Systemic Approach for the Maintenance of Complex Structural Systems

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/233692630 Systemic approach for the maintenance of complex structural systems Article in Structure and Infrastructure Engineering · July 2007 DOI: 10.1080/15732470601155235 CITATIONS READS 31 32 3 authors: Franco Bontempi Konstantinos Gkoumas 209 PUBLICATIONS 772 CITATIONS 64 PUBLICATIONS 212 CITATIONS Sapienza University of Rome SEE PROFILE Sapienza University of Rome SEE PROFILE Stefania Arangio Sapienza University of Rome 38 PUBLICATIONS 163 CITATIONS SEE PROFILE All content following this page was uploaded by Franco Bontempi on 20 January 2017. The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document and are linked to publications on ResearchGate, letting you access and read them immediately. This article was downloaded by: [Arangio, Stefania] On: 23 May 2010 Access details: Access Details: [subscription number 790726812] Publisher Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 3741 Mortimer Street, London W1T 3JH, UK Structure and Infrastructure Engineering Publication details, including instructions for authors and subscription information: http://www.informaworld.com/smpp/title~content=t713683556 Systemic approach for the maintenance of complex structural systems Franco Bontempi a; Konstantinos Gkoumas a;Stefania Arangio a a Department of Structural and Geotechnical Engineering, Faculty of Engineering, University of Rome 'La Sapienza', Italy First published on: 14 September 2007 To cite this Article Bontempi, Franco , Gkoumas, Konstantinos andArangio, Stefania(2008) 'Systemic approach for the maintenance of complex structural systems', Structure and Infrastructure Engineering, 4: 2, 77 — 94, First published on: 14 September 2007 (iFirst) To link to this Article: DOI: 10.1080/15732470601155235 URL: http://dx.doi.org/10.1080/15732470601155235 PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf This article may be used for research, teaching and private study purposes. Any substantial or systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material. Structure and Infrastructure Engineering, Vol. 4, No. 2, April 2008, 77 – 94 Systemic approach for the maintenance of complex structural systems FRANCO BONTEMPI, KONSTANTINOS GKOUMAS and STEFANIA ARANGIO* Department of Structural and Geotechnical Engineering, Faculty of Engineering, University of Rome ‘La Sapienza’, Via Eudossiana, 18, 00184, Italy Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 (Received 20 January 2005; accepted in revised form 7 September 2006) This paper deals with the general framework for the development and the maintenance of complex structural systems. In the first part, starting with a semantic analysis of the term ‘structure’, the traditional approach to structural problem solving has been reconsidered. Consequently, a systemic approach for the formulation of the different kinds of direct and inverse problems has been framed, particularly with regards to structural design and maintenance. The overall design phase is defined with the aid of the performance-based design (PBD) philosophy, emphasizing the concepts of dependability and enlightening the role of structural identification. The second part of the present work analyses structural health monitoring (SHM) in the systemic way previously introduced. Finally, the techniques related to the implementation of the monitoring process are introduced and a synoptic overview of methods and instruments for structural health monitoring is presented, with particular attention to the ones necessary for structural damage identification. Keywords: Structural systems; Systems engineering; Performance-based design; Dependability; Structural health monitoring; Complexity 1. Introduction The theoretical framework for the design of complex structural systems should be based on a comprehensive evaluation of all the performances. The aim of structural engineering is not only to achieve an ideally good design and a nominal construction, but also to assure, by means of appropriate maintenance, the long term exploitation of the system. In this perspective, the procedures based on the concepts of system engineering can be profitably applied. System engineering is a robust approach to the creation, the design, the realization and the operation of systems. Complex structural systems should be designed using this innovative approach; in fact, a structure is a real physical object inserted in its environment in which a variety of factors should be taken into consideration. The traditional approaches are not able to consider the numerous interrelated aspects (qualitative and quantitative) characteristic of the system complexity. Furthermore, the recent improvement in data measurement and in elaboration technologies has created the proper conditions to improve the decisional tool based on the performance on site, leading to a new systemic design philosophy based on the performances, the so-called performance-based design (PBD). Starting from the identification and the quantification of system goals and requirements, once the performances are fixed, a correct and robust design can be implemented. When the design is complete, a verification phase must take place to validate and to assess the design in accordance with the specifications previously chosen. Under this premise, dependability provision and assessment during the whole life of the structure should be performed for the system and *Corresponding author. Email: stefania.arangio@uniroma1.it Structure and Infrastructure Engineering ISSN 1573-2479 print/ISSN 1744-8980 online ª 2008 Taylor & Francis http://www.tandf.co.uk/journals DOI: 10.1080/15732470601155235 78 F. Bontempi et al. its components. In this way, serviceability and safety can be assured and the overall service life can be increased. In this context, a structural health monitoring system assumes an essential role and should be designed as an integral part of the global structural project. In the following, we will show that structural health monitoring can be implemented according to two, often intertwined, approaches: one low level approach which considers the materials (or the basic components) that comprise the structure, and one high level approach which considers the structure (or the sub-structures within the structure) as a whole. 2. The structural system as an extension of the concept of the structure Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 2.1 Traditional approach to structural problem solving There are different ways to introduce the concept of ‘structure’; a good definition is a device for channelling loads that result from the use and/or presence of the building to the ground (Schodek 1998). Traditionally, the diverse problems related to the structures are classified as direct and inverse, as shown in figure 1 (Arangio 2004). A direct problem is an analysis problem; it consists of the evaluation of the response of a structure immersed in the so-called design environment, i.e. under assigned external actions and other boundary conditions, and the agreement with the fulfilment of all the design constrains, by using a suitable model. Inverse problems are, on the other hand, those for which the structural response constitutes available known data. Yet, with reference to figure 1 and according to the nature of the other unknowns as shown in figure 2, the inverse problems can be classified into: Synthesis problems: given the actions and the constraints, the structure is designed to obtain a specific structural response; Control problems: given a description for the structure and the mandatory structural response, an appropriate action to generate that response is searched; Identification problems: given both the actions and the structural response, the model is looked for. 2.2 The systemic framework for structural problems Traditionally, the solution of the structural problems is obtained from the input data by applying a set of rules or equations that lead to the desired output. Nowadays, an approach to the resolution of structural problems that is strictly related to the application of general rules or equations really seems to be both a restrictive and not very effective one. In fact, the necessity to build more and more complex structures emphasizes the presence and correlation of numerous factors that contribute to determining its characteristics. A global approach considers the structure as a real physical object inserted in its environment where a variety of factors should be taken into consideration. In this way, it is possible to contemplate aspects associated both with