Name: CadenceTECHTALK: Water Wading Simulations and One-Way FSI with the SPH Solver
Uploaded: 2026-09-10T16:16:47.677Z
Duration: 48 min 8 s
Description: CadenceTECHTALK: Water Wading Simulations and One-Way FSI with the SPH Solver

Transcript for "CadenceTECHTALK: Water Wading Simulations and One-Way FSI with the SPH Solver": Alright. Good morning. Good evening wherever you guys are from. We're just gonna let some people trickle on in before we get started, so hang in there, and we'll get started momentarily. Thank you. Alright. Well, thank you everybody for joining us today. On behalf of Cadence, I'd like to welcome you to today's webinar, Cadence Tech Talk Water Waiting Simulations and One Way FSI with SPH Solver. My name is Jet Villanueva. I'm a field marketer here at Cadence. Before I introduce today's presenter, I would like to go over a few housekeeping items. All attendee lines are muted to prevent background noise. Today's webinar is being recorded, and a link to the recording will be emailed to you in the coming days. There will be a q and a section, segment at the conclusion of today's webinar. If you guys have any questions during the presentation, you may type it into the ask a question section at the bottom right of your screen. We will attempt to answer all questions at the end of the webinar. However, if we run out of time, questions may be answered via email within the next few days. Please ask your questions early so we could get them queued up and ready for the q and a session. For your convenience, there's a Zoom toggle you could access on the bottom right of your screen if needed. Final note, we will control the movement of the slides. Some slides include graphics and may take some time to advance depending on the speed of your connection. Today's presenter is Stelios can Stelios can canopolas working as a lead product engineer here at Cadence. So at this time, I'll turn the event over to Stelios. Floor is yours, man. Thank you, Jeff, for the kind introduction. Hello, everyone, and welcome to today's, webinar. This is the second part, of the online webinars we have, for the SPH simulation. Today's webinar, as Jed said, is, has to do with water weighting simulations and one way, coupling the SPH solver with external structural solvers for FSI simulations. So, let me just, start, just, with a small space statement. But, any, let's say, a feature, future let's say, features that might be, discussed about in this section might not be, okay. Let's say, contrast plans, but, yeah. Let me just start over, right away with this, the outline of the presentation. As we did in, the first webinar, we started the the SPH, with SPH solvers with, the thermal effects in gearbox simulations. And in this second part, we will be focusing on the widening channel simulation. We will see how to define multi body constraints and multi body, the model. There are body sensors that might be, needed for the simulation and coupling, the solver with the external and structural, solver. So starting off, with this model, for those of you, that are not familiar, maybe a wedding channel test is a test that is designed to evaluate the performance and integrity of, the vehicle and certain vehicle components when driven through various steps of water at low speeds. This is a very common practice for automotive, customers. So this is one of the main, let's say, applications that we use, DSP itself solvent for, which can be used, of course, for a large variety of, applications. To start off the simulation, as we did last time, we need to define, the kinetic bodies, that will take part in the simulation. The kinetic bodies, they all have to be, basically all the parts of the model that we want to be included in the VSPH simulation and to the kinetic simulation. In this case, as you can see, on the left hand side, we have, we start from the ground. The ground, of course, is the wading channel in this case. And the wading channel and the ground bodies, in general cannot move, so they are removable objects. And the rest of the kinetic bodies that we define are the moving parts, of, the the car itself, of the internal model that we will be using here. And in this definition, we have the four, the set of, the four wheels that will be used and will be in contact, with the channel. Of course, what we want to define now in the kinetics of this model is to start the the simulation. The model needs to pass through this channel. It needs to, successfully, decline this, slope right here and, get out of the other side. So for that, we need to use, apart from, the kinetic bodies and the motion on them, we need to define certain, let's say, suspension, as which would do in a model like this. So starting, after defining all the kinetic bodies, we need to define, the different parts of the joints that, the different joints that will be, part of this, simulation. Now for this, simulation, we have defined, slider, joints. This slider joints, four of them, of course. One in each wheel, they are used, for the support system. So for the suspension that, we will define later on. And we define these sliders, in relation to the support system and the car itself. So to balance, the body of the car in the support system, that is connected to the wheels as well. The Revolut joints, are imposed directly at the rims and the wheels themselves. The Revolut system is, the joints, are the ones that allow the wheels to rotate on their axis and rotate according to the motion that we will define later on. The wheels now are rigidly connected to the rings, so they take the rings and the the wheels perform exactly the same movement. Additionally, as I mentioned, we need to define suspension for these models. And for this, we have applied the certain magnetic force, which is, the spring number, oscillators. And we apply four different oscillators for, each of the suspension on, each of the wheel, of, the car itself so, that the model is more realistic, of