing) (Betz 2003), vibrothermography (Uhl et al. 2008) and electromechanical impedance (Bhalla and Soh 2004; Park et al. 2003). The state-of-art of the vibrothermography and electromechanical impedance methods can be found, respectively, in Chapters 9 and 6.
There are many other methods that can also be classified as local SHM (Chung 2001) but those methods are mostly used for more specific applications. Classical NDT methods that rely on the characteristics of ultrasound waves propagating in solid bodies can be used in combination with different signal processing and damage imaging techniques.
In the context of SHM, however, the waves are generated by permanently installed actuators that are integrated within a structure. The response is measured by a built-in set of piezoelectric sensors. There are many different techniques employed for excitation and sensing GWs. Recent results in the area of piezoelectric transducers made of macro-fibre composite (MFC) are presented in Chapter 5.
In thin plate-like structures elastic waves can propagate in the form of Lamb waves and the methods making use of these type of GWs are one of the most often proposed local methods in SHM. GW based techniques can be applied for both metallic and composite structures. Generally, local GW based methods require dense sensor networks (to provide a large number of measurement points distributed in space) which generate a large amount of data that have to be processed in order to detect, localize and assess structural damage. The cost of installation of such sensor networks is usually much higher than that for the global methods. Using two-dimensional (2D) phased arrays, capable of electronic beamforming, instead for the dense sensor networks has been proposed as a feasible solution to this problem; details are given in Chapter 7.
Local GW based methods have a number of advantages, the most important is their ability to monitor structural parts without the need for disassembly. Also, since the wavelength of GWs is in the same range as damage dimension they are sensitive to small damages. However, they also have an essential disadvantage – they require dense sensor networks or sophisticated phased arrays that need to be located in proximity to the potential damage, which means that knowledge of the critical damage location (hot spot) is of primary importance. Therefore, local methods are applied when critical structures are to be tested and early phase of damage has to be detected, and the high cost of the SHM system is acceptable.
1.5.2 Global SHM Methods
The global methods are performed if global motion of the structure is induced during its operation (Adams 2007; Balageas et al. 2006). Vibration based methods belong to this class. The global methods make use of the fact that local damage, for instance, local stiffness reduction, has an influence on the global structure’s behaviour in terms of time and space.
In comparison with local methods, global methods have a number of essential advantages: they can monitor the whole structure using a rough sensor network, sensors do not necessarily have to be located close to damage, and only a limited knowledge about critical location is sufficient. Obviously, global methods also have disadvantages, e.g. the wavelength of the naturally excited vibrations is approximately equal to the dimension of the structure or component and so they have relatively low sensitivity to small damages (especially for lower vibration modes).
Although global methods give only a rough estimation of damage location and size they can be successfully used for damage detection. The most commonly used global methods are vibration based methods (Balageas et al. 2006). Low frequency vibrations have been applied for diagnostic purposes for many years (Inman et al. 2005; Wenzel 2005). The effects of material defects, supporting structures’ failures or geometry defects on vibration response of a structure are well known. The relationship between structural vibration and damages of structures is used in their health assessment.
Two types of methods can be distinguished among global methods: signal based (Inman et al. 2005) and model based (Natke and Cempel 2000). The signal based methods utilize relations between measured responses of the structure after ambient excitation and possible damages. Signal features in frequency, time and time/frequency domains are the most popular now. The methods are very commonly applied in rotating and reciprocating machinery diagnostics for damage detection, but localization and damage assessment need additional information.
The model based methods employ many different types of models of a monitored structure to detect and localize damage in the structure using relations between the model parameters and particular damages. The idea behind the method is shown in Figure 1.6 (Uhl 2004).
Figure 1.6 Schematic diagram of model based global SHM method
Models of undamaged structure and damaged structure are compared for their parameters or output and differences (residues) are related to given damage and help to localize it. One of the most commonly used models in SHM is a modal model, which can be identified on a real structure with the use of measured external excitation (or vibration excitation caused by operation) and measurements of structural responses at many points.
The modal methods monitor the whole structure by detecting shifts of natural frequencies, increases in damping or changes of vibration modes’ shapes. The selected feature should be damage-sensitive. Modal model based techniques can be classified into the following groups (Uhl 2004):
methods based on perturbation of modal parameters (natural frequency, modal damping);
methods based on frequency response function (FRF) (stiffness and compliance) variation detection;
methods based on mode shape analysis;
methods based on detection of modes’ energy;
methods based on finite element (FE) model updating.