the intricacy and the uncertainties of the problem at hand. Some of these aspects, and it is worth realizing that these are the most important ones, belong to economics or Figure 1. Direct and inverse problems in structural engineering. Maintenance of complex structural systems 79 Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 Figure 2. Overview of data related to structural problems. politics, i.e. to the social spheres (Blanchard 1997). The fact that different points of view interact reciprocally implicates that a substantial variation of any of these points may eventually change the characteristics of the system as a whole. Just these connections form one of the signs of complexity and an approach that does not take into account this reality may be short-sighted. It is useful therefore to reconsider the previously given definition of structure; a structure is better defined as ‘a physical entity having a unitary character that can be conceived of as an organization of positioned constituent elements in space in which the character of the whole dominates the interrelationship of the parts’ (Schodek 1998). This definition highlights that a modern approach has to evolve from the idea of ‘structure’ as a simple device for channelling loads, to the idea of a ‘structural system’ as ‘a set of interrelated components which interact with one another in an organised fashion towards a common purpose’ (NASA 1995). This systemic approach includes a set of activities that lead and control the overall design, implementation and integration of a complex set of interacting components. It has been said that the notion of structural systems is a ‘marriage of structural engineering and systems science’ (Skelton 2002). From the operative point of view, the previous definition of structural system points out some key concepts related to the systemic approach, those of reductionism and integration (Sydenham 2004). Effectively, in the presence of complex structural problems, the systemic approach applies the widely used human thinking process where a large complex topic is broken down into smaller parts. This paradigm of problem solving is known as reductionism and constitutes a basic thinking methodology of engineering; it is the top-down process of forward engineering and it can be represented graphically (as shown in the left hand side of figure 3) by a pyramid, set up with various objects positioned in a hierarchic manner. The peak of the pyramid represents the goal (the whole structural problem), the lower levels represent a description of fractional objects (the sub-problems in which the problem can be divided), and the base corresponds to basic details. As a result, a project is intended and estimated with increasing levels of detail one level at a time. On the other hand, in those situations where the details comprise a starting point, a bottom-up approach is used for the integration of low-level objectives into more complex, higher-level objectives (as shown in the right hand side of figure 3). In common practice, however, real world problems are unclear and lack not only of straightforward solution process, but also a single point of view. In fact, frequently in the structural domain, the goals and the details are known, while uncertainties lie in the activities that take place in the intermediate levels. These activities are mostly based on experience and are carried out through successive refinements. In this case, the strategy (illustrated in the central part of figure 3) becomes a mixed recipe of top-down and bottom-up procedures that may be used alternately with the techniques of reverse-engineering. In this way, one analyses iteratively a system in order to identify its components and their interrelationships, possibly creating alternative useful representations of the system. During the solution procedure, the various activities are identified and conceptual models are generated and compared with the real situation. Extreme differences are then rectified in the conceptual model and the implementation is adjusted. The overall loop is then closed by implementing the best available plan, with the process being repeated until reasonable success is achieved. This approach allows us to analyse the problem in a holistic manner, working with the whole system rather than just parts, but it requires an exact definition of the performance Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 80 F. Bontempi et al. Figure 3. Top-down and bottom-up approach to problem solving. requirements and of the analysis, modelling and grade of complexity of the system. 2.3 Systemic approach to structural design and optimization Universally, demand for structural systems requires three dominant aspects to be optimized. These are generally described as the cost, time and performance factors (CPT). Attempting to optimize all three factors simultaneously is a very difficult task. However, the adoption of improved systemic processes seems to significantly improve all three at the same time. In fact, the objective of a systemic approach is that the system is designed, built and operated in the most cost-effective way possible. It means that a costeffective system must provide a particular kind of balance between effectiveness and cost; the system must provide the most effectiveness for the resources expended or, equivalently, it must be the least expensive for the effectiveness it provides. This condition is a weak one, because there are usually many designs that meet the constraints. Each possible design can be represented as a point in the trade-off space between effectiveness and cost. A graph, plotting the maximum achievable effectiveness of available design with current technology as a function of cost, would in general yield a curved line such as the one shown in figure 4. The curved line represents the envelope of the currently available technology in terms of cost-effectiveness. In addition, this curve shows the saturation effect that is Figure 4. Uncertainty in the cost-effective solutions (adapted from NASA (1995)). usually encountered as the highest levels of performances are approached. Points above the line cannot be achieved with currently available technology and they represent currently unachievable designs, although some of these points may be feasible in the future when further technological advances will be made. Points inside the envelope are feasible, but are dominated by designs whose combined cost and effectiveness lie on the envelope. Considering the starting point D0 for the design inside the envelope, there are alternatives that reduce costs without Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 Maintenance of complex structural systems decreasing any aspect of effectiveness (design point D1) or that increase some aspects of the effectiveness without decreasing others or without increasing costs (design point D2). For these reasons, the projects represented by the points on the envelope are called the cost-effective solutions. The process of finding the most cost-effective design is additionally complicated by the influence of uncertainty. The exact outcomes achieved by a particular system design cannot be known in advance with certainty, so the cost and the effectiveness of a design are better described by a probability distribution than by a point. With reference to figure 4, distributions resulting from a design which has little uncertainty are dense and highly compact, as is shown for concept A, while distributions associated with risky designs may have significant probabilities of producing highly undesirable outcomes, as is suggested by the presence of an additional low effectiveness/high cost cloud for concept C. Concept B represents an intermediate situation. It is interesting to consider aspects of the structural system cost as a function of the care of the design. In figure 5, the overall ownership cost curve Ct is represented as the sum of the design cost Cd, the maintenance cost Cm and the operation cost Co. If greater effort is put into obtaining a better and more robust design then, with an obvious increment of the design cost, it is usually the case that the maintenance cost will drop while the cost of operation will not change markedly. The process of minimizing the expected total cost is defined as life-cycle optimization (Frangopol et al. 2001). In order to apply the systemic approach to the design of the structural system it is necessary to take into account the design role within the phases of the system life-cycle. The 81 procedure starts with an interaction between the designer and the customer to establish the needs, which then