Hello everyone welcome to our. February special interest group workshop for G technical analysis. Today we're going to talk about simulating show laboratory testing. To determine clutches soil. Model parameters, I'm going to start in about 1 or 2 minutes, so that everybody can join in, she only loves people. Already blocked in so that's great just waiting just another minute here, so everyone can join then I'm going to start. Thank you. So hello everyone. Again, welcome to our February 2020 special interest group. Virtual workshop for geotechnical analysis today will be talking about simulating so laboratory test for. Fletcher show a model parameter determination. Or how to better optimize the parameters for your analysis. So. Today's webinar. It will compare will be around 45 to 60 minutes. I thought we recorded and it will be available through your personal link and IMO linked you got today. Will contain also the link to the on demand if you later. You'll also be awarded a warm personal development hour, which will be applied to your Bentley transcript if there are issues with this. You can also contact bentley.com. And then at the end of the session. I think quickly after session the. Tickets will be made available online as far as I know an otherwise just completely contact are Bentley Dot Institute. Has been linked com email? For issues regarding this and of course, the seek our seats are complementary. And so just enjoy the show an we hope we can help you understand everything a little bit better. Summer. Additional tips here while watching this special interest group video all of the windows, you can resize and can't move. So just make sure that you can see it at the best way you want to see it. There is a research list available at the bottom there's some useful links there to get some more additional information. Home if you have questions, please submit them in the question question and answer window and then I will. See if I can get to the questions during the during the presentation most likely I will save them up until the end. But don't hesitate to ask your question just put in there an hour answer them at the end of the. Presentation. So today, we're going to talk about simulating short laboratory testing plexus. 1st question of course, is why do we want to do a short laboratory simulation. Most of them should already parameters just want to start do analysis and get the results but I will show you why it's. A good idea to do this and also how to do this and the concerns to the next question. How to start this so tests? Laboratory simulation. I'm quite easy, I will show you how to do this and then also we are all need to. Optimize our model parameters to replicate the real life, saving behavior of the soil. After optimizing We have to get results from this dedicated tool into our. Updated uh sure model parameters data set back into a finite element calculation. So that's also what I will show you during a live demonstration. And we will end with a summary of what we discussed this. This applies to Plexus 2D N Plex is 3D basically more using the shorter test facility here. It can help you. Get this simulation done so that was applies both to pledges 2D and 3D. So what happens in normal cases, you will have your shorter thing if you're lucky sometimes people just have to run based on? Some correlations But ideally we have shorter thing can be instituted testing or laboratory testing combined with your experience you will determine the parameters needed. 2. Model your uh. Chittenango analysis once we have to determine the parameters, we can start up Lexus modeling. That's sort of the typical K Workflow that we have, but you have 3 lives with this is different between. More parameters an model parameters, so parameters. These are always determined through all the behavior of the soil in some real test. So there is a physical soil in a physical test that we actually do the test and based on the test results. We can determine the behavior and describe these behavior true some parameters. On the other hand, will shift model parameters and these parameters. These are required by the causative model that. Approximate. Grill soil behavior as good as possible. For instance, not all models capture all the effects and most of the models for instance, do not. Incorporate creep effects. Say did you want to model that they have a short which is? That shows crib behavior than if you try to sort of approximate this using a another showing model you might. Have to make some assumptions or limit the scope of the model. So you have to then fit these parameters structure that will fit that kind of behavior. So also be aware if user cost of model what the possibilities and limitations are such that you can make a proper assessment. Of which consider showing model you want to choose. So again real soil in a real test to make sure she'll certain so behavior that is not included in the context of model used to model that soil. Now we're to give the example of creepy haviar. Hum. You have to take it into account can also be there, they have like effect like the bonding or. The structuring. Uh, which were not included in the cost of model, but here is to have to. Replicate that behavior to use the correct show behavior in your G technical analysis model. So. Using so parameter directly as a model parameter may may not always what it may give a different behavior of the shore. According to the model compared to the real soil. And by using a model which is simplification of reality. We might overlook a few effects, but yeah, you can maybe tweak the parameters such that it. Gives a good resemblance of the actual behavior. Why should I do a short test we already concluded that the acoustic models cannot include all possible behaviors of real show. Still, model. Um showed the the IT might deviate a little bit from real soil behavior, so we need to check. We need at the end, we need to have the check if the conceptual model and their chosen parameters. In these gift gift similar behavior compared to our real soil. And we want to make sure that at least it works for the stand situations for which we have. Data available. Give me a large liberal reported laboratory test can be a few test, um something like that search that. That model will not give real weird behavior compared to what you expect so some important checks that we want to do. Uh of course, want to make sure that you strength is correct, we do not want to overestimate or answer. On their estimate. Real real strength of the soil and you might then overestimate actually the effect of safety or you might. Over a overestimate the strength of the soil or even underestimate their strength through then the model predicts fairly. Too quickly, which of course, is also not a good way to predict what's going to happen. Also think about strain levels. If we Give wrong straight levels, which are then loading we will probably also not do good prediction on deformation behavior. So if you do a performance based design. We're actually the definition are very important you do not want to. Be really off with the string level, we get from a certain type of loading so. This was a very important to check that. So how can we do this inflexus we have an plexus we have course you can just do a numerical simulation using a final. Model and then run the calculation, but as used quite cumbersome. You have to select a geometry, it generated mesh after running. Relations go to output program. Gather the results may be different defies the defining your own graphs. Which of course is quite cumbersome and is not easy and quick to do so for that for that? Specific feature of just simulating a laboratory test. Create a tool called a Plex server test and with that, we can quickly simulate standard laboratory test for. Chosen material set. We can then also easily compare the results of the simulation with the results of the real level of protest. Hum. And out of this is that we are, we start with the model parameters equal to the soil parameters determined from the test on the real soil. This is of course, our best first estimate of what we want to do. Then we run the simulation. So we can actually do the numerical simulation of the short test we can compare the results. With the rule of