The methods based on modes’ shape analysis, such as strain energy analysis or mode shape curvature analysis, are preferred despite the fact that the required SHM system is then more complex than for SHM systems based on natural frequency and modal damping.
The global model based SHM procedures require neither dense sensor network nor sensors located in the vicinity of damage. These methods, however, are less sensitive and have lower spatial resolution compared with the local ones. Their sensitivity and spatial resolution can be improved by a computational model that interprets changes of dynamic properties of a structure.
The global model based methods are employed mainly for the SHM of civil structures. There are several issues that limit the application of these methods, the first is the high cost of a monitoring system caused by very complex cabling. The second is the relatively high influence of environmental conditions on structural dynamic properties, which means that this influence may sometimes dominate in comparison with that caused by serious damage of the structure. The first problem can be solved by a wireless sensor based monitoring system (Uhl et al. 2007) and the second one by a special environmental filter which is based on a modal filter [Mendrok and Uhl (2008) and Chapter 8].
Since the global methods are much less sensitive to damage than the local ones, in practical applications, they are used only for damage detection.
1.6 Aspects Related to SHM Systems Design
The entire process of monitoring for possible damage in engineering structures depends on structural design concepts. This can be illustrated using aircraft design. Current design principles of aircraft structures are based on the safe-life concept. Load spectra representative of typical operational conditions are first determined. This requires a significant amount of data related to mission profiles, mass distributions and many other parameters. The load spectra and fracture mechanics are then used to evaluate structural components in terms of their service fatigue life. This is followed by a series of fatigue tests of materials, coupons, elements, subcomponents and components, leading finally to the Major Airframe Fatigue Test (MAFT).
In practice, the scatter in design input data (e.g. unknown parameters, change of load conditions, variation of material properties, quality of manufacturing, human errors or structural modifications in service) is quite significant. Therefore various safety factors are imposed on the structure to guarantee the safe fatigue life. The structure is designed for a specific number of flight hours and retired from service afterwards even if no failure occurs.
The estimation of operational life of ageing aircraft is even more difficult. The safe-life design concept, illustrated in Figure 1.7, leads in practice to structures that are safe but over-designed. This is not desirable if economy and performance are analysed. Noncritical structural components that are exposed to multiple load paths are often designed using the fail-safe concept. Even if these components develop damage, the structural integrity is not jeopardized due to the assumption that damage can be detected before any catastrophic failure. This requires periodic inspections of components.
Figure 1.7 Safe-life aircraft design concept
Monitoring techniques offering reliable detection, location, estimation of severity and prognosis of damage can lead to the damage-tolerance design concept. Detected damage is monitored to maintain the safe life of aircraft in this design concept. Although significant inspection effort is required, this concept can lead to lighter structures and better performance. In fact, the prevention of crack initiation behind the safe-life design concept does not prevent catastrophic failures. Therefore maintenance and inspection of aircraft structures is very important whatever the design concept. It is important to note that fatigue of materials in Aerospace Engineering has significantly contributed to structural design. The safe-life and fail-safe design concepts, introduced in aerospace, are widely used in many other areas of engineering.
1.6.1 Design Principles
The design of SHM systems requires dedicated procedures and tools because of the costs, the high level of responsibility and the interdisciplinary nature of system design. The procedure for SHM system design consists of the following steps:
A careful description of the structure, environmental conditions and expected load ranges as well as a list of possible damages are necessary to choose appropriate physical phenomena which can be used for structural monitoring. The chosen phenomenon has to be measurable for a particular structure and operation conditions, however, its normal operation should not be disturbed. It is preferred that the phenomenon can be excited by ambient excitation, but in other cases a special actuation system should be designed to excite the structure. The excitation level should be as low as possible because of the limited power supply and high enough to cause measurable responses of the tested structure. While formulating the algorithm for correct damage detection results, the required computational power and volume of data from measurements have to be considered. Minimal computational power and minimization of the amount of required measurement data are important due to the limited availability of power supply and limited memory of the embedded computer power, especially when a wireless embedded SHM system is under design.