course. Now the final part that we need to have for this, simulation is to apply certain motion. And the motion that we apply in this case, we apply, after calculating, the velocity that we want to achieve and with the radius of the wheel, this can be achieved through our special tool in the, CFD ribbon. You can, calculate, what the rotational speed needs to be, and we apply this RPM rotation, into the RIMS directory in each of the four wheels. Before moving on to the SPH simulation, we just verify that, the kinetic setup is is set up correctly. And as you can see here, we see that the wheels are moving and, the movement, of course, is connected to the rest of the cap that moves along, with the movement of the wheels. Another important aspect that we need to define for this type of simulation, is the sensors that will be used for the simulation. Specifically, for the in order to perform the one way FSI coupling with the SPH solver, we need to define surface sensor, which is the surface pressure force. Now this surface pressure force evaluates, the the pressure forces, from the particles onto the rigid bodies, and it uses a mess to directly, save, the information of the pressure forces that then can be translated into, transient loads for, the natural deck, let's say, that we will be using. So, in order to assess the pressure, for some of this, underbody panel, we have used, a uniform mesh, for the structural simulation of, five millimeters. We have applied it to all of this model, and we load the separate pressure force on the underbody front, rigid body. As I mentioned, there is no need for additional mapping. So the pressure forces that we will get from the SPH simulation can be directly used, in the structural model, with a certain function that I will show you, later. As, it was a case for the, the gearbox simulation, again, in order to, start the simulation, we need to define, an initial fluid, for our simulation. In this case, we will fill the water, the waiting channel, with a set of particles, in order to achieve a height of, the water at 15 centimeters. This relates to approximately 10,000,000 particles for our model that we will be using, five millimeters of, particle radius in order to make the simulation also, run a bit faster. Now the overview of the simulation and the parameters that we use, as I said, five millimeter uniform particle radius, 20, 15 centimeters of the total volume field, in the channel. For the simulation, we have used an NVIDIA a 4,500, graphics card with 24 gigabytes of, video memory. The time time step is three hundred seventy five microseconds, which is calculated automatically, according to the particle values that we select. And this whole simulation took four hours, for five seconds of physical time for the simulation. The structural simulation on the other end needed approximately twenty minutes to, to compute. Now to to see some, results that can be acquired from this. Of course, the waiting, channel experiments and, the waiting channel model can be used, to to measure various different things, on the car body, itself. We can use it to, check if any water gets into the engine, of the air intake of the engine, to test the ceiling of the car, the hydrodynamic forces, to measure, the forces that the water, puts on the water of the vehicle as we will do, in this presentation, evaluate and analyze the splash patterns, and see what where the water goes and assess, how robust and effective the underbody seals are for protecting, any typical parts of the vehicle. Now after, running the SPH simulation, we get the results, directly, from the SPH simulation, gives us, basically, the pressure forces, on the rigid body that we defined previously. And in order to, transfer this information into us to the structural model that we will use later on, we need to, calculate the forces and export these forces, for a structural mess. For this, part, we use the export forces function that can be found underneath, on the functions underneath the fluid simulation window. And this directly, converts, the rigid body model into the structural, model with some of the information needed to to set up the structural simulation. Now all this is done using the the NASDAQ, and we will use, NASDAQ to to set up, the NASDAQ to set up a structural simulation. And with that, we will run, the structural, simulation. All the post processing, that has been done for the, structural simulation can be done entirely in meta, and you can load the results of the structural model and evaluate, let's say, different types of, forces and see any informations that might occur. In this presentation and this webinar, we will not cover the Metapath as most of the functions of the SPH server can be done in answer directly. Now the on the simulation, results after running the structural, simulation, we can, see, the effects that the the water can have onto the model. And we can, as I said, evaluate through meta and create, and see, exactly, where some critical forces might occur that, could, result in failure, in case, this happens. So, this was, the part of, the presentation. Now I will open Anza, to show you some of the things, that we have discussed directly, into the software. Please submit any questions that you might have in, the q and a panel, to cover them in the q and a section at the end of this presentation. So opening, AMSA, can give us, we can use AMSA to directly start, this setup. Now the, kinetic model, on this, has been already set up. So, if we open the database in NAMSA So by opening the database, we can go to the kinetics and see all the different, kinetic bodies that we have defined, starting from the ground, the different, the ways that we will be using, the support systems, and everything. The one of the most important, parts in, the weather widening simulation as they are also simulations that are running two way. So we are coupling, two way the DSPH over with the Kinetics server. So we