become the key in the process; these needs are consequently used as references and driving statements for carrying out a requirements analysis. The next step is to break down the task until its constituents become clear. This is the requirement analysis step. A functional analysis with the allocation of functions and collection of elements is then developed. Consequently, potential solutions are synthesized, by the aggregation of various parts previously collected. If, after this step, the results are unsatisfactory, then the loop has to be repeated. Iterations regarding the requirements, the design and the validation may be required for several loops. This whole process is called the systems engineering process (SEP), or simply the egg (see figure 6). The whole structural design process can be reviewed within this systemic view, particularly considering the recent improvements in measurement and elaboration data technologies. These have created the proper conditions to advance the decisional tools through the enhancement of the information related both to the structural performances on site and to the measurement systems. The enormous advantages that can be drawn by the results of these measurements and the development of these technologies have led to the birth of activities inside the domain of the so-called performance-based structural engineering (Smith 2001, Silvestri 2002). In particular, the design of complex structures may draw benefits from this new systemic design philosophy based on the performances, with the possibility of obtaining perfection of the design solution in the context of the theory of excellence, a really comprehensive generalization of the re-engineering technique (Catallo et al. 2003). Figure 5. Costs associated with optimization (adapted from Sydenham (2004)). 82 F. Bontempi et al. The first five phases shown in figure 7 are traditionally considered by the structural design procedure and lead to the ‘as built’ construction. They are: Formulation of the problem; Synthesis of the solution; Analysis of the proposed solution; Evaluation of the solution performances; Construction. Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 1) 2) 3) 4) 5) Difficulties associated with this kind of approach are evident. The ‘as built’ structure could be different from the ‘as designed’ one due to different factors such as fabrication mistakes or unexpected conditions during the construction phase, or can also be due to inappropriate design assumptions. The latter could create doubts about the accuracy of the analysis results leading to a predicted behaviour that does not correspond to the real one. A way to avoid these difficulties can be achieved using a Figure 6. System engineering process (adapted from Sydenham (2004)). Figure 7. Extended scheme for the design process (adapted from Smith (2001)). Maintenance of complex structural systems monitoring system. In this case, three further steps must be added to the traditional design phases in figure 7: 6) 7) 8) Monitoring of the real construction; Comparison between results from monitoring and expected behaviour results; Increase in the accuracy of expectation of the future structural behaviour. These three more steps are the starting point of a truly systemic performance-based approach to structural engineering, leading to the further additions of figure 8: Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 9) Reformulation. Development of advanced probabilistic methods for a more accurate description of the required behaviour and of the required performance; 10) Weak Evaluation. This methodology assumes that the analysis is exact and that all the actions are known exactly from the probabilistic viewpoint; 11) Model improvement. The necessity connected to the models improvement comes from experiences based on monitoring, where expected and measured behaviours on ‘as built’ structures are directly compared; 12) Strong Evaluation. A third kind of evaluation, called strong, becomes possible when the improvement (see previous point) aims to assign more accurate values to the parameters used and to define a more accurate modelling hypothesis. 2.4 The concept of dependability and the need for monitoring For complex structural systems, or large scale projects where there are significant dependencies between elements 83 or sub-systems, it is important to have a solid knowledge both of how the system works as a whole, and also of how the elements behave singularly. The goal is to achieve established limits of the acceptable service and safety levels. Therefore, it becomes important to define a global concept that groups the attributes assumed to be relevant. Dependability can be defined as the collective term used to describe the availability performance and its influencing factors. Hence, dependability is a more comprehensive concept than reliability that, in the case of complex structural systems, may include some or all of the following attributes (Bentley 1993, Catallo 2005): . Availability, as the degree to which the system is operable; . Serviceability, as the quality of being able to provide good service; . Safety, as the absence of dangerous situations with possible catastrophic consequences on the users or on the environment; . Reliability, as the ability of the system to perform safety and serviceability during time; . Robustness, as the capability of not being damaged by accidental loads, e.g. fire, explosions, aeroplane impact or consequences of human error, to an extent disproportionate to the severity of the triggering event (see also Catallo 2005); . Maintainability, as the ease and the speed by which any activity related to the preservation and upgrade of the structure can be carried out on the system or on its parts. In a generic manner, one can presume that the occurrence of negative events is highly correlated with the perceived complexity of a system. This affirmation asks for a case to Figure 8. Further phases in PBD (adapted from Smith (2001)). 84 F. Bontempi et al. case evaluation, with regards to the type of structural system under investigation and the level of performance required. There are a number of reasons for dependability provision and assessment with regards to the maintenance of structural systems: Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 . Ever-increasing requirements from users in terms of quality; . High performance expectation from the final product; . Elevated interdependencies; . Elevated costs due to low quality parts or sub-systems; . Collateral administrative or financial issues (e.g. insurances). All these aspects can be summarized in one word: structural complexity. It is clear that the dependability provision is subjected to the function for which the structural system is commended and the dependability of the system can be addressed within certain ‘a priori’ imposed limits, often obtained by heuristic judgement. The threats for system dependability can be sub-divided into: . Failure; . Error; . Fault. It is important to distinguish the meaning of these terms. In the case of failure, the service that the system delivers deviates from compliance with the system specifications for a specified period of time. In the case of error, the system is in an incorrect state, it may or may not cause failure. On the other hand, a fault is a defect and represents a potential cause of error, active or dormant. The means for dependability assessment include (Avizienis et al. 2004): . Fault forecasting, concerning the estimation of the presence, the creation and the consequences of faults; . Fault prevention, aiming to prevent the occurrence or the introduction of faults; . Fault tolerance, as the ability of the system to continue normal operation despite the presence of faults; . Fault removal, aiming to reduce the number or the severity of faults. Beginning with the assessment of possible failures, local or global, in the structural system, it is possible to determine the effects of a failure in the global performances with a failure modes and effect analysis (FMEA), while a quantitative evaluation can be linked to these effects within a failure modes and effect critical analysis (FMECA). These are the bottom-up techniques. Event trees pictorially represent the logical order in which events in a system can occur (Stewart and Melchers 1997). They begin with an initiating event, and then the consequences of the event are followed through a series of possible paths. Each path is assigned a probability of occurrence. Therefore, the probability of the various possible outcomes can be calculated. Event tree analysis is based on binary logic, in which an event has either happened or not, or a component has failed or has not. It is valuable to analyse the consequences arising from a failure or undesired event. Fault trees look like a complement to event trees. The idea is to begin with a general conclusion (event) and, using a top-down approach, to generate a logic model that provides for both qualitative and quantitative evaluation of the system reliability. Failure tree analysis (FTA) is a top-down search method that consists of the construction of the failure trees of the events and the relative evaluation in terms of probability. In a FTA, a logical analysis takes place that consists of the construction of the trees and their transformation into Boolean expressions. It should be noted that FTA is a method for analysing causes of hazards to the system, not for identifying them. Therefore, the top event must have been foreseen and thus identified by means of other techniques. It is important to realize that regarding complexity, since it is necessary to set a synthetic view by graphical language, it is essential to abandon the pure equation-based description of the structure (Haimes 1998). Nowadays commercial software is available to implement system analysis as above. 