 Hi everyone, as professor said I'm working at the faculty of civil engineering in Zagreb and I'll present our case study which is partly has been presented on some of the previous meetings because it all started with STSM back in 2016 and it continue to a new STSM which I've been in February. So just have a short introduction for you who haven't seen it already. So my presentation is named application of bridge weight motion measurements in assessment of existing bridges. I divided the presentation as most fathers did so here I will just give a short introduction about the bridge and decision scenario based on assessment strategy then short introduction about monitoring method applied and then data and results which we obtain and in the fourth chapter the value of information and associated costs. So for start this is the bridge the bridge is located in Slovenia it's a highway bridge small one only one single span of 20 almost 25 meters cross section of the bridge is classic pre-stressed high type girder connected with monolithic depth. We had all the original designs and drawings and we did a visual inspection back in 2016 and we built a 3D model in Sophistic software for assessment of the bridge. Bridge is 26 years old so it's not all the bridge but we use this one because we had all the data and monitoring data. So main objective of this case study is to prove that initial investment in monitoring in this case weight in motion will result in extended bridge service life reduction of overall maintenance costs and overall optimization of bridge management process. So back in 2015 I think we started working on multi-level assessment method for existing bridges in Croatia because a lot of them smaller bridges are built in the 60s and 70s and they are very bad state and we started using for the first time weight in motion measurements so we developed like three-level assessment method where we have first one without monitoring data then with short-term monitoring and with low-term monitoring. So the point is if bridge passes the assessment on the first level then you don't have to do the next levels because every each level is much more complicated and of course it costs more. So in the first step just basic numerical model based on visual inspection and with traffic loads based on design code for new bridges as there are no assessment codes or criteria in Croatia or also in Europe. Bridge rating is based on deterministic or probabilistic approach it's based on the detail of the analysis and if the bridge doesn't pass assessment on this level then we can use the strategy with short-term monitoring which is done with bridge weight in motion for a short period of time and if it doesn't pass then we can take a long-term monitoring data but it takes a lot longer time to be collected. So about the bridge weight in motion for those of you who are not familiar it's a method that measures every vehicle that passes the bridge. It uses bridges as weighing scales so it has advantages over stationary systems because stationary web systems are built in the pavement so you need to close the bridge you need to dig in the pavement and they are permanent this one is fully portable so you can apply it on the bridge measure it then move it to another bridge and it plays under the bridge so there is no interruption in traffic which is very important especially if you are talking about highway bridges and also it has higher accuracy and it besides traffic data it provides structural information. Information about how the bridge is responding to traffic laws which we use you will see in this first step. Disadvantages is that it requires a certain level of knowledge about the bridges so you need an experienced bridge designer to decide where to put sensors when to measure and how long to measure etc. The bridge weight in motion we used is developed in Slovenia since the early 90s and it's called Sibym and its sensors look like this they are placed under the bridge here we can see a girder bridge so we have sensors on each girder on the middle of the bridge and on the beginning and the end of the bridge here is one small slab bridge also its sensors so it's now used in more than 25 countries and the data is used in some countries for assessment but in some countries it is used only for some statistical statistics about traffic and here we will show what you can use the data for so traffic data acquired with this system is speed number of axles weight and spacing of each axle for each vehicle that passes the bridge but it also gives what is important the structural response of the bridge which gives the influence lines the realistic ones not the theoretical ones distribution of load over if we have a multi girder bridge and also dynamic characteristics of the bridge which can be important parameter and especially based on pavement condition etc so the application of the data is the most important thing because a lot of countries has the measurements but they are using it only in traffic analysis but we can use it for pavement and bridge design or assessment and also in some countries it is used for selection of overloaded vehicles so the assessment strategies which are including weight and motion measurements are one with the short term measurement and with the long term in short term we can only get the structural data the influence lines and the load distribution but it takes only a few hours for a few hundred vehicles so you can do monitoring on an existing bridge in practically one day or a couple of days if we need a longer calibration for traffic data and dynamic characteristics we need a lot longer period at least two months or 100,000 vehicles but I'm talking only about heavy vehicles so everything which is which weight is below four tons is neglected so you need some time to collect 100,000 vehicles especially it's not very frequent road so for structural data as we said we have realistic influence line which usually shows reduction over the theoretical influence line here we can see the reduction and as all of your engineers you know that this reduction leads to the reduction of bending moment in the middle of the span which already attributes to the bridge