apply these, forces. So these four forces are used to are used as a suspension model, in this, case. Now by opening, the fluid simulation window, of course, we import the after defining with a volume free tool, defining the initial height that we want, for the, for the, the fluid that we want to, to fill this, waiting channel with. We import it directly with the initial sets, and we use the fluid that we want in this case. We're gonna use one of the default ones, which is water, and we use this one to directly, run the simulation. Now on the sensors, thing that, is quite important, by right clicking on the sensors to add a new sensor, you need to, in this case, as we did, we, use the surface sensor on the pressure forces. And this, is the, let's say, the type of sensor that we are using here to evaluate, this, let's say, pressure forces from, this speed simulation of the structural model. For the moldy body, of course, we are using, as I said, to the two way coupling model. And as we did in the last case, we split into two groups to assign the different collision groups. So one group is the car itself and the other group is the ground in order to correctly define and recognize the contacts between the different parts of the model. Now for the simulation part, as I said, we used, five millimeters of, particle radius, for five seconds in total. And let me load these, results to see, how that looks once we load it. So, whilst this, this is loading, I will check to see if, for any questions. So, case, Saha, you say that on slide 15, they might show the different, SPH public account than the one mentioned. Yeah. This was a draft, let's say, picture that was, we used for the presentation. But the the actual, the actual one that we are using is the one, shown right here. Because if you see on the presentation as well, the resulting particles were not close to 10,000,000. It was, roughly 900,000 something like that. So this was just, as a draft picture to to make it. So loading, another important aspect of the simulation also is the to output, to XMF format, the results. This output, allows you to visualize the to visualize the flow right after the simulation and go through the simulation according to how many results per second you have saved on your model. And, we have also we are planning into the version that will be released either in December or next summer. We want to, allow the user to directly if you have not used any sensors, you can extract, information. Since all the information is there, you can extract service information directly from these results. So in post processing, you can evaluate even more things, without needing, let's say, a setup before, the simulation. So once, we have loaded these results, we can go through the simulation, see how that looks. Let me deactivate the visibility of the model itself beforehand. So as I mentioned, we can see, the there's plus patterns, and anything, like that, from this simulation. At any point, we might want, we can, stop this simulation. We can move, to any state, the simulation. And in the visualization tab, we can change the visibility of, certain, let's say, aspects, both for the fluids and the rigid bodies themselves. So in the appearance of the, phase zero fluid, so the basic fluid that we are using in this simulation, we can change, to one of the different, visualization options as I showed before. So either pick fluid, which basically creates a small surface to the using, a cluster, let's say, of particles, or mesh, which is the more, stable, let's say, or a better visibility, for a fluid like that. It creates a closed surface mess, like an isosurface, around the the vehicle around the, sorry, around the fluid particles and creates, a better looking mixing result. In this case, also, I can use, transparency and I can have lesser thickness to make it, more transparent. And on rigid bodies, something that could be useful as well is, in this case, I can go to my, to the other body panel. I can and I can visualize the pressure forces as well. This is something that is visual at the moment, but this is, this will be used later on. So based from the start of the simulation, I can see, the effect of the the fluid particles onto the body as the simulation progresses. Now once the simulation is complete, and I have, saved all of this, information as I mentioned. Let me just, start the simulation of this pie at this point. So any, the sensor information that, we have run, in this case so all the sensor information, I start directly in the sensor results that can be found underneath sensor definition. So the sensor results are saved similar to the the results output as we did from the flow. It works in a similar way. So you can choose how many results per second you want to save or you could save all the steps of the simulation. But, just keep in mind that, saving all the steps or saving a lot of results per second, it will result in a larger hands on file in the end because you save a lot of more information in the database of, your hands on, model that you are using. So by double clicking on, any sensor results that we have run, we can, see how this looks. So in this case, we have the pressure force information for a different for all the different, let's say, grid points, on our model. And all of these, as the sensor data is stored, in this grid. In this case, I have used also a linear velocity, option. So I have measured the velocity of the car so you can see how the velocity varies, according to the, let's say, the, how it finds, the liquid and the fluid, the water itself. It makes it harder for the car to move. So, it drops a bit velocity and pretty much it's being kept around four meters per second. Now once, I have used these results, as I said under the fluid simulation window, we can use the export forces button. And with the export forces button, it's a pretty, straightforward window. Sorry for this. So it's a pretty straightforward window