 I'm going to talk today about Frameworks for Structural Reliability Assessment and Risk Management in Incorporating Structural Health Monitoring Data, and a few other things which I will explain as I go. I will be metaphorically eating into the time we have for lunch, so I will try to be quick. The presentation is qualitative. It's a series of perspectives on SHM and its value. There will be no equations. All right, so again, okay, all right, but nevertheless, all right, it makes sense to start with a definition of what SHM is. There are many such definitions. We like to work within one that was given by Eminaktan and his colleagues more than 10 years ago, and they said that SHM is basically about tracking in situ. Keywords are in bold font, tracking in situ structural performance or health by measuring data and interpreting them using application-specific knowledge so that aspects such as structural performance, condition, and reliability can be quantified objectively. It's a good definition. It's suitably broad. It seeks for a deeper knowledge, which, an information that can be extracted from SHM in terms of condition, reliability, and its quantification. It is useful because it doesn't say that SHM is equal to damage detection. A lot of real-life problems are really concerned with aspects of performance or perhaps serviceability in mid-state. The structure doesn't have to be necessarily damaged, and well, damage detection is difficult. I think we can reliably perhaps at that stage assess performance. Anyway, series of slides, series of ellipses, we try to put SHM in a broader context. In its focused view, SHM is about sensing and data processing. Of course, that's not the end of what we want to do with SHM. We want to create some information and knowledge about structural performance condition and so on, so forth, so to realize the value of that information. But again, a broader context of the view is, and we are here to think how value can be actually extracted. But again, this value will only be a theoretical concept if it's not linked to an even broader look at the SHM, and this is how we can actually realize this value in the whole asset management decision-making process. Another view of SHM in that broader context, this time linear one, is in the slide. We start with monitoring of, say, a wind turbine blade with some sensors, or we look at a K-joint in a truss structure, perhaps a bridge or similar structure. Of course, any component of a structure like that is part of a whole structural system, a wind turbine or a bridge, for example. We are quite comfortable in moving between the two scales, looking at some hotspots within a structure and looking how particular areas which we monitor function as a part of the system, where the links are not that well developed and the scale is, and we don't scale up probably that effectively as we should, is going to a bigger scale. This wind turbine will obviously be a part of a whole farm with many, many similar structures. A bridge like that will be part of a transportation system serving particular city or a particular region. Now, from the point of view of how any health monitoring data will be used, it will be used by owners, operators, stewards of infrastructure systems in decision-making process, and they will very likely think at that level rather than here. So, those links need to be strengthened and develop an SHM to realize its latent potential needs to be clearly anchored in those broader contexts. Otherwise, it's just very up in the air, not being properly interfaced with objectives of those stakeholders. A couple of slides which summarize areas where structural health monitoring can make a difference. Typically, for new structures, innovative design materials and so on where we construction techniques, where we don't have enough knowledge about how they really function in real environment. Structures and assets with poorly understood risks, and there's a whole spectrum of those risks, geological, seismic, meteorological, environmental quality assurance, and so on and so forth. New or existing structures, so-called indicative structures, you may say, which are representative of a larger population of similar structures, where hopefully information can be extrapolated to a wider population. That's more easy for, say, mechanical system, perhaps wind turbines, for hardcore civil infrastructure, like bridges, as a challenge. Well, each structure is quite unique, but nevertheless some mileage perhaps can be developed here, depending on which sector you are talking about. New or existing structures that are crucial or critical, sorry, at the system network level, where failure or deficiency would have a serious impact on the system network functioning. Concept of critical infrastructure was covered in the second presentation. Then existing structures, where we know there are deficiencies already and problems, they have low rating and so on. However, we want to continue their operation, extend their useful life. For example, many offshore facilities in the North Sea are already reaching their design life, but, well, it would be very useful and important to extend this life. And finally, candidates for replacement or refurbishment, where we are able to, using data to assess the real need for intervention and consider SS efficiency of repairs after we have conducted them. The OSHM can make impacts in several areas by reducing uncertainty about structural condition and performance. Once you start monitoring a real structure, most of them are designed with considerable conservatism, so you will find that it has quite a lot of hidden structural reserve, which is good news. On the other hand, but of course we don't know everything about structures, we will often discover, well, perhaps not that often, but from time to time discover deficiencies that may be missed by traditional assessment techniques. In both cases, you are winning because you know better about how your structure really behaves rather than what you assume and can manage the risks in a more efficient way, cheaper, but at the same time keeping some risk under check. Increasing safety and reliability, ensuring long-term quality of aging infrastructure, we hope to achieve better informed asset management of stocks of structures. And generally speaking from the academic point of view, increased knowledge about institutional structural performance, which when later trickled down to, say, code calibrations and recommendations for design. So these are the expectations. In some cases, there are already some results, but there are challenges. This figure is from Franklin Moon of Drexel