Creo Flow Analysis - Multiphase CFD in Creo

Multiphase CFD in Creo — Webcast Transcript
 

Welcome and Speaker Introductions

Dana (PTC, Moderator): Good morning and good afternoon, everyone. Welcome to the Multiphase CFD in Creo webcast. My name is Dana, and I'll be moderating today's webcast as well as recording it for replay purposes. On the call with us today, we have Todd Craft and Tom Quaglia from PTC, and Alex Zhang from Simerics.

If you experience technical difficulties, you can contact me via chat. If you'd like to ask a question, you can use the Q&A box on your control panel, and feel free to submit them at any time. Now I'll turn the call over to Todd Craft. Todd, it's all yours.

Todd Craft (PTC): Thank you, Dana. Welcome everyone to our session here on Multiphase CFD in Creo. Let's go ahead and get started.

Creo Flow Analysis Overview

Packages, Support, and Training

First of all, if any of you are not familiar with Creo Flow Analysis, here's a quick slide I show to most people before we start. It's really analyst-level functionality of CFD inside of Creo. There are three levels: a base, an advanced, and a premium. Today we're going to be talking about multiphase, which is part of the premium package. There's really not a lot of limits to what we can't do with Creo Flow Analysis. It's a partner product developed by Simerics, and they've integrated it into Creo to make really high-performance, high-capability — and even low-capability — CFD, integrated into Creo.

This is the eighth presentation, or webinar, we've done. Down below you can see there are eight of them — actually, I think there are nine in that list because there was one that was done. There's also the how-to series. All these are located in the PTC Community. So if you go to the PTC Community and type in "CFD," "CFD 101," or "CFD for all," you'll find all these recordings, and sometimes you'll find some tutorial models as well.

Just to remind everybody, we're in Creo 10 currently. In Creo 9 we did a ton of development for Creo Flow Analysis — multi-project, usability improvements, behavioral modeling, gyroids, post-processing. I sort of call this Creo Flow Analysis 2.0, because it's been available since Creo 4 — we back-ported it there — but Creo 9 was a very significant release.

I want to make sure that everybody's aware there's unlimited support from Simerics to all the customers. There's also custom training. If you want to take your company through your own examples and train a bunch of people at one time, Simerics will do that. They do that a lot, actually.

What is Multiphase CFD?

Volume of Fluid, Phases, and Free Surface Flow

So today we're going to talk about multiphase. Multiphase is part of the premium package. You use it in mixing — typically gas and liquid, or liquid and liquid together that are different.

Some of these slides are from Simerics, so I'm going to read some of them. The Creo Flow Analysis premium package has multiphase flow, which also includes heat transfer and a volume of fluid approach. There are a couple of other approaches that can be done based on what you're doing — both implicit and explicit approaches. They're very fast, very robust, with excellent mass and volume conservation for each phase.

Comprehensive physics: when you do run multiphase flow, you can add in heat transfer if you want to, surface tension, and liquid-to-solid surface contact angles. It's also extremely accurate. Whenever we run Creo Flow Analysis, you can be assured that you're getting analyst-level, prototype-level results — excellent correlation with test data over the full operating range, always with excellent mass, volume, and energy conservation. You're always going to get that fast meshing and fast simulation speed based on the proprietary numerical algorithm, with significantly faster runtime and faster convergence compared to our competitors.

And lastly, a robust process. This is always the case whenever you're defining things in Creo Flow Analysis. We have a very robust, very powerful mesher, and it understands the physics that's applied and does a lot of the work for you. It conforms the mesh and gives you results that you'd be expecting.

When we talk about multiphase, here are a few examples. We have a gear housing over there with some oil sloshing. Then we have some open flow ones here with a pump, where we're moving the liquid around. Again, we're combining air and liquid in a lot of these cases. Then we have a flushing example over here that combines air and water, which is exactly what multiphase does.

If you think about multiphase, it's really many flows that have more than one fluid present — different substances like oil and water or air and methane, and different phases (like water and ice, for example, of the same substance). But the question is: how are they mixed? If it's at a molecular level, that would be considered multispecies, which is part of the advanced package of Creo Flow Analysis — gases like air and methane, for example. But if it's a macroscopic type of mixture that's very visible to the eye, that would be considered multiphase. So it has identifiable boundaries between the two phases.