order test and if we're happy. We can use these parameters in the analysis and if you think that OK. Part of the the response is not as I would like to see it, then you have to adjust the model parameters. Shut you said you could do a numerical fit to run the in the shortest simulation. Finally, of course, if you find the found the best fit? Just have a look if. The multiple rooms are realistic, it could be that you ever perfect fit, but model parents are completely unrealistic and then if you go out of scope of the sole test facility. Then you might still get weird behavior or unexpected behavior and still overestimate the source trains under predicted informations. Um jump also always usually engineering judgment after fitting the model parameters. Otherwise, you might end up with non physical results. At the end of your finite element Simulation, so this short test is to help you. Better determine uh. Model parameters needed to run a numerical simulation to get there to replicate the real short behavior. So to start the actual test we can easily do that, if we are in a material data set. A window we can from database. We can click the short test button, so there's one in our material sets window. Or if we actually define the parameters, we can also have used the bottom, which is at the Top toolbar. And then we can click on this icon this Eigen Rep Jamals. And an image of interaction test, so it's also easy to recognize Prudential Test next to it on the bottom here. Should it also helps you to do that and it's fairly easy to use which I will show you in a minute. So what kind of options do we have in the black soil test we have the interactual test we have cyclic directional test? Which has been added in the latest version? Don't test at standard test concentrate of string test direction for share test. Also, cyclic direct simple shear now, especially if you want to for instance, model. Effects likely correction you can use this slightly direction push here to model cyclic effects and we have finally have the general option with the general option. We can control our boundary conditions in loading in the exact way that we want to do this you have to realize that this package showed test facility. It's basically a one stress point. Calculation so we just have a stress strain relationship to just do for one. Stress point as well as with the calculation is really quick. A downside of it that we can not only to find any models, which have a physical dimension necessary for instance. If you look at the consultation process. The consultation process requires a certain drains length combined with the real probability. Set thermostat quickly you are actually poor pressure will dissipate this is not possible in the Plex Server Test facility. Simply because it is a one. Stretch point calculation so it's very quick, most of the standard tests are available, but some features that actually. Need a physical dimension like constellation effects. These are not possible in short test. So looking at the actual test on the right hand side you'll see actually the apparatus and the measuring. In tools. Will get you these typical results that you see here on the left side, so it's uh a Top one is actual strange virtual Digital Express. On the bottom point we see a actual stranger phrases volume strain. An wishy different, more circle square with which we can use to determine the strength. So here we have a more. Visual representation of this, it's basically yeah, we have a source cylindrical soul, shampoo, we loaded with shell pressure. And then we're going to increase the load on the Top which will generate an actual strength and that that will. And give us some kind of deformations. Also in this presentation for today. I will only focus here on the actual test and how to quickly operate this. For sure, we don't have time to look at all. The other sort of facilities, but on our wiki whether a lot of. Other examples how to use that? So for this case here. We get the threshold test result an for the first part of our actual strangers. David work stress we can determine the stiffness. Ann. Also, based on the maximum load that we can put we can also determine the strength. From the bottom one he can determine compaction the latency, so that also helps us determine these additional parameters. That we actually need for our final calculation. So I see somebody has issues, with not seeing the slide if you do. Please reconnect and then actually try to reconnect to refresh this. The items. So the traction test here these are giving us the. These results. An based on these results, we can determine our sheer strength and can be effective parameters cohesion friction angle on the latency. Or if we do, and Andre Interactual test we can determine the undrained shear strength. We can also determine the stiffness. For me, it's a sheer stiffness from Interactual Test. But not a pure stare sheer stiffness, the soil sample. Also on the ghost some compression as well. While we load the. So simple it's not only that it's loading share but also it will be on the will also see some. Compression there. Also, depending on the quality of the device that you're using? You can also have a look at the small spring stiffness measurements, so that also will help you too. To detect and determine the small frame stiffness. So here in this simulation I'm going to eat. Define drink traction test we start off with them or column, so we start with the most simple model. I'll show you how the responses and compare that with the actual to actual test so it's a drained. Azure Tropic. Fully consolidated interactual tests loaded in compression, we start with the sell pressure 100 kilopascal. So did to determine some kind of the soul parameters, if we look at this graph here. We can fringes determine. Based on the more common criteria we can determine the. Strength parameters cohesion and friction angle. So based on. The more circles radius and slope of the more column line we can determine these parameters. Now this is of course, one equation with two variables. Which of course if we just have one test result it will not be sufficient since we're doing as traction test for sand. Here we're going to assume collision 0. And then the friction angle here can be computed using this equation here, so it's just a Sigma. Effective shake my one minus the fitted sheet my tree divided by Sigma, one plus Sigma Tree. So this gives us the signs of the friction angle. We can also turn RT 50. This is our sequence stiffness at 50% of the ultimate load. So we can use these 2 we can from this Top graph we can determine these 2 parameters. So, in this example, we just going to focus on this equation on this graph. These are the actual. This result because from a drain through actual test on send. Show pressures on the Clippers score. Hum. Now the first thing to notice is that? Yeah, we can. Determine the ultimate strength here it's a big strength at 225 Clipper score. We can also see, there's a residual strength at 250, two 115 kilopascal. So then comes the first engineering question here is which fell do we want to use? If you would expect that you were. Stress levels in the in the model will not exceed 4%. 3%. Four percent, yeah, we can still use the big strength because this will lead to a certain amount of. Strength in the mall if we expect to load. Further say did you want to determine some kind of bearing capacity for instance, then actually you want to use the residual strength because this will determine. Final strength at Philly. So it depends a bit on what kind of strain levels. You are expecting which strength we will use of course. It's not something there, something new, but it's just something to consider when getting these graphs and also to realize that. Once you get the results back from a laboratory? What kind of strength value do I get into the? Speak strength the Is the residual strength so that's something you have to figure out? In this case with the first analysis that we want to do here. I'm just going with the big strength. Now, with the peak strength, assuming in machine that I have 0 cohesion. I can determine friction angle using is Rachel of Sigma 1 minus symmetry over Sigma, one plus signatory so in this graph. We found that Sigma 1 minus Sigma 3. Is 225 kilopascal since then over the sale proceeds 100 kilopascal then Sigma one plus Sigma 3? Is for another 25 caliber school and this gives me a friction angle of 32 degrees? Next we need to determine the stiffness. So, in this case, I want to use the E 50. So what will do will take here, the UM. The maximum strength then will determine where it. Intersex. With the with my loading curve and actually here, so here at on the 12 point, 5K Pasco. We have the Intersection point and actually we're going to the user to determine the sequence stiffness. As my E 50. So we can see this line here. And then we can. Um. Calculate the E 50 or the Stephens here based on the angle. Usually extended a little bit here further so I've. I have a bit more better way to read an my reading of the result here is not that. Dependent on Some small accuracies so that also helped me to get a better estimate. In this case, my E 50 W 125 Clipper score over .7% will be roughly. 