2.5 Structural identification As previously stated, this structural identification process finds structural characteristics, models or actions through the measurement of the structural response. Traditionally, structural identification was developed in the field of mechanics and aeronautics, where experimental investigations can be conducted directly in the laboratory, mostly by scale models of the structure under inquiry. Recently, the implementation of structural identification techniques has experienced a huge growth. An important task in this process is the determination of the modal parameters. In fact, dynamic properties represent an intrinsic characteristic of the structure and therefore can represent its integrity more accurately. As a consequence, they can be useful during the testing of a new or rehabilitated structure or in the structural health monitoring of an existing one. Structural identification represents an appropriate way of relating experimental results to analytic ones. The identification of the mechanical properties of a structural system can be extremely useful in structural health monitoring or in diagnostic procedures. These allow Maintenance of complex structural systems global assessment and appraisal of the potential occurrence of damage phenomena. Structural identification is carried out in the presence of damage states, or during the service life of the structure. The way to deal with this problem is by using theoretical models, corrected on the basis of experimental results by model updating procedures. The aim of model updating is to minimize the difference between the calculated values of the theoretical model and the experimental measurements of the dynamic parameters of the structure. Structural damage may be identified with a side by side comparison of the structural model, both in intact and damaged conditions. In order to achieve an exact evaluation of the structural system performances, the structural identification, based on the monitoring data, assumes a role of primary importance. Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 3. Implementation of the structural monitoring process The maintenance process is firmly correlated to the structural monitoring process. Recent progress in the field of monitoring systems (the reduction of the price of hardware, the development of more and more accurate and reliable software, not to mention the decrease in size of the equipment) form the premise of developing, constructing, managing, recovering and dismantling the majority of civil engineering systems on the basis of data measurement. All these lead to an evolution of the concept of the structural monitoring process, which is no longer intended to be the passive exploration of deterioration or damaged states originating by observing the structure. On the contrary, the new approach is about the principle of controlling the structural system in a proactive way, in order to reveal the presence of any undesired fault and to understand the causes of failure. These days, structural (health) monitoring is an essential tool for the verification of the effective accomplishment of the expected performance and for updating the system reliability (Papadimitriou et al. 2001). Therefore, the monitoring process should be planned during the design phase. 3.1 Structural monitoring hierarchies It is important to note that the planning of monitoring processes is strictly related to the complexity of the structure. In fact, the cost, as a parameter, has to be taken into consideration. Three major implementations of the monitoring process can be defined: . Visual inspection; . Monitoring ‘by event’; . Continuous monitoring. Visual inspections and monitoring by event concern the main implementation for the overseeing of the structural 85 system, and they are organised as an integral part of the structural management program. On the other hand, continuous monitoring, as the name suggests, takes place with the continuous (with respect to time) reading of data, obtained from different instruments installed on the structure. It is essential for the assessment, in real or almost real time, of the integrity, reliability, durability and safety of the structural system. The process of monitoring by event differs from that of continuous monitoring due to the fact that it is carried out only for the time period strictly necessary for the completion of the requested assessment. The instrumentation is required from the first phases of the project in order to achieve a successful integration with the structure. In some cases, wherever there can be expected continuous variations of the global behaviour, or in the case of aggressive environments, the installation may be of a permanent type. Nevertheless, the present situation suggests that structures are still designed in a conventional way. In these cases, monitoring systems are employed merely on structural systems with innovative or unique characteristics, like complex infrastructural systems, long span bridges, dams, and generally, critical structures or systems subjected to significant accidental or environmental actions. In design, the implementation of a permanent control system is rare. This leads to the downgrading of the monitoring process to a secondary role that typically has as a primary scope the programming of visual inspections once degrading states or structural damages are revealed from local and/or specific measures in the course of an event monitoring process. Only in the event of excessive damage is a continuous monitoring system is scheduled, typically by means of temporarily installed equipment on the structure, as a consequence of the difficulty in identifying the origin of the damage itself. A simple classification of situations includes two categories: complex structural systems that can justify a continuous monitoring system as a consequence of long term cost reduction and serviceability, and ordinary structures, for which such a system, for the added cost, is not considered important. For the first class, structural health monitoring incorporates all the implementations previously mentioned. Parameters under observation are those subjected to the variations caused by environmental or accidental actions. Collected data is inserted into a database, in order to be elaborated and transmitted. The next step includes elaboration from skilled personnel, aiming for the assessment of the structural behaviour. In this way, numerical results can be integrated with normal visual inspections, a process that allows further intervention in case of malfunctions or structural damage. An integration with numerical models leads to the verification of the required performance, established during the design phase. 86 F. Bontempi et al. 3.2 Monitoring protracted in time Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 Another important aspect that should not be neglected is the role of the various implementations of monitoring (visual inspections, by event and continuous monitoring) for the different phases of the structure lifespan: design, construction, testing, service life, and dismissing, as shown in figures 9 and 10 (Senaud 2004). In these flow-charts, for every phase, two different paths can be noted; one is related to ordinary structures, and the other is related to complex structural systems. In this way it is possible to distinguish the main differences in the monitoring process, relating it to the complexity of the structures. 