University. It captures quite well the history, perhaps in a harsh way, but captures quite well the history of structural health monitoring applications so far. Time is here on the horizontal axis and benefit to owner or steward, the photo of the gentleman in my second slide is here on the vertical axis. The red line is the real capacity of what the structural health monitoring system can achieve. And as time passes due to research, due to experience with practical applications, yes, the real objective benefits are better and better. However, what is important here, and, well, Franklin uses words like bubble, snake oil effect and so on, when there's a lot of buzz in the past about what SHM can deliver, leading to overly optimistic expectations. Industry subsequently realized that those benefits were really not there. It's a challenge to extract benefits from health monitoring. That led to backlash here. Overly pessimistic assessment much lower than the real capabilities of SHM systems. So the task, hard task for us is to really bring the expectations to close the gap between expectations or perceptions of our SHM to reality. That's, of course, a considerable change. I think as this network sets up for a journey to quantify value of information, it is very important that we are as realistic as possible in our attempts to avoid those potential pitfalls and learn from the past. So there is a need for realistic assessment of SHM capabilities. In other words, you can say that it's about the realistic assessment of the value of SHM. And then follow up is the strategic plan deployment that is closely integrated into asset management and emergency response processes. I'm talking in many of the sites, but asset management slash emergency response to acknowledge that if you deal with bridges, you are operating with two timescales. One is slow related to aging, deterioration, corrosion, fatigue and so on. You make decision in the timescale of years or decades. The other reality depends if you are a right person at the right time. So to speak, and you have to deal with, with emerging situations such as earthquake or perhaps flood or similar natural disaster, when your timescale is significantly shortened and compressed. But generally speaking, those assess those at the level of what I'm presenting here, asset management and emergency response processes and integration of SHM proceeds along the same, along the same line. All right, so this is the outline of the rest of my presentation. I have done already the introduction. I will now talk about two frameworks for such strategic integration of my, of my strategic integration of SHM versus based on prioritization of structures for SHM. The second one sees the SHM as a, in a value chain of technologies. I will explain briefly the concept later. And a couple of additional remarks. Seeing SHM as big data is an emerging and very likely important, growing in importance concept for the years to come. And I will wrap up with a simple example of monitoring of a major bridge. I'm a structural engineer. Most of my work in SHM has been in the, in, in, in bridge area. So it tins my presentation and tins it towards the bridge applications. But I will try to keep the discussions general so that they are applicable to others, other structures, other systems as well. So first framework is about strategic strategies for integration of SHM into asset management and or, or emergency response and prioritization of structures for SHM, which follows in a natural way. The building blocks in that, which we proposed for that strategic integration, it starts with prioritization of bridges for application of SHM based on bridge importance in the network and a broad spectrum of risks. It is, of course, well, too expensive. It's, it's impractical. It's even not necessary to instrument each and every structure. And it's even not necessary to probably instrument majority of structures. So you have to be strategic, given that you, but you always have limited resources and try to implement the, the, the monitoring techniques on those structures or somehow choose them. The concept of risk is a useful tool to, to make those decisions. Then you can optimize the resources. And then, then the value of, which you can extract from SHM will be higher. Parts two and three are the core activities within SHM. Guidelines for instrumentation to be installed on bridge structures and their vicinity. It's quite often not only the structure, but its foundation. It's the soil. It can be the whole network where you will have to know the, the, the traffic patterns in your network or whole hydrological system. If this is a bridge which you, which has a scour risk, right? Most effective hardware platforms are necessary here. Relatively simple measures, I think, which can help in assessment. From the point of view of quantitative, using quantitative tools for assessing the value of this information. I think we definitely need a better quantify, better quantification statistical sense of all the methods performance. For example, what is the minimum damage size, detectability, and so on, so more important because, well, it's easy to do it on the computer. It works not that well in the lab for real structures. Most of the method, I think it's fair to say that they will have considerable uncertainty in terms of, in terms of, in terms of outline. And we have to be realistic about what, what, what the methods we have under our belt can really deliver. Then, of course, methodologies for reliable condition, damage, and performance assessment are necessary. But, so this is the SHM as such. Beyond that, there is strong need for integration of those SHM assisted assessment into broad asset management, and necessary, and emergency planning. So that it's, it's well integrated and functions within the practices and policies of organizations responsible for functionality and transportation. Functionality of transportation system, if we are to think about bridges. Otherwise, there will be again no, no real benefit beyond academic curiosity. We tried to come up with a simple, simple way of prioritizing those bridges, in a, bridges in a network based on bridge risk to the functioning of the transportation system. We understood, risk understood here in the classical sense, probability of failure times, times consequences. But we, we added another dimension, and this is bridge criticality, right? Some bridges have such, such large consequences of failure that it, they, they should really, really stand up. The, the concept of criticality was again mentioned by, was explored in the second presentation. That enabled us to specify three different