32 Mega Pascal. So these are my input parameters that I can use this on my soil parameters. So here we go to the training through actual test so actually I'm going to show you life. How to use this? I'm going to start Plex Estudi. Same thing in Plashes 3D. So here we are collectors, 2D connect edition will start in your project. And let's call this test for directional. Simulation. Say OK. And through what are you going to start the material database and clicking on the icon to show materials? Here we have the material state sets you can see here. We have the short test button. It's grayed out of course, because we have not selected material data set it's not everything. It didn't even create 1 yet. Show I'm going to create a new one. Knew let's call it sent. More Cologne. He's drained material who did it rain through actual test so also considering material. Um unsaturated saturated short wait and just give features values. Since the shortest facilities are one stress point. Analysis we do not use the material weight in that calculation. So actually this values will not be used in a shorter facility. We have to specify a stress level. During the analysis. So that determines the stress in the soil sample, so next will be stiffness. We determine that 32. Megapascal persons ratio for more column if you do it for primary loading it's usually between .3 and point. 354 drained material in this case, I will just go and use .3. Cohesion we shouldn't just be 0 so let's keep it 0 friction angle will be 32 degrees and also we have to. The latency angle the langey angle will just reuse. The rule of thumb. We don't have additional data. We just do an? Engineering judgment Ingeni Ring guess using a rule of thump rooftop, there will be friction angle minus 30 degrees. So I have to share it at 2 degrees. OK. Should we create them their data set here? I'm going to start here shortly test one click on them? I can hear. And here's the application, so we see here here on the left side we see. Where is the shortest window shortest? Structural test. Here we are. And we could in here at the. Her left hand side which is sent more column, choosing them theorem model mccollom we can have the stiffness. Is 32 uh? Megapascal personal issues .3 friction angle is 32 degrees an Dinesh Angles to degrees just as we filled in. In our data set. Now here at the Top you can see the different test that we have that are actual test. A cycling track should test admitted tests and we can have a multi stage loading here. Concentrate of stress test where we actually have. So we can have different string levels here that we can specify we have here direction procedures. Also, the cyclic version of that one and the the last tab sheet. We ask the general test where we could specify our own boundary conditions an? Also, our own increments. But. Here we're doing the threshold test this is the shore lap simulation that we have. So. In this case are self pressure here is Arnold kilopascal. A certain maximum string level of 10%. And we do this in 100 calculation steps and so this also determines the accuracy the more steps we take. The more accurate deeper analysis will be usually 100 steps is sufficient and in some cases when you hide. Not nearly behavior probably you want to have more steps, but socially is to increase at the Top here, we can also select the type of test. We can do a jointer under interest. Under in test? How do they drink to actual test in compression comes the extension? But in this case, we do compression. And we do in Hydrotropic test and it comes to a keynote. Control data test and what we can fill in here if he cannot value so actually do then. Can have a different ratio for any source stress state of shell pressure? But maybe a higher vertical or lower vertical load. Depending on of course, the real Institute stressed that you want to simulate this case. I'm going to use an Azure Tropic. A drink compression through actual tests on the present honest, Clippers cultural pressure and when you going to run the analysis. A wide bottom part with the command line here, which we don't use it in this example here so let's run it. And here we have the results. Can I hear double click on any graph that I have chosen to actually show. Up a bit bigger. So first party will be an actual string. Epsilon why, why have as you can see also here in the image here after the Y direction is there. Vertical direction here, so this is my. Vertical actual strain and this is Sigma 1 minus symmetry insurances defeat Orix train and that we're going to load until. Failure and then we have as the more common model. Predict San Elasto and linear lastic perfectly plastic model. We also see here, the volume strangers, actual strength. We actually we have first compression compassion and their worth. Native behavior, please realize that plexus uses the classical mechanics sign convention, which means compression is negative tension is positive. So compressive. Stress will look with negative sign if you have compassion. It's also negative strain if you have. Positive strain is actually. Sort of expansion or little late if behavior. He don't have signal one Sigma 3 plots and plot here in PQ plane and of course, also very important. The UM. He more circle, the more circle here is just Sigma freestyle and actually we see here, the the. Someone made a major principle session, Sigma treat minor principle stress, which should be equal of course to kilopascal. Which of course is the sell pressure? So we're looking at this result, I can roll double click, and double double clicking home. I'll get more data. I get to get larger, graph and also see that they will the results here. So what do we see here? Nice thing is uh. Should we in our desk. We said OK, we want to use the peak strength? And in this case actually you can see here also that the maximum. Take 1 minus symmetry here is. At 225 point, 5K Pascal is also our input value. We said that they actually did the value. We want to achieve so also $225 so the model here protects predicts the same strength. Runs or something to be expected, it's also good to check. Another nice thing is here is that we can based on this we can determine the stiffness stiffness is here just the. Steepness with the line here. Uh so we can read the value sharing computed. But it's also nice is that? We can just click a point on the line. The Plex, a short test tool will determine a secant and attention for this line as long as there's one line. On this location. The program can calculate the Tangent Secant. He just needs to hear the origin and the point that you click. And here from the bottom part here. You can see it here, she contingent are both determined at 32. 1000. Since our units are set to kilo Newton meters is actually 32, Mega Pascal so this is also. Hum. This directly resembles our input values that we put in the morning as we would expect. And then of course comes the question. Um how good is this compared to the real short test. So I'm going to. Go back to the presentation slides. And actually here we have the? We see here are the actual results. The real soil soil test. It's in the blue line. And the red line, which is the results we got from the more column. Analysis. But we can clearly see that OK, we reach the same peak level and 50. It's the same thing it's the. Basically, it's the she can hear through this point here, but we can see it, especially in strain levels of. Like a point. 