where I choose the name of the model that I want to export. I select the bodies, the rigid bodies that took part in the sensor evaluation. In this case, it's the body number 21, the underbody front, model body. I need to select the results file, that I want to extract this information from because, as you can understand here, I can run multiple, SPH simulation, and they can be stored under the same ANZAR database in different simulate DSPH simulation window. They can be stored, under the sensor results. So for each case, if I have multiple, I need to select the correct one. In this case, I have just the one for five seconds. So I select this one. And also, I can select the phase, but in this case, I have just phase zero. I do not have any other option. So by pressing on the export button, this, gets the information in the NASDAQ. Of course, here, since I'm using the FidelityFlow, deck, I need to switch to the Nafran deck in order for this, function to work. So under the deck, I switch to Nafran. I now export. And you see, I get I can wait a couple of seconds, and I get the the result that I want. So once this is, finalized and I get that, everything is completed, I can close the, speed simulation window. And in this model now, I have, two models. And the one that is active is a structural model. So I will press the show only button on this, enable the visibility, and I can save this model, by since it's used as set active, and it's the only one that is active at the moment in Anza, I can, use the save as option to save it into a different, file. Now by saving this, file, I'm directly opening it in another Amazon. And the information that is already, in the database of this again, let me switch to the last one deck. The information that I get directly from this is, the t loads. So the transient load, that is saved in its grid grid in its grid point, from the SPF simulation, will be used, for the structural simulation in order to evaluate, the structural forces that are applied on this model. Some, preprocessing needs to be done in this case before, starting the simulation. Of course, we need to switch to the analysis, structural deck. And, apart from the d below deload, we need to, some preprocessing needs. We need to define the deload, the boundary conditions, the material that we want for the simulation, And and National Endowment header, of course, and a time step for the simulation that we want, according to the time step and the total, iterations that we want to, to use for this simulation. In the National Endowment header, of course, we need to define it, so that it creates, the transient, case. So we will use Sol, one online to simulate the transient loads of this case. And a lot of times, this is something that can be done, of course, by a a structural engineer to, let's say, most most probably more probably, calculate this. The advantage of, using the answer SVH solver is that you do not need to, use the results mapping, as it is done in other solvers. So you directly get the, the pressure, and the forces that you want for the structural model, onto your, model. So once, this is the preprocessing is done for the structural simulation, we can use, the solve option, in Nustrand. This can be, pretty, pretty, pretty easy. If you just need to define a new model, you need to define the main file, the results directory, and everything, and, you can run it, like this and do all the post processing later on in meta. So that, would conclude the, presentation and to the, small demo that we have in, in today's webinar. We can, I think that we can go to the, q and a section now? I think we have some questions already. Perfect. Yeah. Go ahead. Go into the q and a. Folks, if you guys would like to ask a question, please submit the questions into the q and a panel, and we'll get to it. Okay. So, so for through that, let me go through some of the questions. So can the forces also case, how you ask, if can the forces be exported to any other solver like, Ansys Mechanical? At the moment, the only tech that we can use, for the export forces button is the NASDAQ. We plan to support, Abacus in the future, but, this is something that is still under development. So, it works, at the moment just for, Nastran. Any other post processing, I guess, you can do, but you will have to do it manually to get the information from Nasrond and, maybe change it to a different kind of solver. So to fund, you're asking, are there simpler examples or tutorials similar to this study available for, customers? Yes. So, at the moment in, the answer documentation and, the answer, let's say, library, we have some tutorials, which are about to be, let's say, updated to the most recent version. This, we are also planning to have, some, let's say, rapid adoption kit for the solver. So you can have some video, examples for each of the different applications of, yeah, of the solver. And one of that, of course, is, this application. So, Nathan, you're asking how a heat exchanger typically modeled, and has Cadence contacted any studies to evaluate, the impact on undercooked thermal, wetting. So, heat exchangers, in the speed solver, you would mean, I guess, to add also thermal equations and thermal, solver, if I'm not mistaken. Right? So if that is the case, so if that is the case, in the SPF solver currently, you can, you can use the SPF solver with the thermal option as we showed in the the first webinar with the, gearbox simulation. Okay. But yeah. The thermal option is something that you can use, but at the moment, it works just, from, transferring, let's say, heat from, the surfaces that is bodies to the fluid and within the fluid itself. We are planning for next year to support the CHT analysis as well. And, yes, this is, something that we, that we would want to have, also have rigid body heat transfer, for that to evaluate even, let's say, cooling of transmission or any part of the the car. So for the for the wetting, you're correcting, the