3.2.1 Monitoring during design. For the entire design phase, the location site of the structure is monitored, mostly implementing ‘by event’ techniques. Data collection aims to gather geotechnical information, such as the stratigraphy, the physical and mechanical characteristics of the soil and the underground water table. Furthermore, during this phase, the testing of the materials constituting the structure’s components and the scale models of the structure usually takes place. The design of the monitoring system advances with the design of the structure. Firstly, the parameters to be monitored are selected. Subsequently there are choices to be made regarding the sensors to be used, their quantity and location, the method of data transmission, the lodging of the central data collecting units, the location of the control and processing units, and finally, the method of an eventual data transfer via the internet to other remote locations for storage and further analysis. 3.2.2 Monitoring during construction. During the construction period, special care is to be taken for the provision of a temporary monitoring system, as a substitute for the main one, which is still in the preparation stage. This phase should be watched closely in order to intervene in case of errors. The geotechnical instrumentation, used in the previous design period, should be maintained in the proximity of the construction site, in order to check possible subsidence. The realization of the monitoring system proceeds side by side with the one of the construction and the positioning of the various instruments and sensors should take place as intended. Figure 9. Structural monitoring during design, construction and final inspection (Senaud 2004). 87 Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 Maintenance of complex structural systems Figure 10. Monitoring during service and dismissing (Senaud 2004). 3.2.3 Monitoring during testing. During the testing period, an important role is reserved both for visual inspections and for by event monitoring techniques. In particular, both destructive and non-destructive tests may take place, in order to assess the mechanical properties of the materials, the condition of the steel reinforcement, and other characteristics. Additional load tests may take place, in order to assess the global behaviour, the applied loads and the resulting stresses. Continuous monitoring takes place as soon as the sensors network is completed. The purpose is to record displacements and deformations during the load tests. This may be a further occasion to test the monitoring system as well, calibrating the sensors and verifying their accuracy. In the case where a permanent monitoring system is not planned, there should be provision for a temporary network of sensors to conduct the load tests. 3.2.4 Monitoring during service. During the service period, the approach is fairly different if one refers to a simple structure or to a complex structural system (see figure 10). In the first case, visual inspections should be carried out, integrated if necessary with by event monitoring. Continuous monitoring is employed only in special cases. Complex structural systems, on the other hand, necessitate further attention. In these cases, visual inspections and continuous monitoring are mutually integrated, and only in Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 88 F. Bontempi et al. the case of discordance or necessity does by event monitoring techniques take place. The monitoring plant starts to work at the beginning of the service period,. From that moment forward, data gathering is initiated. This process should go on until the end of the service period of the structural system, preferably without interruptions. The data collection may be continuous or at defined intervals. It is very important not to confuse the significance of continuous monitoring with that of the continuous gathering of data. In the first case one means that the structure is monitored for its whole lifespan and not only in the case of necessity. On the other hand, continuous data gathering means that the sensors measure continuously, time after time. A significant difference in terms of accumulated data exists between the continuous and the discontinuous data collection. Nevertheless, in both cases a central facility is necessary for the process of storage and elaboration of the gathered data. This aims to reduce and transform a large amount of data, difficult to manage data, into a quantity easier to interpret and exploit. The information obtained can be integrated with results from numerical analysis in order to compare the effective and the evaluated performances of the structure. In this way, it is possible to improve the accuracy of the models as well. 3.2.5 Monitoring during dismissing. Every time a structure is dismissed, the monitoring system has to be dismantled as well. The sensors should remain active for as long as possible, in order to monitor the effective maintenance of safety. This final period, in the case of complex structures, is of great importance, not only for safety reasons, but also for the appropriate recovery of the construction site and the eventual recycle of materials. 3.3 The individuality of the monitoring process The use of new methods based on the concepts of PBD and dependability, as well as the development of new techniques and technologies correlated to the monitoring process, allows the reliable assessment of structural behaviour. On this basis, and with careful management based on knowledge, further understanding of the design, construction and degradation process may be gained, allowing for a further increase of the safety level. This does not mean that all structures shall be equipped with the maximum technology available. In fact, the restriction in terms of cost is important. Therefore, it is important to create a monitoring process, specific for each structural system, and congruent with the available resources and the complexity of the system. According to these concepts, it is necessary to review the classification adopted in figures 9 and 10, in which the structures are divided into two large categories: simple structural systems and complex structural systems. In reality, considering the specificity of the monitoring process, this is an oversimplified classification that can be improved. A more refined classification includes: . . . . . . Ordinary structures; Selected structures; Special structures; Strategic structures; Active structures; Smart structures. The different classes are characterized by different levels of complexity in terms of information technology, as schematically shown in figure 11 by means of increasing size circles. The ordinary structures are those that do not need permanent monitoring equipment. For these, the usual monitoring consists of periodical visual inspections. The selected structures are provided with sensors for load tests in order to assess the compliance with the effective performances. The special structures require periodical planned controls through visual inspections and provisional instruments (monitoring by event). The data collected is useful for the calibration of the analytical model developed in the design phase, and provides information regarding the structural behaviour during the service life. The other three classes require continuous monitoring, which has to be planned from the design phase. The strategic structures require a monitoring system for assessing the structural behaviour over time. The active structures, as the name suggests, are provided with an active control of the structural behaviour that ensures the fulfilment of performances. The smart structures represent the maximum in terms of strategic management of complex structural systems. Their advantage consists of the possibility to modify the structural behaviour through feedback actions. In this sense, it is possible to apply structural field techniques from the artificial intelligence field, such as artificial neural networks or genetic algorithms (e.g. Sgambi (2005)). With this in mind, a specific scheme is shown in figure 12 (Senaud 2004). The five life phases of the charts in figures 9 and 10 are considered in a single chart, and it has been extruded to a third dimension, to take into account the complexity. In this way, the various planes represent different complexity levels (Z axis), while on the X and Y axes, the phases of the lifespan of the structure and the different implementations of the monitoring process are represented. The results obtained from different monitoring processes may be used in the organisation of the project phases and the maintenance activities, allowing for the complete assessment of the structural behaviour and the prevention Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 Maintenance of complex structural systems 89 Figure 11. Relationship between classification of structures and characteristics of the monitoring process: sensors, quantity of measurements, and feedback. of future deterioration or damage, thus reducing the risk of conducting onerous interventions on the structural system. 