8% to let's say 3%. That the stress strain relationship is not that good. So if you do a performance based design, you can quickly see here is that we will not reach the correctness strain levels. With a shirt and loading condition. No no especially also hear the the analysis is much too. As much. Too much stiffness and this part so you underestimate the. The formations and then suddenly here we have plastic deformations so if you do. And ultimately limit state calculation and you actually go to bearing capacity analysis, probably will not be that bad. But if you really want to get the deformations predicted properly that's something that you really have to do. To check So what's next? Yeah, it then we chain, then should not do some more color model, but the more advanced model and for instance, we can use the hardening. But for that of course, we need to have some additional parameters. One of these additional parameters are is the diameter stiffness. We also need to look at using unloading reloading stiffness. And there's an input parameter called Power M, which determines the stress dependency of the model. We do all from the analysis that we already did we already determined the 50. So that she can stiffness at 50% of failure. So. That's already something we have, but adult descriptions we do not have it. Typically, you will get it from your compression tester admitted test or might you might want to use correlations to determine this. Hum. Stiffness parameter 4 primary compression. In this case here for simplicity. I'm just going to assume udometer is the same as the Earth as the 50. So that's Just an assumption here that we're going to do because we do not have more data in this case, we just have only this. One through actual graph. The you are the only reloading stiffness. We do, we can get it. We have an unloading reloading path. In the actual test. So we can determine the stiffness from that point. The Next One Power M. This gives a relationship for the stress dependency, you can see also here. The equation on the bottom right side. You can see we have a? Reference stiffness the E 50 ref. Which is the 50 E 52nd stiffness at our reference stretch level? If you have a different stress level, so basically that means the variation of Sigma tree or for the track. That means different cell pressure. We will get a different. E 50 she can stiffness an using this equation there's a power law involved. And that will be used to determine the sequence stiffness. At different. Stress levels. So with this weekend determine our stress dependency for class usually it will typically be around 1:00 this power. Ann. Um. For Sands usually lays in Rays of .5 to .6. Of course, these are typical range is getting results for Realtor actual test will help you better determine these actual parameters. So uhm. Yeah, so, based on these parameters, we here in this case, we all assumed udometer end power M already defined. And we have to determine are unloading reloading statements. So back to our drink directional test result. To determine my EUR. We take these 2 points here, an we make a we just make a line through this and based on. This English of the line, we can also determine are you are extended the deadline is a bit too. Hum. He had to be accessed in here 2D200 Clippers goal line. With that I can have a better range and with it, I get. Unlearning reader statement above 86.5 mega Pasco at 100 kilopascals sell pressure that's nice so by Pier F. Um for the stress dependency, it follows the same power law. As for the E 15 so if I take my. Reference threat level at 100 kilopascals directly here have my own learning reloading statements. So let's go to the analysis here. I go back to the plexus application and gone immature dated base, I will create a new data set. Um. I've got the shan't battalions dehydrating soil. Model. Have to change it here from more column to hiding so and there are different models available depending on what you need. In this case, I'm just going to use the hiding soil, too. Determined. The model parameters. Again. I've just put it in the features wait which. Um. It's not necessary for the soil test facility. So they had defined the 50 rough. Clutch it automatically suggests. Uh the 50 in the you don't have to be equal in at the ER is 3 * 850. But OK if you've ruled Thursday to you can better not use this correlation and also depending on the titles. Open till that you have you might also want to use a different different ratio between these 2 in this case we found that the? Unloading reloading stiffness. Was set at 86.5? Make up call. Collision said to zero friction angle 32 degrees and the latest angle 2 degrees there from default advanced stiffness, but one of these will be the PDF. Here, which is a set the tone of Cooper School. Um Pier F this is just the reference stress level for which we will determine our stiffness parameters. Can you can see as they cannot add normal conservative state using Orange Jaquish Formula One minus minus five to determine the? Default value will show up here. A person's ratio for unloading reader unloading reloading which is typically in the range of 15, two point 2. He would just going to take the default value will be .2. So again let's directly go to the soil test facility so he ran it up. I have his few buttons. And the last one is the shortest button. So here we see the input parameters, Harding soil E 50 MAT. ER there's just dependency. Fashions Rachel came out and she and the pier F so the reference stretched level where we determined his parameters here. Collision. Friction angle of the latency. And channel properties of the weight. So, in this case, I'm just do a trend compression isotropic. I download. Kilopascal show pressure to Maxim strain had the maximum strain here is dependent on test that you want to simulate. So if you test goes to 6%, yeah, you can just do it like this is a default to 10%. Which usually will give you enough information number of steps? We're going to assume that our scientists know more normally consolidated so there's no free consultation stress and also the. The mobilized relative shear strength. Is also just switched off so we do not have any uh relative share string mobilized shear strength and we start from. Normally control date state. So. Well, running this. We can see the same graphs again. Hum TV torex transfers actual string here, we have the volume strength is actual string and so we first compaction here. And then actually deleted behavior because of the langey angle. Here we have the Sigma one Sigma 3 plot so we just there without here vertically in this. Plot them will keep the shell precious Sigma tree and the vertical stretch will be increased. QQ plot one over 3. Information here and here we have our Marshall. Totally looks very nice. I can also look at the analysis here. Ann. Maximum strength here is defined at 225K, Pasco Digital Express.' Let's also actually what we Set has a peak strength. So that's wonderful you, also want to have an here. We can also determine are seeking stiffness or that will do it just zoom in a little bit. Yeah, I can just. I'm clicking this button here with the plus I can. Join in the area. I want to see an at around 12.5, which we love it here. I can click. What else here is that my secant is 31.2 Mega Pascal and my tangent is? 18 Mega Pascal. If you want a little bit, we can also see this happening here. Chill tickets definitely bit lower than we would expect the question is how good is this in this analysis. So let's go back. Today. Slide and actually going to put the results are measures are good to show you. How to get the results from this so I will double click on this window here? Yeah, we here we have the graph here with the table. Using this copy to clipboard option, I can choose copy the graph to the clipboard. Or I can copy the results table to the clipboard. So that means there when I had the kindest export the results and I can maybe put them in France is Excel or any. Plotting option and we can actually compare the 2 results. So back to the. Slide where we actually see the. The result. So here we have the? Analysis here done so the blue