The next step in the design procedure is testing the formulated solution by simulation, which requires creating models of the monitored structure, damage scenario, sensors and actuators. The goal of this simulation is to create tools that enable testing of the correctness of the formulated method and choosing a measurement system that will enable measurements of the structural response with sensitivity matched to the monitored damages. Sensitivity analysis of a structure’s response to damage accuracy is a basic tool which helps to design a monitoring system. With the use of the model and its simulation the POD can be tested too (Mendrok and Uhl 2008). For a chosen SHM system, the POD should be as high as possible in order to consider the formulated method as useful in the design of SHM for a given structure. But due to undefined localization of the damage and fixed position of the applied sensors, the model based assessment of POD can be a difficult task [Gallina et al. (2011) and Chapter 3].
A much more accurate determination of POD can be done by applying an experimental laboratory test. In such a test damage is defined by a sample with known damage location and dimension and well known properties of a healthy structure. The laboratory test should confirm the sensitivity and probability level of the correct damage detection.
The next step is implementation of the designed SHM system, which requires conformity with related standards that depends on the application area, e.g. aviation standards are completely different from those that have to be obeyed in civil engineering. This is the main reason for completely different designs of SHM systems in both branches.
The modularity of SHM due to the requirement of scalability is an important feature during the implementation phase. In subsequent phases the SHM system is installed on a real structure in the operation environment. The operation test is always required to confirm the correctness of the system design.
A further development of SHM systems requires new automatic algorithms of damage detection, localization and assessment, new state prognosis methods and algorithms, and development of self-diagnosis and self-healing of critical structures.
References
Adams D 2007 Health Monitoring of Structural Materials and Components. John Wiley & Sons, Ltd.
Adams D and Farrar C 2002 Identifying linear and nonlinear damage using frequency domain ARX models. Structural Health Monitoring1, 185–201.
Balageas J, Fritzen C and Guemes A (eds) 2006 Structural Health Monitoring Systems. ISTE.
Betz D 2003 Acousto-ultrasonic sensing using fiber Bragg grating. Smart Materials and Structures12(1), 122–128.
Bhalla S and Soh C 2004 Structural health monitoring by piezo-impedance transducers. ASCE Journal of Aerospace Engineering35, 154–165.
Cempel C 1991 Vibroacoustic Condition Monitoring. Ellis Horwood.
Chung D 2001 Structural health monitoring by electrical resistance measurements. Smart Materials and Structures10, 624–636.
Dargie W and Poellabauer C (eds) 2010 Fundamentals of Wireless Sensor Networks: Theory and Practice. John Wiley & Sons, Ltd.
Doebling S, Farrar C, Prime M and Daniel W 1996 Damage identification and health monitoring of mechanical systems from changes of their vibration characteristics a literature review. Technical report, LA-13070 MS.
Farrar CR, Lieven NAJ and Bement M 2005 An introduction to prognosis. In Damage Prognosis for Aerospace, Civil and Mechanical Systems. John Wiley & Sons, Ltd.
Gallina A, Packo, PP, Ambrozinski L, Uhl T and Staszewski WJ 2011 Model assisted probability of detection evaluation of a health monitoring system by using CUDA technology. Proceedings of IWSHM 2011 (ed. Chan FK), Stanford University.
Grimberg R, Premel D, Savin A, Le Bihan Y and Placko D 2001 Eddy current holography evaluation of delamination in carbon epoxy composites. Insight34, 260–264.
Inman D, Farrar C, Lopes V and Steffen J (eds) 2005 Damage Prognosis for Aerospace, Civil and Mechanical Systems. John Wiley & Sons.
Lynch J and Loh K 2006 A summary review of wireless sensors and sensor networks for SHM. The Shock and Vibration Digest38(2), 91–128.
Maldague X 2007 Nondestructive Testing of Materials Using Infrared Thermography. Springer.
Mendrok K and Uhl T 2008 Modal filtration for damage detection and localization. Structural Health Monitoring 2008: Proceedings of the Fourth European Workshop. Kraków, Poland.
Natke G and Cempel C 2000 Model Based Diagnostics. Springer.
Packo P and Uhl T 2011 Multiscale approach to structure damage modelling. Journal of Theoretical and Applied Mechanics49, 243–264.
Pao Y 1978 Theory of acoustic emission. Elastic Waves and Non-destructive Testing of Materials29, 107.