4. Structural health monitoring techniques A structural health monitoring system should be able to monitor the various loads to which the structure is subjected to (e.g. Gkoumas 2004), such as natural (ambient), anthropic and entropic loads, as well as the consequent structural response to these loads and the change in the state variables (as shown in figure 13 in the case of a long span bridge). Over a longer period of time, the entropic actions (decay of the materials) that influence the structural behaviour should be assessed. Most significant environmental loads involve temperature, wind, earthquake, rain, snow and, in the case of offshore structures, wave loads. In designing a structural monitoring system, it is convenient to differentiate ‘slow’ environmental loads (such as temperature or humidity) from those that vary rapidly (such as wind or earthquake). Anthropic loads, in the case of bridges for example, consist of pedestrian, highway or railway loading. In particular, railway loading can be considered deterministic while, in the case of highway loading, the knowledge of the representative spectrum of vehicle configurations and traffic patterns, together with a provision for the projected increased loads, is essential both for reliability and fatigue analysis. Structural response monitoring can be implemented generally with two often intertwined approaches: one considers the materials that comprise the structure, while the other considers the structure (or the sub-systems of the structure) as a whole. In the first case, the monitoring techniques are used to enquire about the local properties of the materials (concrete, steel, wood, composite materials) in order to evaluate their performance under different types of loads, temperature variations, ageing etc. In the second case, the structure is observed as a whole, with general techniques, from a global, geometric point of view. Dynamic characteristics of the structure may be used in order to determine if damage is present in the structure, to determine its approximate geometric location, to assess its extension, and eventually, to predict the remaining service life of the structure (Rytter 1993). The basic premise of vibration-based damage detection is that the damage will alter the stiffness, mass or energy dissipation properties of a system, which, in turn, will alter the measured dynamic response of the system (Farrar et al. 2001). Global methods, in contrast with local ones, do not Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 90 F. Bontempi et al. Figure 12. The individuality of the monitoring process (Senaud 2004). Figure 13. Principal monitoring issues for a long span bridge. require either direct access to the structural elements or the knowledge of their prior state. The choice of monitoring method relies, apart from the discretion of the project manager, on various parameters, typically the: Most importantly of all is a case by case evaluation that considers both the objectives to accomplish, and the relevancy of the structural system under investigation. Typology and complexity of the structure under inquiry; Age of the structure; Condition of the structure; Available resources. For the purposes of local structural health monitoring, there is the need for sensors to sample the local state of the materials of the structural elements. The parameters under investigation are related to the physical, chemical and . . . . 4.1 Local structural health monitoring techniques Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 Maintenance of complex structural systems tensional states of the materials. For strain or crack monitoring, it is necessary to adopt strain gauges or fibre optics. In the first case, the sensing takes place by measuring physical quantities, mostly by means of the variation of the electricity that passes through them. Fibre optics, on the other hand, represent the evolution of sensing. The basis of their operation is the total internal reflection of light rays travelling through tiny optical fibres. Fibre optic sensors can be classified as extrinsic or intrinsic ones. In an extrinsic sensor, the sensing takes place in a region outside the fibre. The fibre itself essentially serves as a conduit for the efficient transmission of light to the sensing region. In an intrinsic sensor, one or more of the physical properties of the fibre undergo a change, in this way leading to the desired evaluation. For structural health monitoring purposes, various types of sensors have been presented. Some of them are still in a research state, while others, such as those based on the Fabry-Perot interferometer, low coherence interferometry or fibre Bragg gratings, have been installed and tested in a variety of structures. The sensors may be embedded or may be externally mounted on the structural element. The advantages of implementing fibre optic sensors over classic strain gauges consists of the absence of electromagnetic interferences, the smaller size, the higher temperature capability, the greater multiplexing potential, and the demodulation to longer distances and the greater resistance to corrosion (Schultz et al. 1998). Most significant disadvantages include the higher cost and the sometimes limited accuracy. For the detection and localization of failures in high strength steel wire, strand or cable, acoustic, ultrasonic or magnetic resonance methods can be applied. Other than corrosion, high resistance steel, like the one used in prestressed concrete for reinforcement in post-tensioned structures or in supporting elements in cable-stayed or suspension bridges, is subjected to degradation due to the considerable tension applied, i.e. fatigue and hydrogen embrittlement phenomena. These actions are the predominant cause of ductility loss, with the potential consequence that the deficiency may be of a fragile type. The latest technology sensors, like those based on acoustic wave emission, allow for continuous, non-destructive monitoring, combined with the liable determination of the exact time and location of the failure. The registration and the evaluation of the acoustic events take place in real time, while a data processing unit is responsible for their automatic filtration, under prefixed criteria. For corrosion monitoring, embedded sensors are necessary. In recent years, several monitoring systems have been implemented in major infrastructure projects (Sørensen et al. 2004), based on sensors capable of deducing the corrosion risk of the steel reinforcement inside the concrete structural elements. These sensors indicate the depth of the 91 critical chloride content and the depth of the carbonation front initiating corrosion. Thus, the time to corrosion can be determined, enabling the owners of the structures to initiate preventive protection measures before cracks and spalling occur. The most common type of sensors is those embedded inside the fresh concrete during the construction. The basic measuring principle is to place electrodes at different depths related to the concrete surface, and to measure the onset of corrosion of these electrodes one by one. The measuring electrodes are made of steel, with a similar composition as the reinforcing steel, to ensure that they will start to corrode themselves at the same time as a bar situated at the same depth. Measurements take place manually, typically every six months or less. A similar sensor that can be inserted at a later stage is used to monitor corrosion on existing structures. Studies have shown that its performance is in accordance with that of the embedded sensors (Baessler et al. 2000). 