Gas and liquid or liquid and liquid are the most common — water and air is probably the most common. There are also instances of gas and solid or liquid and solid; those are also possible.

Some common examples are a fuel tank sloshing and overall free surface flow, where you have water or a liquid moving and it's taking up the area where air is — so they're mixing together. Other examples include a sprayer, filling of tanks or containers with various liquids, mixing vessels, oil and gas extraction, separators, pumps, dryers, evaporators, food production, and boiling and condensation, and many more. Whenever there are multiple phases or multiple substances — or similar substances — that's really what this is for.

Just to comment, going back five years being a product manager for Creo Flow Analysis, a lot of people said multiphase is too complicated, it's too difficult, only analysts should be doing it. But what we've proven is that Creo Flow Analysis and our customer base is very capable of solving for this, and it's a very common use case we're finding. You don't have to have any specialized education or training to do this. You should know how to do it, obviously, but we're finding that a lot of customers are very capable of doing multiphase inside of Creo.

Real-World Applications

Bioink 3D Printing and Semiconductor Reactor Shower Heads

Something that Creo Flow Analysis and Simerics have been involved in is the printing of bioink. This is actually a very complicated procedure, and there are some images here of those results.

Another very complicated thing Simerics has been able to do is around the semiconductor reactor shower head. You have flow, you have heat, you have mixing. Here are a few examples of what that looks like.

I promised Alex I would introduce this topic for about ten minutes, and that's where we are. So I'm going to now turn it over to Alex. Alex Zhang is really the expert at Simerics. He's an application engineer. He's going to run you through a bunch of examples.

What I want you to do during this session is fill the chat with questions while Alex is performing those demos and walking you through the product. That would be great. I'll stop Alex if I have to, or if I can type in the answers myself, I'll go ahead and do that in the chat. But please feel free to ask questions. This is really the part where Alex is going to show you inside of Creo how to do a multiphase simulation. So Alex, if you could take control of the screen, I'll watch the chat box while you're presenting.

Live Demo: Shower Head Model and Fluid Domain Prep

Alex Zhang (Simerics): Sounds great. Really appreciate it, Todd. Let's go ahead and hop right into Creo and take a look at a full workflow, start to finish, implementing the multiphase module in Creo Flow Analysis.

As far as multiphase goes, some major industries that use it are consumer appliances (like you saw in Todd's slides), automotive, and aerospace — really anywhere you're modeling any kind of fluid, so liquid or gases.

In this case, we're looking at the example of a shower head. Of course, we want to model how the water comes out of the shower head and then how it interacts with the surrounding air. First, we'll start off with our Creo model. We can go ahead and take a look here.

Nowadays, basically every shower head has an adjustable setting where you can have it in three modes: a jet mode, an overall mode, and a combination between the two. In this case, we're going to model the jet mode, where this disc here is rotated so that only the water travels through these channels here.

I went ahead and created the fluid volume. You can use Creo Flow Analysis to extract the fluid volume, or if you're more savvy on the Creo Parametric side, as most users are, you can actually go ahead and create this by hand. This can be done very easily using some revolve features. You don't have to be precise in where the fluid domain is because then we can use Boolean operations to cut off the hanging end. Something like this I was able to create in two minutes, and we'll go ahead and use it for our simulation.

Simulation Setup Wizard

Multiphase, Gravity, and the Fluids Material Library

To set up our simulation, first we need to launch Creo Flow Analysis, and that's as easy as going to Applications > Flow Analysis. You can see all we do is we add a ribbon and we add a tree, but we're still here in Creo Parametric. If we flip back to the model tree, it's all still here. You can think of it as almost like an overlay where we add in the digital fluids to your digital prototype.

For first-time users, intermittent users, or users who prefer a more structured workflow, we have this wizard workflow. What it does is break down all the steps and properties you need to set up your analysis, and lays it out in a nice, simple, guided workflow. We're going to walk through this together. As Todd mentioned, we are using the premium package, so we're going to be modeling fluid flows, turbulence, and multiphase.