data collection and associated SHM application, bands, core, intermediate, and advanced. For core data collection, this is for low risk and criticality bridges. Visual inspections don't have to be, to be very, very often, conducted very often, and SHM will probably not be used, or maybe sparingly. Intermediate visual inspection following two, three year typical cycle, supplemented by some SHM. Advanced data collection, where bridge criticality is high. Visual inspection will be more frequent, have some individual timeline for them, and they will need to be supplemented by advanced SHM. So that provides a simple way of deciding which bridge I really want to, or bridge asset I really want to consider SHM, and where it won't be necessary, and will probably be just a waste of money. The second framework of seeing, or way of seeing SHM in a broader context is SHM in a value chain of technologies. This is a concept which goes back to a paper by Wong and Yao in computer-aided civil and infrastructure engineering nearly 15 years ago. What they, what they proposed is seeing SHM as follows. SHM is here. It starts with monitoring data collect, and data collection using sensing methodologies, signal processing, using appropriate tools, and finally data analysis. For example, for data damage detection or some other forms of statistical processing of the data. But so what, right? What can we do with those results of the analysis? The next logical step is reliability, safety, and risk assessment, right? Using appropriate science and art, because it's quite often art as much as science. Once we are, we, we, we completed this task. We have, we can start to, to manage our risks. We need some decision making to us. And only with that, this is the real value to the infrastructure stakeholders, right? They want, they want, or their task is to, or desire and, well, and the obligation is to deliver safe, reliable, and efficient infrastructure at a minimum cost, right? They don't care what is happening here. They, they, they care at the other end. But so far, the information is not flowing sufficiently through this, through this value chain, where a continuous link, or you may say gap between SHM and the, the, the next in the chain of the enabling technologies. And this link needs to be developed if we want to really, really deliver the value of SHM to the, to the stakeholders, and clearly articulate it. This will enable us to clearly articulate that, that value of SHM. And it seems to be a condition for white adoption of SHM. Another, another issue which I got interested in recently is the big data perspective on SHM, right? If you start monitoring any, any structure, you will immediately be overwhelmed with, with data. Here is an example. It might be an extreme example, but nevertheless, it's a real example. Stone Cutter Bridge in Hong Kong from paper by Nian, Nian, Nian Wong. This bridge has 15, more than 1500 sensors, accelerometers, temperatures, strain gauges, versus WIM system, and many, many other corrosion, digital cameras, and so on, so forth, right? So as you can see, this is a lot of sensors and a lot of data being collected. Is SHM data big data? Well, big data is characterized typically by free extremes or extraordinary qualities. Volume, velocity, variety, and uncertainty. Volume, as you can see, for large bridges, volume and velocity, for large bridges, there is in the order of thousands of sensors and so on, so I think it does put pressure on, perhaps not so much on storage. Storage becomes cheaper and cheaper every day these days, but while transfer, if you have wireless networks, which are becoming more popular, it's a huge, huge bottleneck. But more importantly, timely and efficient interrogation puts quite a lot of pressure on available analytical techniques. So the question is, how much data to collect? Again, if you manage to assess, before you start doing any monitoring, value of data that will be a very, very useful concept because there is a tendency that let's have more sensors and so on. Is it really necessary and so on, right? Yeah, OK. Variety and veracity, lots of things can be said. Comprehensive SHM systems will have data sampled at different intervals, different sensors, different accuracy, diverse technologies. It needs to be merged with other forms of data from visual inspections. They will always remain for bridges, for us, for a visible future. SHM will not replace them as such. It can help to create additional information. And other forms of data, such as data stored as drawings or descriptive or qualitative report, we also need to think about how all the data mining techniques can be integrated with expertise and judgment by engineers. So I can synergize between different data. Digital twin, which is a concept which is becoming quite popular in aerospace industry, it's effectively and will be enabled by big data from SHM monitoring. Digital twin integrates high fidelity, multi-physics and multi-scale models and simulations with SHM data, maintenance history, and all available historical data. It enables to create much more realistic digital twins with what we are doing now, the presence of a lot of or ample data from monitoring. And new and enhanced levels of safety and reliability with over-designing of structures seem to be more and more realistic for us. Okay, and finally, I will probably skip this part. Some of you have seen it perhaps more than once, I'm sure, about monitoring of a bridge with it. Just briefly, we are trying to, we have a system which has around 90 channels of data. We try to create, effectively, a digital twin as much as we can by calibrating both models for both dynamic, short-term response, as well as long-term response related to CRIP and shrinkage and CRIP data to have both calibrated time-dependent structural models for reliability simulations and probabilistic models of actual responses, loads and other effects. All right, I will skip some numerical data. Summary conclusion, clear link to reliability and assessment of structure for an overall asset management necessary for SHM to realize its latent potential to frameworks. I briefly touched on big data, presents emerging challenges but also opportunities and the example of a bridge which I didn't have time to show much. This is based partly on my research but quite a lot of it is just literature survey with some, I tried to digest this so I would like to acknowledge the many resources I used, more of them are in the paper. And finally, I would like to thank my supporters as well as the many collaborators which devoted themselves to various aspects of this research. Thank you very much. Thank you.