load to resistance ratio and it also gives us distribution of loads over girders so you can define or indicate a critical section of the bridge because not every girder is in the same condition not every girder is loaded the same and on these bridges almost every girder as they are prefabricated they have the same reinforcement and tendons so you can indicate which one is in critical condition it also reveals non-visible cracks and changes in the stiffness so here are the just quick ratio of resistance and loading on each girder this dark blue this is done without any monitoring and this is done with only one day of monitoring and applying the influence line and distribution factor so already here we can see that it attributes to overall ratio then the next step if bridge doesn't pass its assessment on this level is to use traffic data which takes a longer time so at least 100 vehicles but with enough data you can build a site specific traffic load model and with that you can apply it on your numerical model instead of your design traffic model here is the ratio also resistance load so this is without any monitoring this is with shorter monitoring and with longer monitoring so you can see it's almost double the ratio of resistance against load and also you get dynamic characteristics which can reveal the real state of the pavement which is also important for brief design and this is all deterministic approach and of course all this can be done and it should be done on probabilistic level with a calculation of probability of failure and values of ability in this so this year analysis was done using forum method I'm now working with professor tons on a big metal model to do this all with Monte Carlo simulations to get even better results I think it will be done in coming months and now the most important part is the part about how much this costs and how much the bridge owners can save using this monitoring method this is also not finished yet as this is posterior decision tree we need to relate the cost of bridge with cost of measurements which is a bit hard because not every bridge of course cause the same consequences of bridge closing or failure are different depending on the bridge type etc so we build a this is a simple visualization of decision tree where we can we have assessment without any monitoring with shorter monitoring and longer monitoring so the related costs which we defined or we are in process of defining is cost of bridge failure cost of bridge repair and cost of short and long-term monitoring but the important thing is that these two costs come this are direct cost but you also have indirect cost of course if bridge is closed then you need to take into account the other natural alternate routes the traffic jams the reputation loss etc and in case of bridge failure you can also have extensive damage not only material there can also be human lives in danger so it's a bit hard to to define them in the model for bridge in general we can we can speak in some ratios and we decided to do it like to take a bridge value as one and then indirect costs are one portion of bridge value cost of bridge barriers other portion of bridge value and we are in the process of finishing that model and then we will have exact number in value of information analysis but what we know from previous experience mostly in Slovenia where they are using bridge weight in motion for over 20 years they are using it for decision making and they in general they came to the conclusion that overall costs are minimal comparing to the fact that the bridge does not have to be strengthened or replaced or closed for at least 5 to 10 or even 10 to 20 years and if you take into account the cost of the bridge repair and all the indirect costs then costs of these measurements and analysis are really minimal here are some of the costs but they are really general and they also depend on which country are looking for but for example installation of bridge weight in motion for a short bridge is 75,000 euros which when we think about the prices of bridge design and bridge repair is really not much for short term data post processing is really not much because we only have a couple of hours of measurements and the bridge analysis is also not so much but of course it depends on the bridge size bridge type etc. for longer term the installation costs the same but maintenance and post processing are more expensive and we need to take into account the time that we lost doing measurements and of course the analysis is also a bit more expensive due to size specific load models etc. but when we compare this to the bridge value and to the direct and indirect costs of bridge failure or bridge repair bridge closing they are really minimal when we know what and the first presentations the amounts that colleague from Brisa was speaking about how much they spend how much are they spending every year on maintenance then this is really neglectable cost and I think that's it so we have prepared a couple of questions for decision makers so first one is do you perform any type of weighted motion measurements if you do the most important thing is what you use the data for a lot of countries are using the data only for statistics and you can use it for much more but you need to have some rich engineer who knows how to use it of course is this example enough for you to invest if you're not using already and do you use any other measurements monitoring technique for assessment and of course are you interested in a pilot project because if you are they are willing to do a measurement to do an assessment if you will if you are ready to finance so that's it presentation and of course the questions for the audience and especially for the bridge engineers and we would like to start probably with the first one are you using it here in Portugal do you use it more than for statistical purposes is anybody who would like to answer this question is it the reason the question yes so we're not using you are not using okay simple answer not using yeah so very first so that's right have you heard for it no no yes I knew that so this is something you so a study of last question is valid yes this is the most