With multiphase — to get a little more into the numerics — what we're doing is tracking the fractions of however many fluids you have in the simulation through each and every cell, and we're also tracking the free surface interface between the two. Multiphase has the assumption that the components are immiscible. You can think of it as oil and water: you can mix it, but they're not actually going to be uniformly blended. They're still going to be their own separate chemical components.

You see here, the wizard will update as we make selections. In this case, we chose multiphase, so we just want to identify what components we're modeling. In this case, it's going to be air and water. Then we have basically every model you could ask for, for multiphase — implicit or explicit modeling, as well as various time schemes. I'll say that the default here, using the explicit method, which is a Runge-Kutta second order, is pretty universal as far as what you'll want.

We do want to model gravity here, and that is based off the assembly origin. You're going to want to orient everything to that, and we do repaint it here for convenience. Our gravity direction is in the negative Y direction, so we're going to set it here. The reason we take it as a vector like this is that we could also have gravity change with respect to time. This is a way to model movement, translation, turning, or changes in incline. Of course, not particularly relevant to a shower head, but for the example of a fuel tank in a vehicle, you can actually model any maneuvers that you need to see, using an acceleration curve input here.

We don't need to create the fluid domains, although if you do prefer Creo Flow Analysis to automatically extract this for you, you would just select the Internal option here, select your inlets and outlets, and it would actually generate the exact same geometry. I just prefer to do this by hand because it gives me a little more control over what I want to include or not include.

Now we're selecting our domains. With Creo Flow Analysis, especially if you use Windchill or some other PLM solution, you don't want to be messing with the master model too much — like deleting or suppressing extra components that don't matter to the simulation. Instead, we can just pick whichever ones we actually want to use without having to affect anything else in the model.

We've got our fluid bodies; we're just going to apply air and water from the fluids material library. This is built into Creo Flow Analysis. We have the major liquids and gases in here. If there's anything you need that you don't have, we can just add it as a custom material. You can see the types of inputs we take. These are the default ones — so density and viscosity are always needed, and of course you can add the thermal properties as well if that's relevant.

On top of constant density and compressible gases and liquids, we also support non-Newtonian fluids, pressure-dependent properties, pressure- and temperature-dependent properties, and all the way to an equation of state — so real gas modeling from these tables, basically the full spectrum as far as what you would want to model in CFD.

Boundary Conditions, Surface Tension, and Initial Conditions

Now we'll set up our boundary conditions. Tell us what you know at the beginning and end, maybe some stuff in the middle, and then the solver takes care of the rest.

We're going to go ahead and first select this surface as our inlet boundary. Industry standard is to have these things running at 1.8 gallons per minute. I'm going to set a specified volumetric flow rate of 1.8 gallons per minute, and we're going to say that it's all water coming through this boundary and no air. So we're assuming there's no air entrants anywhere upstream.

For our outlet, that's just going to be the outer surfaces of the cylinder. I created this cylinder to represent the ambient environment of the room, basically. Select these boundaries, and we'll just say zero pressure boundaries. We actually have gauge pressure enabled, so that's already good to go. And then, of course, these outer boundaries start out as air, no water.

At least in this case, we're about done with the setup. No other boundaries to apply boundary conditions to. We just want to make sure that our initial condition is correct, so that's going to be set in volume parameters. I'm going to go to Initial Conditions, and we're going to start it completely empty, so all air, zero water.

We can also incorporate surface tension. Of course, water has one of the highest surface tensions, so it's definitely a place to consider adding it if need be.

Pseudo Steady-State Approach and Meshing Best Practices

Body-Fitted Mesh

Creo Flow Analysis does support a pseudo steady-state simulation for multiphase. Typically, you'll want to model it in transient, but in the case that you're looking at things like spray — typically the primary one where there are no moving components and it's all static geometry — we can actually use a pseudo steady-state to get a solution much faster, with very reasonable accuracy compared to a full-on transient time-dependent run. When you're looking for your final results, you do want to use a transient simulation. But we can see how we leverage this initial pseudo steady-state to make the entire process better.