Park G, Sohn H, Farrar CR and Inman DJ 2003 Overview of piezoelectric impedance based health monitoring and path forward. The Shock and Vibration Digest35, 451–463.
Pietrzyk, A and Uhl T 2005 Use of RCM methodology for railway equipment maintenance optimization. Archives of Transport49(1), 6583.
Priya S and Inman D (eds) 2010 Energy Harvesting Technologies. Springer.
Raghavan A and Cesnik CES 2007 Review of guided waves structural health monitoring. The Shock and Vibration Digest39, 91–114.
Rytter A 1993 Vibration Based Inspection of Civil Engineering Structures. PhD thesis, Department of Building Technology and Structural Engineering, Aalborg University, Denmark.
Staszewski WJ, Boller C and Tomlinson GR (eds) 2004 Structural Health Monitoring Systems of Aerospace Structures. John Wiley & Sons, Ltd.
Staszewski W and Worden K 2009 Signal processing for damage detection. In Encyclopedia of Structural Health Monitoring. John Wiley & Sons, Ltd. pp. 415–421.
Uhl T 2004 The use and challenge of modal analysis in diagnostics. Diagnostyka30, 151–160.
Uhl T, Hanc A, Tworkowski K and Sekiewicz L 2007 Wireless sensor network based bridge monitoring system. Key Engineering Materials347, 499–504.
Uhl T, Szwedo M and Bednarz J 2008 Application of active thermography for SHM of mechanical structures. Structural Health Monitoring 2008: Proceedings of the Fourth European Workshop. Kraków, Poland.
Wenzel H (ed.) 2005 Ambient Vibration Monitoring. John Wiley & Sons, Ltd.
Worden K, Staszewski WJ and Hensman JJ 2011 Natural computing for mechanical systems research: A tutorial overview. Mechanical Systems and Signal Processing25(1), 4–111.
2
Numerical Simulation of Elastic Wave Propagation
Paweł Paćko
Department of Mechatronics and Robotics, Faculty of Mechanical Engineering and Robotics, AGH University of Science and Technology, Poland
2.1 Introduction
Various physical phenomena are employed for damage detection and localization in health monitoring systems. Recently, elastic waves received a lot of interest due to their inspection potential and nondestructive nature. Particular attention is paid to guided ultrasonic waves, such as Lamb waves (Boller et al. 2009; Staszewski et al. 2004) that are exploited for damage detection in the maintenance of engineering structures. A broad review on the application of Lamb waves for damage detection in different structural components and materials may be found in Raghavan and Cesnik (2007), Staszewski (2004) and Su and Ye (2009). Despite a considerable amount of research effort, real engineering applications of Lamb waves are still limited. This is due to the complexity associated with Lamb waves propagation, i.e. multimodal nature, dispersion, possible reflections from boundaries and other structural features that produce a wave field that is quite difficult to analyse even without taking into account scattering from defects. Thus, methods that help to understand these phenomena are sought.
Numerical simulations are frequently used to predict the possible wave field, providing an insight into mechanisms that drive the wave interaction with structural features. This enables employment of simulation methods in the design, operation, and improvement of SHM and NDT/NDE systems as a mean of optimizing the system’s characteristics, as well as a part of the system to support damage detection and localization. At the design stage of a monitoring system, numerical simulation helps to adjust the settings, choose optimal sensors’ positions or examine the system’s response for various damage configurations. This may be performed virtually, reducing the number of test samples and physical experiments, and minimizing the time required for the monitoring system design. During the operation phase, the simulation may be employed as a part of the system providing a virtual baseline result or in an inverse analysis for the identification problem. Finally, the conclusions drawn during the virtual testing of a monitoring system lead to an improvement of its performance and inspection capabilities.
Considering the inspection procedure of a structure, the task of simulation is to support an SHM system at as many levels as possible. Starting with the fundamental task of an SHM system, damage detection, some demands on the simulation method for wave propagation may be proposed.
First, once the goal of the system is to detect damage at an early stage, it is necessary to employ high frequency waves. This is because the waveform is significantly affected only by an obstacle with a characteristic dimension that is at least of the order of a wavelength. Despite the increased attenuation, short wavelengths must be employed which implies high spatial resolution of a numerical model. Secondly, due to the high wave speeds for a metallic or composite material, high temporal resolution of the simulation method needs to be applied. These two requirements impose serious limitations on the modelling method, leading directly to large model sizes and long computational times.