line gives me the. The real short test and the Orange line here. Give me are my simulation using Harding showing model and you can see this is a much better match compared to the more common model. Might overpredict, On the predict here, the development of strange on this level, but overall, it's a pretty good match so this looks nice. But we Only have the primary loading part of course here. What we do have is that we have loaded until here that my Sigma 1 minus Sigma trees are not as? Cheetah Pascal. So with this we wait might want to do also a simulator, loading and then unload reload. This behavior and then continue loading. So the question will be. How to do that and we can do this using the general test and track should test is just a simple? Uh. Um Bloat. Continued loading on Tori reached the maximum string level. But we can now use the general test what we have to specify in our own boundary conditions to do that. We have to add and unload real part. In the answers. Of course, we started with our boundary conditions. We still have the Azure Tropic. Uh uh starting conditions show my boundary conditions will be that my all my primary. Stress is Sigma one Sigma 2, two Sigma 3 are on the Clippers call. Um noted here, we have to use negative. Such values because these are compressive stresses. The next part is that we need to load until that Sigma 1 minus symmetry is equal to 100. 87 Clipper score basically means that we have to increase our vertical effective thresh with another 87. Then we do the unloading until reach about 20 KB Call Ann. We do that by putting in plus or than 67 kilopascal to simulate the unloading then we again. Reload after which the 87 Clipper Score Digital Express' value showing again do the minus 167. And then we're going to continue loading until we reach failure. And So what you see from this loading path it should be remote controlled. Now we have to load until a certain load value unload to certain load value. An continue loading so we have to stretch increments. So let's start doing that and go back to the. Analysis here. I have to close this one, so let's go to the general test. We use drain test and we don't have any vertical precompilation stress. Uhm I don't like to sleep with biggest with better input. Are my initial stresses were set at minus RN? Killer push Cole Default here for the boundary conditioner should strain increments will have to change them to stretch increments. Also, the sites we keep the shell pressure constant at 100 kilopascals. We do not sort of fixed shape the boundaries like you would do with it. I don't want testing. We have a constraint modules. But in this case, we going to simulate this using a stress increments, so we keep the The X. Direction and that direction would keep them constant and on the Clippers cultural pressure an we've got to increase the vertical. Stress. First, it will be loading. Until 187. Kilopascal. Then I'm doing in size to undo the unloading 200 and still on the 67 which we should. 20 kilopascal in the TV toric stretch. I'm going to add another one over here load again. Until. Failure here now with you. She also per face. We can specify the duration and the number of steps. So the Russians he said the one day steps 100, so dishonest determines the accuracy that we want to use. Calculation will be done. During the load increments in 100 equal intervals and duration doesn't play a role in this analysis. Generally, because the Harding show model is not time dependent so there's running through model itself is not time dependent. So it will not have any influence on this. And finally here, so we're back at the. The point where we started unloading and then did reloading it will reach the same level and now I'm going to load until failure. And I'm just going to add let's see. 60. Kilopascal. OK and let's make it bigger and let's run the analysis. You can see it's done quite quickly. It's already finished for this for faces. Um proper Sigma one further Sigma 3. PQ 'cause here, we have loading and then here we reach the failure line actually we have dilated behavior here. And here we're going to see them or circle at failure. We actually missing here is the graph, which we are interested in which will be actual strain versus. The uh if you toric stress. Through that we can add a new custom chart. Let's call it a 3 actual result. And, of course I have to edit this one. So I have to show the X axis and is already set to absolute one. An only vertical axis. I want to do is shake my one minus symmetry. Now, since we want to put it in the same way that we get our. Results it means that uh since him plexus compressive strains are negative. The graph will go to the go to the left hand side, so if I change. It you will see this happening here. So if I can pick this one here actually she hear that indeed his or her cold. Left side, but actually want to go move it to the other side, so we can just again and this chart. And invert the shine. And because she had go to the other direction. And you can see here. Which is? A larger value of strange here. Open till Then, which of course, is our huge values so they are loading point is we put here just too much for the for the. Analysis where you want to tune down a little bit. I put 2 - 45. Let's run it. And you can see here we go to. Today's value any gotta see a bit of the unloading reloading parts be difficult to see. Here we are so we can see here we have the loading unloading reloading and then actually here we reach already. 10% string so we definitely reach failure here with this horizontal line. So. How good is it again? You can just. Copy the results table to the clipboard. And we can put it in the results. If we go back to the analysis. Hey we put the results are combined. Then we can have the A train blueline again. We see the actual the actual to actual data. Um with the Orange Line is the threshold loading for the Harding showing model with the direct input of the. Show parameters as model parameters. Well then integrate language, mostly overlaps with the Orange Line That's to be expected. Is that UM which is the point? We have the unloading reloading simulation now? What we see here is that? I will load until a certain a digital exchange level. That will be off where which makes sense that we could resolve that the. To put that part stings a bit too high, so also we tried. Downloading originally part will be done. At the Different location. You can see that still that the only report is done. It's very parallel to this one. But you can see it's not very good match. The question is what should we do. To get a better match. So to get a better match what we want to do, yeah, maybe. We still want to use the residual or maybe you want to have a bit of a lot since for this. David that we want to see is a bit too. Stiff we're actually going to lower the stiffness here. To model the tomorrow the? Get a better prediction of the model parameters, which can be slightly different compared to tractor because there might be some effect. And as we didn't take into account in in the hard Lakeshore Mall, which is actually available in the. Real sure. Show to optimize the parameters, I'll go back to the shortest simulation. And we are I will close this one first. I'm going to save this result. I'm going to call it just test 1. And the next part is I was just going to load this. Then I finished it. He can see with Green. That. And also visual in a little bit you can. CD analysis here chilling also a green backdrop, which basically means we can sort of like the reference value for this through weather. Improve the result. George had before we won't change the friction angle to 31 degrees. Let's keep the shame of term so I. Also, going to lower the latency angle bit and I said, before we want to have a bit softer behavior. So let's assume we going to use 30 a kilopascal 30 megapascal. Through the megapascals so should should dirty. E3 to 30, * 10 to the Power 3, Kunin per square meter. So we change the parameters little bit here and let's run this simulation and see what we get. So, in this. Bolt we can see different behavior that let's zoom zoom in a little bit here. So we can see it better what's going on. So we can see here is that the blue line here represents my current. An optimized