Continuing on, now we're on to meshing. With meshing, as you saw from the past slides, we use a body-fitted mesh. Your primary controls are going to be the largest the cell can be and the smallest the cell can be. What I recommend for multiphase is having the max cell size in cells and on surfaces be the same value, as that gives us a uniform mesh throughout the entire model. The uniform mesh is fairly important because we are tracking the fractions of components between each and every cell. A nice uniform mesh gives us a nice uniform result.

I'll go ahead and start up the mesh here.

Q&A During Meshing

Todd Craft (PTC): Quick question, Alex. There are a few questions in the chat, but one easy one.

Q (from chat): Is the wall condition adjustable — like wall roughness?

A — Alex Zhang: Yes, it is. You can specify for any particular boundary, or any subset of that boundary, a specific roughness height. For the roughness height, you can use the sand roughness chart as a reference.

Todd Craft (PTC): Okay. And then while it's solving — or do you want to show the results while it's going? That's still meshing. We've got time for another question. This is sort of complicated, but:

Q (from chat): Can you use Creo Flow Analysis where it's impacted by corners and turbulence?

A — Alex Zhang: Yes, absolutely. We would not be a competitive CFD package if we could not.

Q (from chat): And what about angular spray and multiple outputs?

A — Alex Zhang: Yes, absolutely. And actually, in this case, we are doing the spray cone for this shower head.

Running the Simulation and Live Isosurface Visualization

Alex Zhang (Simerics): Excellent questions — please keep them coming. In the meantime, we'll go ahead and wrap up the setup here and start looking at some of the results.

In this case, what I'm doing is creating a pretty small simulation domain, because I'm just interested in the initial spray cone of each of the nozzles. We want to make sure there's a reasonable balance. Of course, it's always a trade-off in design between balance and modularity — in this case, being able to change between three different modes. We just want to make sure our jets are relatively even, that they're all going in relatively the same direction. We also want to see the pressure drop that is required to run this design. So we're getting some very macroscopic ideas of what the design is, so we can go ahead — really early on in the design process — and iron out those larger issues that will come up.

We know the pressure at the outlet; that's just going to be atmospheric pressure. So at the inlet, we're going to output our pressure. We know the flow rate, and that'll give us the restriction. Another exercise we can do — I'll go ahead and turn on these outputs and explain more in just a moment. Go ahead and turn on the volumetric flow rate for the water as well.

Excellent. Let's go ahead and start the simulation, and we can view the results right off the bat. One big advantage of Creo Flow Analysis is that you can render the results as the simulation runs. So there's no need to wait for your simulation to finish running and then load back your results in a separate post-processing suite. It's all integrated into the same workspace.

What we're looking at here is an iso surface. What the iso surface does is track some kind of criteria. In this case, we're asking it to look for every cell that has above 1% water in it, and it'll render it on the screen like you see here. We can actually see how the water fills our shower head. Because we have the channel here, of course this one's going to fill up before this one. We can see how that may or may not affect the symmetricality of our model — having the offset can cause some recirculation, some low- versus high-pressure zones that can really offset our flow. We can see, like here on the right side, it's filling a little faster. Of course, end of the day, all of this happens within like 0.2 seconds, so we're just splitting hairs there.

What we do want to see is the final result. We can just watch as our jets develop. For a very straightforward exercise like this, fifteen minutes start to finish, we can already get a lot of information. We can get the ideas of our fill characteristics in the system, as well as a good idea of what the water profile is going to look like coming inside. Actually, it looks pretty good to me. Of course, we're going to have a little bit of an off balance here on the left, again because of the offset, but that's also due to the modularity we need to bake into this design — having three different flow times.