The crucial aspect of the simulation of the SHM, as well as NDE/NDT, systems is the modelling of wave interaction with damage. Depending on the material of a structural component, different damage types are considered. Typical damages in metallic structures are fatigue cracks, cavities (e.g. from the casting process), folds and laps introduced in forming operations, highly deformed regions (shear band localizations), degraded regions, corrosion, etc. Cracks may produce a highly nonlinear response to waves due to small scale plasticity effects, interfacial friction at crack faces, thermoelastic effects and others (Buehler 2008; Klepka et al. 2011). On the other hand, the most common damages in composite materials are delaminations, debondings, folds, fibre, matrix, and fibre/matrix interface failure, degradation of material properties due to environmental conditions, the temperature dependence of the structure’s properties, etc. A reliable damage model is key to the success of the simulation method for SHM.
The wave interaction with damage may be thought of as an interaction with an interface of two media of high impedance mismatch that may additionally be a source of nonlinear effects. This requires special boundary conditions and a very accurate modelling approach, otherwise the structure’s response to damage will be of no use and the modelling method discredited.
The possibility to simulate such complex phenomena and the wave interaction with damage allow for full exploitation of the monitoring system’s potential. This requires sophisticated modelling techniques which are described in the following section.
2.2 Modelling Methods
This section reviews the selected numerical methods that are employed for the simulation of wave propagation phenomena for health monitoring applications. The key ideas and the necessary mathematical background are provided. Since the main emphasis is on the time evolution of mechanical systems, the time integration techniques are also briefly covered.
2.2.1 Finite Difference Method
The finite difference method (FDM) is one of the most popular methods for wave propagation simulation. The framework for the method was given by Courant, Friedrichs and Lewy in the 1930s (Courant et al. 1928). It was further developed during the Second World War and later in the 1950s and 1960s along with the development of computing units. Significant progress for time-dependent problems was made by Crank and Nicolson (1947), Lax and Wendroff (Lax 1961; Lax and Wendroff 1960) and Newmark (1959) by establishing time integration methods and stability criteria. Fundamental contributions to systematization of the theory were made by Collatz (1966, 1986). The method has been widely used in geophysics since then (Abramovici and Alterman 1965; Alford et al. 1974; Alterman and Karal 1968; Twizell 1979).
The FDM is an approach to solving differential equations by substituting derivatives by finite difference formulas. Therefore the original equation is transformed into an algebraic equation. There is a wide variety of difference formulas that can be used for spatial and temporal derivatives (Strikwerda 2004). The fundamental relation for deriving difference operators is the Taylor series:
(2.1)
which relates the unknown value of a function for a point x+h to the known function derivatives at x and the distance at which the value is evaluated, h.
Equation (2.1) may be truncated after a certain number of terms producing a truncation error, and, assuming that the function values are known, used to evaluate the function’s derivative. The simplest calculations obtained by truncation of series (2.1) after the second term are the so-called forward difference estimate or backward difference estimate (Woolfson and Pert 1999):
(2.2)
Equation (2.1) may be also combined for different points to obtain various finite difference formulas (Strikwerda 2004; Woolfson and Pert 1999).
It is worth noting that formulas obtained by using Equation (2.1) may be used for time or space discretization.
Once a given differential equation is transformed into a finite difference algebraic equation, it is implicitly assumed that both, time and space, have been discretized with certain steps Δt and Δx, respectively. The discretization process leads to discrete steps in time and the domain subdivision into small pieces, called elements or cells. In FDMs it is frequently assumed that the considered domain is composed of cuboidal cells with possibly different edge lengths. Nevertheless, due to the implementation issues it is sometimes convenient to deal with cubic cells. This means that the governing equation of a given problem is evaluated in its discrete form only in nodal points of the grid.
Such an approach to domain discretization leads to a well-known ‘staircase’ problem of curvature representation. A significant drawback of a regular mesh is the fact that the whole domain of interest is covered by cells of the same size, and cannot be refined where needed, i.e. in corners, fillets, etc., and coarse in other regions of the model, without a special treatment.
On the other hand, the FDM produces a number of relatively simple algebraic equations which can be processed very efficiently by a parallel computing unit. Moreover, due to the same equation structure for each grid point and regular grid, each processor is calculating a similar thread which reduces the computational time significantly.