under hood wetting. We do not have any study, exactly for that, but this is something that we could simulate with the SPEIT solver. So you could simulate, of course, you would need a higher height depending on your model. You would need to have more more fluid to to be able to reach, to the top of the underhood. Yeah. But this is something that is possible, with this page, solver. It's okay. So how so, the first, for any of you that, have missed the first webinar, there is a recording and you can request, to access that, by, through our webinar, page, I guess. So, that is, your question regarding the the licensing. The only difference, that has been in the recent Now because we have released a new type of licenses, which is the Empire license. So in with this license, you can use, DSPH solver with a uniform, let's say, credits. So you have a credit pool when you can use different parts, that are included into the MKI packets. So one of them is the SPH solver. So you could use the SPH solver with the, let's say, uniform with the MKI package, license. So you can use, directly, 200 credits of SPH, for SPH but using the uniform credit pool. Also, Kishaha, you are asking if you we can do leakage analysis, using the SPH solven. Yes. This is something that, it can be done. Of course, you need to drop a lot to to the, the particle radius. It needs to be, the particles need to be small enough to fit through any gaps, and you can evaluate you can evaluate that. A way that you can do that, more successfully, would be the adaptive, particle sizing, which is a feature that we have implemented into the latest version, the two thousand twenty six six dot one, for the SPH solver. But at the moment, it mostly works, for a a waiting simulations. This is a feature that is, the first time that has been implemented. So it will, require additional, let's say, corrections and it will require, more development from the developers, side in order to, to be more robust. But leakage could be, faster done by that. But even with uniform, particularities, you can do it. So, Tufan, you're asking if the the speed of the car, is a limit in this case, or can high speeds be simulated by using higher number of particles or smaller tantrums? No. The the speed of the car, it's not something that is done, and the the slow speeds in this case, it's not something that we do, for stability, let's say, reasons. So it's something that, because mostly the the speeds that are used, with some of our customers for these types of applications is, to the range of that. So at low speed, they simulate. So you can, increase the the the speed, then you can use also high speeds into this, and simulate it, directly. Of course, by increasing the speed, you would use a bit the the time step because the the automatic, calculation of the time step is, some is an equation that uses both, the particle radius, but also the highest, velocity of, that you are using for the particles themselves. So if you expect to have higher velocities, you would increase that and decrease the time stamp. But for sure, this is something that, you could do as well to increase the speed of the the car. So, Remy, you're asking, that, you said that you are interested in, SPH hydraulic configurations. If you can use it to model Pelton turbine flows. So it's not part of this webinar. I cannot show you anything at the moment. If you're interested, we could, have a arrange something. But, you can use the SPF solver for that application. And I actually have a model, that recently we have run, two way coupling. So we have run just, the flow that gets that hits, basically, the Peloton turbine and how this moves, the whole kinetic body. Yeah. This is something that you could simulate. Again, the only thing due to the high speed of, the velocity in, the high speed of, the flows in this case is you would need to to adjust the time step to be quite smaller. But, yeah, by selecting a higher max number and velocity, you can, calculate, let's say, the required one from the server directly, into the server. So, Keeshaha, yes. The, you're asking if the time step can be done dynamically for stability. So we have the adaptive option for the, we have the adaptive options for the time steps on the simulation. You can use an adaptive time step, and this allows the solver, to it uses, of course, a a minimum time step, and then it recalculates according to, the needs of how, it calculates, let's say, the speeds, the the velocities of the the particles and readjust the time step accordingly. This can, increase, or, let's say, allow you to run a bit faster in some case. But it gives you more accurate results, let's say, in high speed, flows. I'm not seeing okay. So let me, dispute solver. So we're asking if it runs also on CPU or just GPU. The, this bit solver, it, can run, it mostly runs in GPU and, more or less any GPU that you might have. Of course, the the newest, let's say, the hardware that you are using, the newest that the the hardware is, the faster the simulations will be. It has the possibility of running on CPU, but, it's not something that we use. But, it's, let's say, it's GPU accelerated, so it's, running, almost complete in GPU. I'm not seeing any more questions, if I'm not mistaken. I think we covered, have answered all of the ones that, have come already. Okay. Alright. Well, we can go ahead and end it there. Thank you so much, Stelios, Yeah. for your presentation and answering all the questions. Ladies and gentlemen, thank you guys so much for joining us today. That concludes today's, webinar. Please take a quick minute, and complete the short survey. On behalf of Cadence. I would like to thank you guys for joining us today, and we'll see you next time. Again, thank you so much, Stelios, for your presentation. Yeah. Thank you, Jit, and thanks for everyone, for attending, and see you on the next, webinar. Alright. Take it easy, everybody. Yeah. Have a great day, everyone. Bye bye.