important for you is there any is there any other here who would like to show the previous previous slide right because with all the last one with the cost yes exactly that one you argue that your cost of monitoring are small which of course they are but the only thinking with that is that you least cost of bridge failure cost of bridge repair right as they are not accounting really for probability for that happens right the probability of that bridge actually fail or meeting repair are what tend to power by s3 that's kind of typical for a typical bridge right so I'm afraid that you are overestimating here by a factor of perhaps a thousand of cost consequences of failure of repair versus the cost of this bridge so this bridge if we could use if we have all the data for some bridge from the 60s or 70s we would use it but unfortunately we use this one because this was the only one we had we had over one year of measurement yeah but I agree but the reality is that even if you take the bridge from the 60s its probability of failure is not one right as would assume because it's still there and probably carry quite a lot of traffic so we have to be careful and I think the gains from monitoring are not that huge those are good example by Daniel Zonta a couple of years ago with Barcelona where it was I left with sort of feeling that it's more or less equal what you pay and what you get of course case specifically thank you for this I'm sure you have thought about the expected cost in terms also the probability that it happened multiply with the cost given that it happens it is a very clear statement are there any other questions are you dealing or maybe we can have the next presentation in the perspective of this that it is not already applied in article or seldomly applied rich rate emotion I think it's important to know that the note models for bridges have derived based on rich rate emotion so the duration of the load models can simply be read up with I think very reasonable costs and I think there's great potential in the knowledge of the actual loading on a specific bridge for a moment and in the perspective of the traffic prognosis for the next year so one addition yes thank you my opinion is still work to be done and I think this is a very good approach but as the exam we heard the previous example the value of this is connected to the to the study of the behavior comparing this data with the with the behavior forecast in the model so in order to understand this kind of measures should trigger alerts or trigger actions again the value of this isolated could be nothing because in fact you should measure you should compare and that should trigger actions and the value because the risk of failure and the need to repair is always the same the difference is between the means you are using to collect information interpretation and should be interpreted of course if you do nothing if you don't put sensors and you just inspect analyze the behavior of the bridge and you reach the conclusion that an intervention should be done in fact in the end you should pay the cost of instrumentation of course instrumentation if as more comfort in the safety point of view you can control better and also you can but you must interpret the results against the behavior models and that should be in this in my opinion if you create a framework this this work should be done in a more and not an effort more effectively we should not only focus on measuring and in fact a lot of technology is arising we are becoming more and more strong collecting data from the bridges the second is to put these in the models because all these information should be interpreted and then should lead to a framework of actions and that framework in fact the safe postings are in that framework of actions yeah I mean I agree that all this data was used to calibrate the existing models because first model was theoretical and then more and more data you build a more precise one and about the costs and how effective this is there's really good example in Slovenia they are using it for over 20 years and they are using it with their own department to do for decision-making and they they have a list table it was on one of the conferences I can't remember they show how much in millions of euros they managed to save I mean of course you can always do the repair and it would be better but the point is in priority direction because you don't have the money to repair all the bridges in country but in order to direct them which one should be repaired first and which one in wait for five years I think in that case this this is important because the best of it to repair or replace all the bridges in the country but you cannot do that so when you have a lot of them then you should rank them and for ranking I think this for sure but I believe that I don't know if you understand well about what I said I believe every owner in the world would like very much to have all the bridges with this kind of intervention have a dashboard with all the states and all the states of all every bridge every single bridge but for that they should have also the trash orders design and to interpret whether the behavior of the bridges more close to the to the limit state or not in order to provide prototypes and then and then in fact repair framework of action thank you so what is actually saying is we need the thresholds the warning the performance indicator and the associated warning and this is an important issue which we have to discuss also in tomorrow again oh my last session I will show you my life thank you very much for this comment this is very much appreciated and goes exactly along the lines we have been setting up this cost action and our decision like this yes the information must trigger actions and yes the information only provided if we can modify the actions actually just need to work through this we have it in our tools yeah maybe I should stop thank you very much that was also the reason that the simple example with no not you have a clear threshold depending and then you can do two simple actions so that that is what what it is important to have actions and so on so thank you very much for the presentation again