 So I'm going to talk about a bit of a different type of structural material which is a graph revetment So But well, the main principles are the same. So that's a hopefully that will come out clearly So very briefly about my PhD in general so what I'm looking at is how can we manage the reliability of flood defenses throughout the life cycle and I'm mainly looking at how do We deal with uncertainty Development of uncertainty. How can we use structural health monitoring and inspections to well keep that in in check how do there how does the change the the strength change and the load change in time and how does that influence the reliability and in terms of heuristics like how can we use the principles of asset management to define Strategies so There's this principle of you. Well, huh? It's Quite common. I think it's pretty much common sense. Hey, you plan do plan something you do it You check did it work and then if it worked you probably will do it again the next time and if it didn't work You'll change something so that's what I'm looking at So and then there's this challenge while we have all kinds of different measures so we can do site investigations to better characterize the dike we can improve them by adding more soil or by adding structures or relieve wells to relieve the pressure of the water Or reduce waves by putting vegetation in front of it. Well, there's all kinds of different things. They all have their respective uncertainties and so my main Topic is how can we get a good strategy from that? so but zooming in a bit for this this short presentation is So what one of the things I want to look at is it wasn't an intro slide But I skip that fairly quickly is can we derive a probability of detection for visual inspection of grass revetments so To give you an idea. So this is a well a nice picture of one of the Delta works But as you can see even though this is a coastal defense. So in coastal defense as you typically covered the dyke body with rocks blocks as felt but also a Fair part of it is mostly covered by grass. So it would be typically something like this So you have a heart revetment and then a grass revetment and that grass revetment also has to have some kind of erosion resistance Or it might feel because waves run up erode this part and eventually things go Go completely wrong or it can overtop and Erode from the backside. So that's the idea. So that grass revetment has to be kept at a certain state How do they do that currently This is so there's they made this sort of inventory of damage mechanisms, which you see here Which is a well quick count probably about 17 or 18 different things that inspectors have to inspect and Practically all the inspections are done visually. So they have to rate defects and rate for all these 17 different mechanisms, which is quite a lot and most likely not not all of them are Critical so this is an example as so you have one of the main issues is that you can have burrowing animals and If it's good, you don't have them, but if it's poor you have like all these Holes in this case is from mice mice are not that bad But you can also have beavers and they can dig away like a cubic meter a day. So that's pretty Pretty bad So there is a way to estimate the erosion properties of the of the grass salt, but it's Not that widely used and most mostly it said well, it looks good. That's That's the highest quality And it's also it's slightly destructive because you in fact you cut the roots You put it put it out and then you pull it until it breaks and the point where it breaks determines the class of the quality of the salt So it's slightly destructive. Yeah, I would say so as I said preferably you do it in another structure way and that would then pull down to doing a visual inspection, but that visual inspection as Daniel said yesterday is often regarded as being In a very this deterministic way. We see there's some burrowing of animals So this revetment is very bad or so this but there's And if someone says it's good, and that's probably the The most difficult part They It might actually not be that good in terms of erosion properties, especially if you have not done this test so my idea and I'm Well, I'm developing the idea. So it's to focus on how reliable is that that's How reliable is our estimate of the salt quality based on a visual inspection? so the other things I also spend some time on but now focus on this and My idea is to set up an experiment with and inspectors probably about 30 To well see how close they get to the Eventual tests that we'll do other at a test site So the thing I wanted to bring forward as a kind of a discussion topic was how should we should I approach this? What constraints should I apply? Well, maybe you can share some experience of out of your work on That I can use to further develop this idea So maybe it's a bit of an open question, but On this part no no not yet so so so one of the things I did I recently do an analysis to show like how important that salt quality or that that grass quality is for the failure probability So once it's this kind of three categories, so it's called closed open and fragmented And once it gets fragmented your failure probability increases by a factor thousand because it essentially loses all its All its erosion resistance because well the parts are separated. So that's the situation you have to avoid at all costs Like Yeah, well, so there's this there's this failure model that models the erosion of the grass cover and the clay layer underneath and After that happens typically especially at coastal areas because the waves are so big your Yeah, your flood defense will be gone within a few hours. So that's that's Well, once that happens, it's a couple of billions typically You What it yeah Well One one of the remarks is I'm not sure if this is the ground truth, but this is as close as we get. That's the that's That's kind of the thing as so so what they do is they pull it and if they pull it well in one one pool it's it falls apart it was apparently fragmented and Fragmented is related to an erosion resistance value And that we can use in a model and then if you have to pull a bit more It's gonna be open and if you can't pull it apart. It's closed. That's sort of the Relationship Yes Some value that you use yeah Yeah, that's me. Yeah It's a Yeah, yeah, yeah Yeah, it's a correct indication or something like that Yeah One is to set up Yeah Yeah, and then I can Yeah, yeah, that's usually also an approach. I like because you can say well apparently You have to have a 80% correct rate to be okay, and Then I can and that's how you do the inspector and say can you meet this? But yeah, yeah, that's a good good point Yeah Yeah, that's a good point there has been a test that spatial variability is always a killer in flood defenses, but There has been an experiment for revetments where they had like a few Tests in a in a couple of square meters and that was quite consistent, but Typically what you get is when you look at this Thing this is typically an anomaly problem, so you have a Sots layer and there's certain points that are damaged, so There's two so so that's also There's a piece of anomaly detection in the inspection and there's a part where you estimate like the general properties So that's also a yeah