4.2 Global structural health monitoring techniques For the purposes of global (or geometric) structural health monitoring, there is the necessity to obtain measurements of the acceleration, the velocity and the displacement of the structural elements, on a regular basis and with an elevated sampling rate. Topographic techniques used in the past (mostly during the testing period) provide highly accurate results, but are not always appropriate for an efficient implementation, fundamentally because of their low sampling level and their limitations in three-dimensional motion. Global monitoring techniques embrace vibrationbased structural damage detection, during which the alteration of the stiffness, the mass or the energy dissipation properties of the system (and consequently the alteration of its measured dynamic response) is expected. Detection methods, using changes in modal parameters to identify damage, can be implemented with two distinct approaches (Lauwagie et al. 2002). The first approach, which is called ‘the response-based approach’, compares modal parameters of the undamaged structure with the modal parameters obtained on the same structure in a damaged condition. The presence and severity of the damage can be assessed by evaluating the changes in natural frequencies and damping ratios. The second approach, ‘the model-based approach’, aims to find a set of model parameters of a mathematical model, in most cases a finite element (FE) model, of the considered structure, in order to have an optimal correlation between the experimentally measured and numerically calculated modal parameters. Damage can then be assessed by investigating the model parameters obtained. The premise of vibration-based structural health monitoring is very promising, even though sometimes it is difficult to accomplish since damage is typically a local phenomenon and may not significantly influence the Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 92 F. Bontempi et al. lower-frequency global response of structures that is normally measured during vibration tests (Farrar and Sohn 2000). Accelerometers are the reference instruments for global structural monitoring and particularly for vibration-based monitoring. They are available in a variety of ranges commercially. A first classification is between active and passive ones. An active accelerometer (e.g. a piezoelectric) gives an output without the need for an external power supply, while a passive accelerometer changes only its electric properties (e.g. capacitance) and requires an external electrical power. In applications, the choice of active or passive types of accelerometers is important, since active sensors cannot measure static or dc mode operations (Eren 1999). Accelerometers record the global behaviour of the structure even in the presence of cracks or fractures. Accelerometers based on MEMS (Microelectromechanical systems) have recently been developed. MEMS are devices and machines fabricated using techniques generally used in microelectronics, often to integrate mechanical or hydraulic functions, etc. with electrical functions. A typical MEMS device combines a sensor with logic to perform a monitoring function. One of the advantages of MEMS is that they can be produced in high quantities, achieving economy of scale. MEMS accelerometers can be separated into two categories: capacitive and piezoresistive ones. Even if both types use an internal seismic in order to perceive the acceleration, their difference lies in the mechanism used to correlate the movement of the mass to the acceleration. Capacitive accelerometers directly measure the acceleration, using balanced differential capacitors formed by capacitive plates that are etched into the perimeters of both the seismic mass and the substrate. On the other hand, traditional piezoresistive accelerometers are generally used for out-ofplane acceleration with a proof mass connected to a short flexure upon which a piezoresistive material has been implanted (Lynch et al. 2003). Capacitive MEMS accelerometers have experienced a superior commercial success with respect to piezoresistive ones, this is due to the significant further reduction in their manufacturing costs. In addition, some problems associated with the temperature influence on the piezoresistive materials led to the use of special shells or compensation circuits, with a further increase in cost. In spite of this, new researches aim to demonstrate the feasibility of cheap piezoresistive MEMS accelerometers. An accelerometer with a planar cantilevered proof mass was fabricated upon a silicon die using deep reactive ion etching (DRIE) techniques, in which the cantilever is a single short flexure that is slender in the in-plane direction limiting the out-of plane response of the proof mass (Partridge et al. 2000). Oblique ion implantation was used for the formation of piezoresistors upon the side walls of the cantilevering element where a region of highly focused strain existed. Over the full dynamic range, the accelerometer was found to exhibit nearly constant sensitivity, resulting in a linear transfer function of the sensor. During a structural monitoring validation test (Lynch et al. 2003), the frequencies of the calculated modes were found to be within 3% of those calculated by the theoretical model. The frequency response function from the high performance piezoresistive accelerometer was noisier than those calculated by the capacitive accelerometers. This was expected because of the dominance at low frequencies of a particular type of noise inherent to the accelerometer’s design. Global Positioning System (GPS) technology represents the ultimate variety of global structural monitoring techniques. The GPS is based on measuring the transit time of radio signals emitted by orbiting satellites. Displacements obtained with the GPS can be continuous, automatic and conducted under diverse weather conditions. The improvement of GPS technology in recent years, combined with the continuous diminution in price of GPS hardware and the automation of the structural monitoring process, makes this method a valid alternative, especially if considered in combination with other sensors (e.g. tiltmeters, accelerometers, etc.). A broad review of the challenges involving the implementation of GPS technology is provided by Rizos (2001). Positioning with GPS can be performed by either point positioning or differential positioning. The point positioning technique employs one GPS receiver that simultaneously tracks four or more GPS satellites in order to determine its own coordinates in a three-dimensional Cartesian coordinate system and an associated ellipsoid (WGS 84). The differential positioning technique consists of determining the position of one (rover) receiver (located at the point of interest), with respect to a fixed (reference) one. The latter technique permits a higher positioning accuracy, thus, it is the one used for structural monitoring purposes. Furthermore, differential positioning GPS can be applied in two distinctive modes: real time or post-processing. In real time differential GPS (RTDGPS), the compensation between stations (receivers) takes place during the measurements. A base station computes, formats and transmits corrections usually through some sort of data link (e.g. VHF radio or cellular telephone) with each new GPS observation. The roving unit requires some sort of data link receiving equipment to collect the transmitted GPS corrections and get them into the GPS receiver so they can be applied to its current observations. The accuracy observed in structural monitoring applications is of the order of 1 cm horizontally and 2 cm vertically. In post-processed differential GPS, the base and the roving receivers have no active data link between them. Instead, each one records the satellite observations that will allow differential correction at a later time. Differential Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 Maintenance of complex structural systems correction software is used to combine and process the data collected from these receivers. This technique is adequate for the monitoring of structures that experience slow displacements and require an elevated accuracy (2 to 4 cm horizontally and 6 to 8 cm vertically). The principal problem in using GPS monitoring techniques for structural monitoring comes from the presence of errors in the measurements. Principal causes include phase ambiguity (the phase measurements delivered are biased by the same phase), multipath (interference caused by reflected GPS signals arriving at the receiver as a result of nearby structures or other reflective surfaces) and, sometimes, troposphere delay (the propagation delay of the GPS signal caused by the non-ionized or neutral atmosphere). There have been various attempts to mitigate these errors, with more or less success. For this reason, their integration with other monitoring devices such as accelerometers has been suggested (Roberts et al. 2000). For very complex structural systems, such as long span cable stayed or suspension bridges, the trend is to implement GPS technology as a part of a broad monitoring system, consisting of further sub-systems, each with an elevated number of sensors (Wong et al. 2001). This is the way to achieve continuity and redundancy in the measurements and, at the same time, to guarantee the maximum dependability of the structural system. 