Mass Conservation, Diffusion Checks, and Hardware Requirements

Laptop Performance and Parallelization

In case you're curious, this is running on my laptop. Let me bring up my Task Manager. It's one of those Dell Precision laptops — eight-core processor, 32 GB of RAM, and then a physical GPU, and the GPU is only used for the post-processing here. The idea is that any Creo workstation can run Creo Flow Analysis. Every license of Creo Flow Analysis comes with up to eight physical cores. If need be, you can also license additional core usage. Our parallelization is excellent and scales almost linearly with the number of cores. You can imagine if I had 32 cores, 64, or 128, we would be getting these results very, very quickly.

Our simulation is done. We can look at the macroscopic results — we get a good idea of the spray profile. Looks pretty solid, actually. We can come over here and look at our pressure drop. So, 47 PSI pressure drop, from the inlet to the outlet as we have 0 PSI here. Just like that, we can already evaluate: is that too high of a requirement, too low? We can go in and start changing diameters and the like, and really play with the model to get what we're looking for.

Diffusion and Mass Conservation Checks

Of course, we do want to make sure we're checking certain things when you're running the simulation. The first one is something called diffusion. Let's go ahead and take a look at our section view, and I'll paint the mesh on here. Because we have the smaller mesh domain, we can actually use a finer mesh throughout, simplifying the usage process — a few less things to think about. As far as rules of thumb, you want at least three cells through your thinnest flow channel, and we accomplish that very, very well here through the nozzles.

What diffusion is, is when we look at the volume fraction of water through here, and then maybe tighten up the range a bit. You can imagine, as we're tracing the water and the air through each of the cells, if the cells aren't fine enough then some of that water will disappear. You can imagine jumping from a small cell to a big cell — tracking 0.5 — then all of a sudden that fraction becomes less meaningful as it comes up. So one thing to look for is just making sure that the amount of water going into the system is the same amount coming out. That's why I turned on the volumetric flow rate for the water at the inlets and the outlets.

What we want here, as an industry standard, is typically within 5% — that's very reasonable. We're looking at a difference of about one cubic centimeter per minute, which is way less than 1%. This is well within the acceptable range as far as mass or volumetric conservation. We're doing constant density, so mass conservation is preserved here.

That's the strength of our solver, mesher, and our multiphase VOF implementation. Of course, it's easy to list off that you have this, this, and this, but the hard part — and why CFD only has a few solutions out there realistically — is the importance of how those work together. That's a much more complicated question. That being said, we have massive advantages in this space. As you see: excellent mass conservation, easy and fast meshing, and very fast and efficient solve. Again, I'm doing this on a laptop, and it's not even a very high-end laptop.

Todd Craft (PTC): You've had it for over a year now, haven't you?

Alex Zhang (Simerics): That's right.

Todd Craft (PTC): And even when we bought it, not the highest end either. It's maybe better than the average, but it's still not something special — not something you wouldn't find off the shelf.

Alex Zhang (Simerics): Absolutely. The important thing here is the clock speed on your processor. The median is around 2.4 to 3 GHz. Mine, for some reason, goes up to 4, which is nice. But you can think — if we scale that linearly, something that takes me five minutes might at worst take you eight or ten minutes. So it's not that big of a difference, realistically speaking. Any CPU can run Creo Flow Analysis. Of course, the more cores you have the better. But even if you have a two-core machine, it'll take a while to solve but it will still scale to that machine. So there's really no barrier of entry to Creo Flow Analysis.

Q&A: Wall Functions, Benchmarks, and Optimization

Behavioral Modeling Extension

Todd Craft (PTC): One question that was asked, Alex, was about droplet breakup handling. Do you want to quickly answer, or is that even a quick answer? How does Creo Flow Analysis handle droplet breakup?

Alex Zhang (Simerics): For sure. And actually, I do have something coming up on that, so hold off just a bit.

Todd Craft (PTC): One more question on this demo:

Q (from chat): The boundary layer mesh is not included. Does the tool provide control with boundary layer meshing and local refinement?

A — Alex Zhang: Sure. For our solver, mesher, and physics, we do not require an explicit boundary layer mesh. We actually use wall functions to solve for the boundary layer on walls. Of course, that goes against conventional understandings of how CFD should be implemented, but those conventional understandings come from decades in the past. One big advantage of Simerics as a company is that we started thirteen years ago, so we got to understand the industry as it was at the time, plus all the advancements you can imagine. In engineering, we're supposed to be at the bleeding edge. Decades of advancement — of course there's going to be some big differences down the road that we should be embracing.