parameters, I'm not sure if there already optimized but let's assume that they are already optimized and the green. One is the one the original run that we did. With the direct input of the solar parameters. Now you can see here we have a bit of a shift here to the left. Visual mean a little bit. We can see. Clearly here. There's a difference still these coloring real parts. There. Parallel should dishonor sickness is OK and we get it. Maybe it's off the behavior shortly point will be a bit here more to the right hand side. So let's then also combine combine this. In the results. So here we have the updated parameters the effect of that, so again the blue line here represents the Realtor actual test. The Gray line. Here overlapping the Orange Line. Is the unloading reloading simulation with the original shore parameters and the yellow line here? We actually did the? Unloading reloading residual. Basically, it's the residual strength you can also see here that the file strength is a little bit lower. And also the I don't want her in the 50. Are Lower 230 megabytes core then skip skills? Are very nice match with most of the curve here? You can see here, the beginning up until we do the only reason part it's falls it fairly closely. Send the unloading reloading parts also follows the other reloading part of the Realtor actual test. Quite closely and we're going to continue loading only after let's Adrian have 4% results are a bit different. In terms of ultimate strength what that makes sense because we just use the visual strength. Up in the morning in the model parameters as the input for the simulation. So this gives us the final situation, which is better optimized. Item. Then of course, what's next optimize the after the optimizing the parameters. We have to include it in the final. Analysis. To do that, we can copy. The optimized with 0IN short tests using the copy to clipboard button. And then we're going to place it in the platters 2D to 3D material window. Things are going to do that. So we have the optimized parameters. And optimize parameters here called back here that's with these are the 30 megapascal stiffness parameters 31 degrees friction angle. I continually option called copy material. I copy it. And then I go back to the Lexus 2D material data set window and actually here. You can see here. It's still at the original values, it's not updated. We have to explicitly update this. Can you search using this button here? To paste it and you can see here at directly changes them. the E 50 ref to 30 megapascals. Shamefully odometer. And here the friction angle should do 1 degrees. And now we can directly use this for instance, too. Create a server layer an apply. Here. Diving show in order to it. So we have a quick way to optimize the parameters. We don't have to travel ourselves during a full final and simulation. Then quickly do it optimize it and then reuse and use these values in a finite element calculation. So it was quite easy to do. So to summarize here. General advice, I would say that you should do always solar thread simulation because the model is just a simplification. Of reality it might over it might not incorporate certain effects. So we have to adjust the parameters a little bit to make sure that it actually resembles real slow behavior. It's very easy to do with inflections in the soil test facility created material data set it supposed is already did. Before clicking the shoulders button run your shorter simulation its calculation to be done here is just a matter of seconds. And then you can also compare the response with your analysis and then compare this with your test data. You might need to require to optimize the little bit to replicate real life behavior. And then finally copy optimized set to as a into the data set into practice 2D and 3D. To use this in your finite element simulation. Before we go to the questions. Um. There's some resources here on R Bentley communities wiki and there are several videos. I want to highlight two of those. It's the general soil test video. This video interactual testing how to use that there's also a verification of the Harding Shore Mall. Interactual testing, which actually is going to use a finite element calculation. To verify Harding soil and compare it with real soul data. Some upcoming plexus events upcoming events February. We have an automation workshop in real classroom in the Netherlands getting off March. We have the advanced course. Computational geotechnics or we also teach more about Flex is a special course on dynamics and seismic to Tango Engineering. Um in the same week. We also have the next special interest group video plexus and black slow too. Simulated groundwater flow. To Register here, the online events are done through bentley.com and then go to the events page should have. Picture at the Top. Uh. The other events, you can go to pledges.com/events where we have the classroom events also next week also have. Another ski technical analysis. Web and R it's about interpretability comes to go to bentley.com again, then go to events overview. Toledo met up menu and then it also sign up for the for that one. So now it's time for some questions so Paul here if you want to have somebody from Bentley contact you. Orificial additional questions just click yes and somebody will contact you could be about plexus or if you're more interested in how to manage your. Troll data we have also applications called Keith or kinetics hallways where he actually can also have this data stored. Show hours he budget questions so that's good little bit overtime here, so overall free 20 minutes to go into your questions. So let's see what we have. Uh bring us here conversation is not possible in plexus Realtors. Indeed, it is quite possible yes. To control time in your test you can just specify the amount of time that shouldn't test needs to be done. Um. Also don't open my screen sharing here. Let's again go to the short test facility. Yes, you. Can. Just feeling. Time. Interval for your face. Of course. They considered model needs to be able to model creep. You can think about social creep model or the script 1S model. So I'm going to the shortest here. Non directional test we don't have time available, but friendship go to the dominant test. We can have had duration of Princess is distended. One day but say that you wanted to have a 2 day test or? Admitted test with a 24 hour loading time you can French or 24 hour response time you can for instance, to like. Azure time interval increment of. Hundreds and then maybe we do additional loading let's say yeah. In. Share today will do is minus 100 and then actually we're going to do a duration of one day. Without any stretch increase so basically yeah, we load and keep their fairly constant just run it. And you can see here. Will she have time effects in this case? Of course is the model is not time dependent the results are constant. But you can just run this to simulate also creepy Asian realize that consolidation is not taken into account so. If you have like a creep. Material, which has a very slow hydraulic nativity, probably crypto also take, will take place at the same time, so in that case, probably do we better? We do a finite element simulation with the real dimensions. Question here from a stream and how do you determine maximum strength maximum strain is something you get from the test? And the maximum strain in Blacks is just assumed to be at 100 kilopascals so we just. We just did something you as the engineer have to decide which dimension restraint. Here is that you want to simulate. OK. How is the moment that defined the model of a city is defined for unconfined compression is confined by definition so it will be different. Creation of actual stretch divided by liberation of actual deformation. Yes, that's right so the odometers indeed is a constraint. Monitors. Told that answers your question. He is the question of the adomet are easy 50 would mean the confinement has no effect on strange. Oh no So what I did here is that the The Udometer. Uh so if you look at how how did the effect is done for the E 50 it's based? Omar the show pressure. Well, that udometer. It's a constraint models, but it's rated based on the vertical stress. So it's not taken the same stretch base as