5. Conclusions The development of new approaches in structural engineering based on performance-based design, system engineering and structural health monitoring, combined with the use of new technologies and new software allows the behaviour of complex structural systems to be correctly assessed with a lower degree of uncertainty. In this study, it has been shown that the design of a complex structural system involves several general aspects, and therefore, their consideration and optimization need to be organised and synthesized by implementing a performance approach, which is opposite to the prescriptive approach. The whole process should be organised within a system engineering ‘modus operandi’ in an organic and hierarchic way, in order to guarantee established dependability levels. Furthermore, the principal phases of the structural health monitoring process are exposed in concise statements, with regard to the maintenance of complex structural systems. It has been shown that for correct implementation of the monitoring process, the monitoring process should be contingent to the structure’s complexity and should be suitably planned for every phase of the structure’s life. The use of latest generation GPS and fibre optic technologies, integrated with additional sensors installed on the structural system, allows a more 93 precise and reliable knowledge if the systemic approach is adopted. Acknowledgements This study is closely based on different theses, submitted in the academic year 2004/2005 at the Department of Structural and Geotechnical Engineering of the Faculty of Engineering, University of Rome ‘La Sapienza’, by S. Arangio, K. Gkoumas, G. Senaud, L. Catallo and L. Sgambi, with F. Bontempi as advisor. Furthermore, the financial support of the University of Rome ‘La Sapienza’, COFIN 2002/2004 and Stretto di Messina S.p.A. is acknowledged. Nevertheless, the opinions and the results presented here are the responsibility of the authors and cannot be assumed to reflect the ones of the University of Rome ‘La Sapienza’ or Stretto di Messina S.p.A. Finally, the considerations of the anonymous reviewers are appreciated. References Arangio, S., Risoluzione di problemi strutturali diretti ed inversi mediante tecniche euristiche. Graduate thesis (in Italian), Department of Structural and Geotechnical Engineering, School of Engineering, University of Rome ‘La Sapienza’, 2004. Avizienis, I., Laprie, J.C. and Randell, B., Dependability and its threats: a taxonomy, in 18th IFIP World Computer Congress, Toulouse (France), Building the Information Society, 2004, 12, pp. 91 – 120 (Kluwer Academic Publishers). Baessler, R., Mietz, J., Raupach, M. and Klinghoffer, O., Corrosion monitoring sensors for durability assessment of concrete structures. In Proceedings of SPIE, edited by S.C. Liu, 3988, pp. 32 – 39, 2000 Smart Structures and Materials 2000, Smart Systems for Bridges, Structures, and Highways (SPIE Press: Bellingham, WA). Bentley, J.P., An Introduction to Reliability and Quality Engineering, 1993 (Longman: Essex). Blanchard, B.S., System Engineering Management, 1997 (WileyInterscience: New York). Catallo, L., Progettazione prestazionale nei sistemi strutturali complessi: fidatezza strutturale. PhD Thesis (in Italian), Department of Structural and Geotechnical Engineering, School of Engineering, University of Rome ‘La Sapienza’, 2005. Catallo, L., Sgambi, L. and Silvestri, M., General aspects of the structural behaviour in the Messina Strait Bridge design. In Proceedings of the Second International Conference on Structural and Construction Engineering, edited by F. Bontempi, pp. 2487 – 2494, 2003 (A.A. Balkema: The Netherlands). Eren, H., Acceleration, Vibration and Shock: the Measurements, Instrumentation and Sensors Handbook, edited by J.G. Webster, 17, pp. 1 – 32, 1999 (CRC Press: New York). Farrar, C.R., Doebling, S.W. and Nix, D.A., Vibration-based structural damage identification. Phil. Trans. R. Soc.: Math., Phys. Eng. Sci., 2001, 359(1778), 131 – 149. Farrar C.R. and Sohn, H., Pattern recognition for structural health monitoring, in Workshop on Mitigation of Earthquake Disaster by Advanced Technologies, Las Vegas, NV, USA, 2000. Frangopol, D.M., Kong, J.S. and Gharaibeh, E.S., Reliability-based lifecycle management of highway bridges. J. Comput. Civ. Eng., 2001, 15(1), 27 – 34. Downloaded By: [Arangio, Stefania] At: 14:48 23 May 2010 94 F. Bontempi et al. Gkoumas, K., Metodologie per il monitoraggio di sistemi strutturali complessi con particolare riferimento al caso dei ponti. Graduate thesis (in Italian), Department of Structural and Geotechnical Engineering, School of Engineering, University of Rome ‘La Sapienza’, 2004. Haimes, Y.Y., Risk Modelling Assessment and Management, 1998 (Wiley: New York). Lauwagie, T., Sol, H. and Dascotte, E., Damage identification in beams using inverse methods. In Proceedings of the International Conference on Noise and Vibration Engineering, edited by P. Sas and B. van Hal, 2002 (on CD-ROM). Lynch, P.J., Partridge, A., Law, H.K., Kenny, W.T., Kiremidjian, S.A. and Karryer, E., Design of piezoresistive MEMS-based accelerometer for integration with wireless sensing unit for structural monitoring. J. Aero. Eng., 2003, 16(3), 108 – 114. National Aeronautics and Space Administration (NASA), Systems Engineering Handbook, 1995. Available online at: www.nasa.gov (accessed 15 May 2006). Papadimitriou, C., Beck, J.L. and Katafygiotis, L.S., Updating robust reliability using structural test data. Prob. Eng. Mech., 2001, 16, 103 – 113. Partridge, A., Reynolds, J.K., Chui, B.W., Chow, E., Fitzgerald, A.M., Zhang, L., Maluf, N.I. and Kenny, T.W., A high-performance planar piezoresistive accelerometer. IEEE J. Microelectro. Mech. Syst., 2000, 9(1), 58 – 65. Rizos, C., Precise GPS positioning: prospects and challenges, in Proceedings of 5th International Symposium on Satellite Navigation Technology & Applications, 2001, paper 37 (on CD-ROM). Roberts, G.W., Dodson, A.H., Brown, C.J., Karuna, R. and Evans, E., Monitoring the height deflections of the Humber Bridge by GPS, GLONASS and finite element modelling. In Proceedings of IAG Symposia 2000, edited by Schwarz, 121, pp. 355 – 360, 2000 (SpringerVerlag: Berlin). Rytter, A., Vibration based inspection of civil engineering structures. PhD thesis, Department of Building Technology and Structural Engineering, University of Aalborg, Denmark, 1993. View publication stats Schultz L.W., Udd E., Seim M.J. and McGill E.G., Advanced fiber grating strain sensor systems for bridges, structures and highways, in Proceedings of SPIE, 3325, pp. 212 – 221, 1998, Smart Structures and Materials 1998: Smart Systems for Bridges, Structures, and Highways. Senaud, G., Strategie per la qualificazione prestazionale ed il monitoraggio strutturale. Graduate thesis (in Italian), Department of Structural and Geotechnical Engineering, School of Engineering, University of Rome ‘La Sapienza’, Italy, 2004. Schodek, D.L., Structures, third edition, 1998 (Prentice Hall: Columbus, Ohio). Sgambi, L., I metodi dell’intelligenza artificiale nell’analisi e nella progettazione dei ponti sospesi. PhD thesis (in Italian), Department of Structural and Geotechnical Engineering, School of Engineering, University of Rome ‘La Sapienza’, 2005. Silvestri, M., Gli aspetti concettuali del performance-based design e la loro applicazione nel progetto dei ponti sospesi. Graduate thesis (in Italian), Department of Structural and Geotechnical Engineering, School of Engineering, University of Rome ‘La Sapienza’, 2002. Skelton, R.E., Structural system: a marriage of structural engineering and system science. J. Struct. Control, 2002, 9, 113 – 133. Smith, I., Increasing knowledge of structural performance. Struct. Eng. Int., 2001, 12(3), 191 – 195. Sørensen, R.E., Buhr, B. and Frølund, T., Sensoring corrosion – the Danish way. In Proceedings of the 2nd International Conference of the International Association for Bridge Maintenance and Safety (IABMAS’04), edited by E. Watanabe, D.M. Frangopol and T. Utsunomiya, 2004. Stewart, M.G. and Melchers, R.E., Probabilistic Risk Assessment of Engineering systems, 1997 (Chapman and Hall: London). Sydenham, P.H., Systems Approach to Engineering Design, 2004 (Artech house: Boston). Wong, K., Man, K. and Chan, W., Real-time kinematic spans the gap. GPS World, 1 July 2001. Available online at: www.gpsworld.com (accessed 15 May 2006).