Boundary layer meshing intuitively makes sense. Our wall functions work extremely well — extremely well-validated against thousands of cases. We have numerous white papers and benchmark validations that we're always happy to share. Our philosophy here is, instead of selling you on a checklist, our philosophy is "seeing is believing." We're always happy to run a benchmark or validation, or point you to reference materials — whatever you need to be confident in Creo Flow Analysis so that you can begin succeeding with Creo Flow Analysis as soon as possible.

Todd Craft (PTC): One last question:

Q (from chat): Do we have any published comparison benchmarks between other applications that are out there, either free ones or commercial ones?

A — Alex Zhang: Yeah, we have quite a few of those. So if you want a particular application space, let us know and we can send you a paper or papers directly in that area. On top of multiphase, we do everything else.

Q (from chat): And with this demo, do you have anything to optimize the path?

A — Alex Zhang: Absolutely — or at least I can speak high level to that. We do have an integration with the Behavioral Modeling Extension in Creo Parametric. With Creo Flow Analysis, we can output basically anything you want. If you're optimizing around the pressure drop or the flow rates, you can output those from any boundary of interest, so whatever might be driving your simulation. In this case, if we want to reduce the pressure drop, then that would output the pressure on the inlet. We'd add this as a parameter that we can use to drive a sensitivity study, if you're interested in that. We can actually map — you can choose a particular dimension in the model and use that to set a range that affects it, or you can let the computer do most of the work. We can say to minimize the pressure, and we can use any number of dimensions in the model to run this.

Todd Craft (PTC): The key is having a robust model that can be changed, with everything updating. But it will work combining with the Behavioral Modeling of Creo.

Alex Zhang (Simerics): Yep.

Todd Craft (PTC): Okay. I think you had one more demo you wanted to show.

Alex Zhang (Simerics): We'll actually continue on here, and then have a few examples for other applications of multiphase.

Todd Craft (PTC): Are you looking at the chat? There are a few more questions there. Let's see here, the last two.

Hey, Jasheen — really appreciate it. Looks like you're helping us out with the questions.

Alex Zhang (Simerics): There's a question about keeping Y-plus at a certain range. It all comes down to implementation and models and, as Jasheen said, wall treatments. In this case, we really don't worry about that at all. Again, "seeing is believing." If you want to see a benchmark, validation, or white paper, happy to share that with you.

For the question about the turbulence model, we are modeling this with turbulence actually right now, with the standard k-epsilon model.

Transient Simulation and Courant Number Tuning

Let's go ahead and wrap up the demo here. Creo Flow Analysis is designed for usage by all levels of engineering and all levels of experience. What we're looking at here is a very easy-to-replicate and set-up simulation that anyone can do — from designers, engineers, to even dedicated analysts. Of course, you can always tier that up. That's the beauty of Creo Flow Analysis: it works for upfront simple simulations all the way to your final validation analyses.

One other thing we'll want to do, if we're looking at that final validation, is use a transient simulation, as this will be inherently more accurate overall. What we're doing is running a time-dependent simulation where we're running a series of simulations that march forward in time. In this case, we're running every one five-hundredth of a second, the simulation, and then running one right after that builds off that solution.

Let me go ahead and start off here. This is the really valuable part of being able to run pseudo steady-state: we can actually use this pseudo steady-state simulation — the results here — as an initial condition for the transient simulation. We already save a lot of solve time and processing power. We can see here how the free surface of the jets themselves are actually resolving themselves even more finely. You can actually see as the jets compress here — that's because transient is going to be more accurate.

If we look at the pressure drop, it's not changing too much actually. So again, looking at those macroscopic values, the pseudo steady-state is perfectly fine. Once you want to get into a little more detail, you can see here it's just a one-click switch to transient.

Courant Number

There is one more thing that comes into play once you switch over to a transient simulation, and that's the Courant number. The Courant number is a measure of — you can think of the individual fluid particles moving through the system — how much time each particle of water spends in each cell. The idea is the perfect value is a value of one or lower, where the water particles are moving one cell at a time, which as you can imagine is ideal.