the reference. So uh. Practically it means that you get at the 50 Rafini odometer ref in plexus is roughly can be roughly the same. Hum. But it's not taking into account the same stress for the reference reference level, so it does give a A different behavior. For that. So it's not the same stiffness now that you're 50 is a stiffness based mostly on sheer well the. There is more of a compressive strength. Now compressive stiffness. So current from Jacobs ask what's different between 8:50 and 8:50 ref so for death. I will go. Back to the uh slides. Let's see for the more common model. There we are. OK, so did you die with their daily 50 and he 50 ref this? The equation that explains the difference? Uh we know from chance if we are going. Deeper and deeper in. In a lower elevation, we know because we had higher stress levels that there's differences higher. So there's a certain stress dependency of the of the stiffness. And with this equation here we can actually determine that So what we do here is that with the E 50 ref. We determined this stiffness at a certain stress level in this case, PRF, which we are not kilopascal, which is our self pressure. For the actual test at least. Now, if we very stress level say here say that you are. Normally sort of would be maybe, as a like 5 meters depth. But if you had 20 meters depth. Sigma tree will be much higher. Compared to the at 5 Level 5 meters depth. So with this equation, we can compute the actual. Stiffness at any stress level and using this equation, so that's the difference. We have a reference. Stiffness value editor than stress level turns to the pier F. And with and with this, this power law that we have here we can compute the Stephens at any stress level. So I hope that answers your question. Trevor asked how about has a small and we can also use the HMO in disorders facility of course. You just have to also take into account the small frame stiffness and you can get it from a different. Test the resonant column test or something Bender Test. Uh or if you have good measuring devices in your traction test you can also get it from that. Somebody asked her if the period should be 200 kilopascals and no, the pier of here that we use. These aren't kilopascal because it's a Sigma tree that's the minor principle stress an in this case when we do a compressive. Structural test it, you can take it at the same as the sell pressure just Sigma 3 will not change. During your threshold test. Exactly. Correction. Test it. Honest. Shell pressure. How for writing this for people who watch the on the month session they can also check their questions later? Uh we did the traction test around Clippers cultural pressure an? Since the uh stressed dependency. Yes. It's taken. Owned. Sigma. 3 value. We can take. Be brief is on 100. Cubicle. Why would ask you also how was it out of turn it was less than the secant and so let's go again? Cody traction test from the analysis of the results. So here we took the she contana stiffener and the tangent of course, what you see here is here because we have this. Is kerf line here is the secret is just the line from the origin to the point here? Which has this definition. But if we zoom zoom in you can see that the tangent to the line here. It's not at my age. It's not as cheap as they seek in there, so that's why the tangent is lower than the chicken. Now it's just because of the shape of the line that the secret here is a higher value than a tangent. Probably for looking at the beginning. Uh we can see here. He says mostly down here it tells you to watch will be lower than the. She can hear. Got here at the beginning. Well. There's honey any difference here. OK. Do changes with the initial void ratio influence results hardening showing model does not depend on the void ratio. So, in that sense, the fortress will not influence the results. So, in this case, the this doesn't have any influence on the results if you use a model another model? Like for instance, the The modified gameplay model for instance, or a software Model T have these effects of default racial but. I just wanted those not so it will not have any influence. Brittany asked could you explain? What is the relative mobilize shear strength we could have included in the test simulation? So the uh. Now I think so model is model, which has she hardening any compression hiding. These are 2 parameters, you can. Influence independently in your shoulders sheltie. Maybe it's good that I. Load on the Model here let's go to Tilia Models Manual. Hopefully, I can quickly show it to you. Let's go to the. We are. So here are the material mulch manual and let's go to the. Ordering showing model. Sorry I don't do it in. Show me what we see here is we have different yield low key for constant village of the Harding parameter. Garraty, which basically just as she heard the name. Anne let me see where we have the other graph. Here is the pre consultation stretch this so this is our vertical pre consolidation stress. And we can fill in the shortest facility, which turns are BP this keeps are compressional. I think so. There are 2. Hardening effects in the Arlington Model 1 is the compression, hardening controlled by our vertical pre consultation stressed. And we also have a mobilized relative shear strength. Basically. They go back to the unit lokkhi graph here this gives us sort of and so, if it's 0. It started here and then it will just take the. Couldn't stretch that where we are? It could also be that you are very close if you already have 30 more configure line than the. Mobilized. Relative shear strength. Is here? One in an order based on loading history. You can be somewhere in between. And then you have to give. Is Rachel? Question here it would be nice to see how it could be done using Python an. In order to avoid many changing for many change of the internal parameters. Hum yeah, you can with the new version of that and we plan to do a special interest group. Video on this later this year, so probably June or July, the program is not fully fixed yet, but we want to. Show how you can optimize things using Python. There's also an optimization to already available in sort of facility. Um. Which she would? We're here. There's also the option where we can do a parameter optimization and then like this role can also do this optimization already for you. Yes, only says we should take the stiffness from the platelet test and not from lab. Yeah, so do engineering judgment, which will determine which you can use your stiffness parameters note that from the laboratory it gets. Information from your specific soil sample well if you do a plate loading test you get this difference parameters from. The complete show mash in city so the play low interest is more of infill test while we can do a laboratory test. The Germany. The individual parameters for the soil sample. Nadim arts here, the question unloading and reloading curve is following the same path where we post to fine tune it first. Loading plot. Different from rezoning Grift Yes, you can do that. But in order to do that. You have to use the small model. Yes. Using the Teacher small model. And as you can use this user so we can have a different unloading and reloading path if you. Used to hardly control. Data small model, we also take small spring stiffness into account. The holding short model by itself is just elastic in the living room. Bart so in that sense, it's taking that not into account and again this is a limitation of the hard links. Model where it cannot. Replicate that kind of gave Chris not part of the conservative description of the model so either have to. Assume and say this is good behavior for the model that I intend to do or you have to. Go to one step more advanced model using the XS model. OK let's see the next question. The web and R can be accessed at a later time you can just use the link from that, they have. Be sent to you. Through the email you can just check it later on the month. Mature or is the question is there any difference when you want to model software that software and hiding soil. Yeah, of course, there are uh limitations in what you can do with the software and harming soil model. Um. Friends if you have really soft soils hunting story model might not be able to allow this summer, Strick