That being said, there is always a trade-off between maintaining that level of one and the practicality of running a simulation like that. The industry standard, as far as what's acceptable, is going to be 50. You want to make sure that all the regions that matter are 50 or below.

We can see here, like this gray blob — again, this is the iso surface — for things like these channels where it's already filled in with water, we're not really too worried about that because there's no real interaction there between the air and water. But we do want to worry here at the nozzles. With a smaller diameter, it's going to a higher velocity, so of course it's going to be a little trickier to resolve, or going to need a little more to resolve.

What I'll do here, with this iso surface up, is come up to our transient setup and increase the number of time steps. We can start by doubling it, and we can again track our iso surface Courant number. We can see doubling it highly reduced it, but it's still not quite where we want it to be. You can see how quickly and iteratively you can do this — we'll just double it again. The entire time, we're watching the free surface of the jets, making sure that doesn't change too much as we modify the Courant number. Now we can see it's down to just a few rogue cells. This is where you kind of make a judgment call where it matters or doesn't. In my experience, at this point, this is very, very good. Of course, if we had unlimited computational resources, we could make it even finer.

I want to use this also as a general reference for those of you using multiphase today or looking at using it. If you have all these bases covered, then we can go from a simple upfront simulation like we see here on the screen — which actually gets us already very far — and we can see the jets very highly resolved. We can see also where we have some crossover here now. We can take that all the way back to the off state — we can see it's actually perfectly diagonal to that off state, which makes sense.

Full Test Stand Results and Industry Applications

Pump Priming, Radiator Filling, and E-Motor Cooling

Engineering is all about trade-offs. We can take all these principles that we just covered here in this example, make this cylinder larger, and we can do a full-on simulation. An industry-standard test stand for a shower head like this is to have a plate perpendicular to the flow about three to four feet away from the shower head itself. You can see here we have the coarse mesh out here, and then a fine mesh along the path of the jets. That allows us to maintain the volume fractions throughout, and then we would verify that by looking at the mass balance of the water — coming from the inlet and going through the outlet.

Using a mesh like this, you can see the level of detail we can achieve. We can see the droplets shearing off of the jet as it approaches the plate. Of course, we can see the splashing as it hits the plate as well. Nice, sharp, clean interface, very high-detail results, very little diffusion as well. Just another angle of the results — you can see again the level of detail you can achieve. The setup is everything you just saw, just extending that out a little more depending on your objectives.

Creo Flow Analysis is a full-blown, analyst-quality tool that is usable by all experience levels and can also cover any use case you might need. It's a tool that you can grow with.

Todd Craft (PTC): There was one more question, Alex:

Q (from chat): Do you see any differences in the final answer when k-epsilon or k-omega are used?

A — Alex Zhang: Not sure if the full context is there for me to fully answer that. It's a little hard to say, but I will say, at least between k-epsilon and k-omega, there really shouldn't be much of a difference.

Todd Craft (PTC): Is this the last demo portion, Alex?

Alex Zhang (Simerics): That'll be the demo. I do have a few other examples here we can look at.

Todd Craft (PTC): High level?

Alex Zhang (Simerics): Sure, other applications.

Liquid Systems: Pump Priming and Radiator Filling

Liquid systems in general — hydraulics, coolant tanks, coolant systems, fuel tanks — all that good stuff is covered by multiphase. You can see a pump priming example here, where we just want to prime the pump and make sure we get all the air bubbles out of the system.

We also have the filling of a radiator here. This is a pretty cool animation. You can see a really large aspect ratio model, with very large aspect ratio differences as well. We have these larger tubes that flow into these smaller channels, but you can see the level of detail here. You can see the water even bubbling up, splashing as it drops down the tubes and as it fills in the system. Very popular application among our automotive customers, as you can imagine.

Validation: Dam Break and Free Surface Flows

This is very well validated. Here are some fundamental validations I want to share with you — for example, this dam break scenario. You can see the actual lab test here and the comparison of our results against the high-speed imaging. You can see how you can actually do these prototypes digitally, from the comfort of your desk.

Of course, the advantages of CFD don't stop there. With lab testing, you have to build the prototype, set it up on the test bench, calibrate your devices, and even then each of your devices is only monitoring one property at one location. With CFD, you're getting every property at every location — pressure, velocity, temperature — anywhere you like. That's where you can see a lot of advantages, both in what information you get from CFD as well as the time savings. Nowadays, we're all very good at prototyping, especially with additive manufacturing — it's been in our design processes for a long time. But running and operating the prototype, having to go through revisions, calibrating devices, measuring them, interpreting those results — it takes a lot of time that you can save quite a bit of by running CFD cycles within your design cycles.

Free surface flows — this kind of open channel flow — very popular application as well. You can see it again, compared against the measurements point by point. Very good, and a very sharp and natural interface, no stair-stepping.

E-Motor Cooling (GM Validation)

The last example I want to show — we've been looking at a lot of lower velocity applications, so I want to share one that's also high velocity, so we cover the full spectrum of operating conditions. Electric motors are all the rage these days. In this case we're modeling the oil distribution through the system as it comes through the pan, goes over the rotors and stators.

E-motors rotate at around 10,000 RPM, give or take a few thousand RPM, so we're dealing with very high rotational speeds as well as very small volume fractions. In this case, not only are we modeling the oil distribution as it's sprayed and sent throughout the system, we're also modeling the heat transfer between the systems. We're looking at wildly different time scales — heat transfer — yet we put all this together, and we compare it to the thermocouple measurements from the customer.

I can say the customer — this is GM, General Motors — because this is from a white paper we published with them a few years back. General Motors, and quite a few other companies, use us as their e-motor simulation solution. You can see comparing against the thermocouple testing — excellent alignment across the board, despite the complexity of the physics involved here.

That wraps up what I have covered here.

Todd Craft (PTC): Question: what's an e-motor? Electric motor is really the answer there.

All right, Alex, that was awesome. I think Tom Quaglia will say something from PTC, and then we'll wrap up.

Closing Q&A, Condensation Roadmap, and PTC Community Resources

Tom Quaglia (PTC): Thanks for having me, Todd. So a lot of you were responding to the invite that I sent out last night with the Tips and Techniques. As you guys know, we do have Tips and Techniques sessions — we used to do them biweekly, but now we're doing them only monthly because we kind of ran out of topics to discuss. There are about fifty of them archived on the community page. I'll drop that in the chat. If you guys are interested in bookmarking it, you'll have them moving forward. Thanks again for joining here. If there are any questions from a commercial standpoint, I'm able to help out with those too, afterwards. Thanks.

Todd Craft (PTC): Great. Thanks, Tom. Condensation capabilities, Alex, inside of Creo Flow Analysis — that's the last question.

A — Alex Zhang: Sure. Right now, what we do is an approximation. If you have some kind of analytical trick for that, we can use our expression editor to generate condensation on specific surfaces depending on what you're looking at. Like windshield de-icing, stuff like that — we do quite often.

That being said, we are developing a full phase change model where a liquid can turn into a gas or a solid dynamically — with movements as well. So keep an eye out for that. It will be coming out very soon, and we are very excited for it. I don't think anyone has a full A-to-Z solution for that yet. We hope to be the first.

Todd Craft (PTC): Perfect. One last note. We have examples to run through an analysis. If you go to the PTC Community, you will find a whole section on Tips and Techniques, and all the tutorials are there. In addition, when we ship Creo, we have a large number of basic and advanced tutorials that are located in the Creo Help. If you go to the Tutorial section there, you'll find a bunch of Creo Flow Analysis examples where you can download the models and run through them.

Thank you, Alex. I will be putting this up on the PTC Community as soon as the recording is available, and add it to the rest of them. So if you want to rewatch this or send it to somebody else, it'll be there. Thank you everybody for attending. Have a great day.

Alex Zhang (Simerics): Thanks everyone. Take care.

Todd Craft (PTC): And thank you, Alex.