Tiens. On the different ratios of the E 50 odometer, and unloading reloading stiffness. It cannot take into account all different combinations in that case you then you should go to the software model. The social model, it just is also stressed dependent but it's linear stress dependency in order ensure model. It has this power law that gives you a bit flexible flexibility. On UM how stressed dependent the model is. So that determines a bit so, based on the show parameters that you actually have so the real soil parameters. It determines basically you want to use a software model or the highly joint model. So someone else can we control rate of loading in a dominant test. If yes, then it can be called consultation test. It depends a bit on what you define as consolidation if the process of consultation means that you will. Generate accessport pressures and this has to be dissipated. This is something we cannot do in Seoul Test. Insult us, we do not have consultation process so the dissipation of export fresh air do through a natural. Electricity is just not something done in short test so if that is the case in your analysis you cannot use. This user find the elements small final model use the actual drains length and compute that to determine the. Actual behavior. Hum. Miss Army Tarses pasta use unconfined compressive strength test using to optimize so Gramma approach has of course, you can do that. You have to use as a general test. Uh and then. Play with the parameters to do that. But she robetta ask can I? Use escalate in shelters that's a very good question. I think with the latest addition, we indeed did it. Um. So I can use the user defined and your friends with Christy. Oh, I can just of course, she was here at the whole pro model. We just released the whole problem, which offering model with here. Just normal pro model. For some features values here. So I can tell it took me 10 gigabytes com .2 uhm. Here we can for instance, uh. Specify. Some parameters and just going to click with some of these using these options here. TS I almost had it too. Let's say 60. Disturbance I will not do anything disturbance here and this gives me Mohawk Brown criteria. Let's go to shortest. Here we are so evil ground. We can also run different tests and in this case you just have to make sure that your boundaries. Dishes are OK. So for low ground you can do that just as desk. They want us we can also use that. Sorry to see Ashley. Model that you asked but you can also use that here to do the Ashley shortest. Here's the question should estimate you dont's attention at Pier F. Adopters definition you should get it from the odometer test so we cannot directly get it from Interactual Test. Um very nice question here should I take per F as Sigma tree or pier, FSP now. PDF is Shake my. Tree. Hum brownian goalie art is there any considered model inflections that can resemble strange softening behavior actually we released the look Brown? The softening behavior. Uh. Let me see if I can. Show you the model here man. So what you see here. We have a hook and Brown model with softening behavior and actually here. Indeed, we have work procedure softening behavior. I want to show you the? Link here. Just just being released. So it's you can also use that one. I'm out of the question what is tentacle of tension cut off is an upper limit that we set for? Tensile strength in the toy model. So for instance, if you specify a little relatively high equation or effusion unread shear strength. You might not want to allow. Tension at all or just a little bit with attention cut off, you have an additional. Plastic limit in that case, it's a tensile strength, so if for instance, you're Sigma 3. No. Get into tension to Plex collision might cut it off at a certain value such that it has to reduce with it stresses. So we also can get tension cut off in your analysis, Indiana in the results to classic points. You can also see tension. Point each other points were actually the filter cutoff. Nothing I asked the question so just have 2 minutes left and she'd still 3 questions so that's good. Um. Can you please share again the link to the video showing? Analysis uhm there should be in the resources uhm. But let me quickly go here holders to access. Short test. Unable to link. So this is the main it should also be available in the resources. But also put it here. Somebody asked about especially hard links older model question darling parameter of the compression rating service depends on the average maximum. To expression or only the vertical starting parameter for compression depends on the. On an equivalent. Mean effective strength, so it's not the uh depends on basically not only on your meaning factors but also. Uh. Uh because it uh. The capture fish so you surface it has an elliptic shape. So it's not only the vertical. Question here from amount is most of the engineers use more color model that's true but as I've shown you in analysis. The Harding soil has some advantage over the more common model French if you just interesting strength more common model will probably be OK. With the cases, but in other cases, it will not do very well prediction of the soil behavior, especially when you do. Dating all analysis and you have to do deal with unloading stiffness reloading stiffness primary loading share loading doing this in a more global. Can be very troublesome to get a good prediction and in that case using and more advanced model would definitely benefit. So uhm. Are strongly suggest sort of to go from the basic model and more advanced model to have a better? Representation of reality. And. And so we also follow up question on this and still ask questions. I have to ask if you have questions. Please go to the service request manager on our Bentley. A Bentley Connection Center. And there, please submit the service request for more detailed questions uhm. I want to ask what's the engine more common deep activation problem. It doesn't take in consideration, loading reloading fetch difference. Can we increase the stiffness of soil to two 2 * 2 to 3 E ref at the bottom? Activation. Only while using Google. If you're dealing with the more gold model. You have to realize more cool model has only one stiffness and you as the engineer have to decide. Which stiffness you will use so depending on the loading path of the analysis maybe you want to do? Put in the unloading reloading stiffness their decisions you have to make as the engineer. So you can indeed increase the stiffness if you know it's going to be unloading you can put the alarm. That's possible but please realize that he was the engineer didn't have to make decisions what's the loading part is. Rather than that the calculation helps you to determine what the loading direction is if its primary loading wereling reloading. Or even smaller stiffness, then, yeah, as the user you have to define it yourself. It's better to use an advanced model for this in my opinion. OK, um. I see the few questions left up, but I'm afraid that we don't have time anymore orbit. Overtime so I thank you for your time here and I hope it was useful. Uh if you have more follow-up questions, please use the service request manager to submit your questions and then. Thank you for your time. And I hope to see you in one of our following sessions. Thank you.

When conducting laboratory test results – such as Triaxial and Oedometer tests – you want to make sure that the behavior of your chosen constitutive soil model captures the test result data. With the SoilTest feature, PLAXIS offers a quick and simple method to simulate these lab tests and verify the model behavior. To learn more, this Geotechnical Special Interest Group virtual workshop is a must see!

The agenda for the one-hour session encompasses:

- How to start a soil lab simulation
- Optimizing your model parameters to replicate real-life behavior
- After optimizing, how to quickly update the soil material definition in PLAXIS

The Geotechnical Analysis SIG is open to all Bentley users, so invite your colleagues!

***The webcast will be streamed through your computer, so there is no dial-in number. Please make sure your computer speakers (or headset) are turned on and the volume is set to an audible level so you can hear the presenters. #EVENTHELPURL#*