Joule Heating Simulations in Ansys, CFD, and Icepak

[This was auto-generated. There may be mispellings.]

Hi, this is Mingyao from Ozen Engineering. In this talk, titled "Joule Heating Simulations in Ansys, CFD and Icepak," I'll be looking at how we set up dual heating simulations in ANSYS Workbench. There are a number of different packages that can do this.

I'll be looking at starting with mechanical, ANSYS CFD, and then Icepak. Dual heating is a condition where when you run a current through a conductor, it generates heat.

So, it's a multi-physics problem that requires us to model both electrical conduction as well as heat transfer and fluid dynamics thermal convection into fluid. I'm starting with a pretty simple model here. You can see that I have a little conductor in the middle surrounded by an air box.

So, this is all inside the workbench environment. The first model I'll show is doing a thermoelectric simulation. Here's my model for a thermoelectric simulation, which is a conduction-based thermal analysis. We will not be needing our air box. Let's set the material for this part.

We'll call it tungsten. There you go. Okay. So, this is tungsten. Okay. So, this is tungsten. Okay. Okay. Okay. Okay. Okay. Okay. Okay. Okay. So, there's a few tungsten alloys available in here. Let's go with the TA10W. We can look at the material properties.

The ANSYS has a pretty large material library now. So, we have metal plasticity values, temperature-dependent, thermal expansion coefficient, different SN curves for fatigue analysis as well as, you know, the Oxygen, the Oxygen, and the Oxygen.

And then, of course, we do have the isotropic resistivity and specific heat and thermal conductivity so we have everything we need for this model. Let's look at the quick mesh. The basic standard mesh is very coarse for this, so I'm just going to go ahead and specify a mesh sizing on this body.

We'll make it maybe half a millimeter. We should give this pretty good resolution on this model. Now we set up the simulation. We want both the thermal and the electrical characteristics. Let's make this bit bigger so I'm going to add some boundaries here.

This will be my current, and I'll say let's go to metric units here. We'll do five amps going into this terminal and we'll set this terminal to zero to have a voltage of zero. This will tell us the resistance across this model.

It's doing a thermal analysis, so we also need to be able to cool this down. I'm going to select all of the surfaces and maybe deselect these parts.

I will say there's a convection boundary on all of these areas, five watts per meter squared degrees Celsius, which is a typical setting for natural convection. Since we're not doing force convection in this model, we're just going to guess at a number and see what type of results we can produce.

You can see within a couple of minutes from beginning to end, we have a prediction of the temperature. It's going to be 85 degrees Celsius, which is how high the heat-up will be. We can look at the voltage. The voltage drop across the temperature is going to be about a quarter of a second.

So, we can see that the current density across this conductor is about 56 millivolts. We can also look at total current density and let's say dual heating. You can see where the heat has been generated. It's kind of related to current density.

So, as the currents are going across these tight curves, we have some hot spots here where there's a large amount of heat, and then it kind of stabilizes. We can turn the joule of the current density into vectors, which shows you the direction.

We can adjust the sizes of the vectors so there's a lot of basic pre-processing and post-processing capabilities to make the heat transfer so we can see that the heat transfer is there.

But in case some of these gains or projected followers will cause these to spin up, we'll take a look if I could also see if that sink is right for example, is this one for example, field, and so this will come and yellow jump.

We are seeing that the air flows that the part where 73 into which is basically in finish fourth. So, we're going to see that the air flows into this part, where it's going in, and we can adjust the fan speed to see what the actual temperature I get. Now let's do a CFD analysis.

So, I'm going to use CFX. It has a pretty easy-to-use thermoelectric analysis capability. Here, we can put in the air box so we actually blow air across our conductor. The same interface for meshing.

I think one of the things one of the things I want to do is set the mesh to be the same on the conductor itself. So, highlight meshing, we're going to put sizing in, and we're going to set a size of 0. 5. One of the things that's nice to have in CFD analysis is a decent boundary layer mesh.

So, we want to have inflation on this body and these surfaces around there. This will be typically, I like to use a 20 to 1 aspect ratio. So, the thickness of that initial boundary layer mesh will be 1/20th of the size of that mesh, and let's do 10 layers. Let's look at the mesh.

It's pretty coarse on the outside, but we care about the mesh around our tungsten wire. And you can see we have the tungsten wire well resolved. I can move this around a little bit so we can see a little bit better here.

And you can see we have nice boundary layer meshing that resolves near-wall flows and this gives us accurate heat transfer coefficients and heat transfer calculations. Let's go ahead and run this CFD analysis. This is a CFX, which is one of our CFD packages.

It's pretty easy to use, so I tend to use this when I can. Now everything is a single domain, so we need a new domain, and we'll call this our wire domain. And we'll say it's a solid and it's inside. So, that's the part inside.

We're going to do thermal energy, will do electrical magnetics, turn on electric potential solver, and that's all we need to do. The aluminum doesn't have a material, and it's not actually a new aluminum. So, we're going to have to insert a new material, which is pretty straightforward.

It's called tungsten, and it's going to be a solid. And the material will be Let's go ahead and copy the information out of ANSYS Workbench now that we have defined tungsten. If I click on this button here, filter engine data, material, I see all the material properties we have for tungsten.

So, we have a density, which is in SI units, and we can mix and match units here. A specific heat capacity when this is required for about 140. And we can I can change the unit globally if needed. Thermal conductivity, it's a solid, so there's no viscosity.

There's a thermal conductivity here, which is 42. And there is a little magnetic conductivity. So, we have Electric electromagnetic resistivity here, so we can put this in as an equation like this. Oh, I need to copy and paste. So, we can always do 1 divided by and the unit is ohm per meter.

Oh, meter. There you go. So, the CFX will automatically do unit conversion for you, which is nice. Let's go ahead and change the wire from aluminum to tungsten, which has our electrical conductivity information. And now that's done. On the fluid dynamics side, let's check the models.

It's going to be thermal energy. We're not going to worry about electromagnetics because we don't really want it to be conducting here. We have a turbulence model set, which looks fine. And now we can go ahead and set our boundaries.

It looks like there's a heat transfer option here, where it wants to make sure we have the right information. Clicking on the error message automatically takes you to the location to fix it. Okay, so let's go ahead and hide this and put in our electrical boundaries first.

So, this will be the current terminal. I'm going to select this surface over here, and we want to put in a current. It's in Amps per inch squared, so we want to specify this will be 5 amps. But it's going to be we need a kind of a volume area averaged.

So, we're going to divide it divide this by the area at current. Okay, and it takes it. If you don't know these commands, you can right-click, so the function will be area, and then the physical locator is current terminal. So, that's what I did to specify that.

And on this side, we'll call it voltage. So, we over here, so we're setting up an apples-to-apples comparison. And the voltage will be zero, so we can pick ground. That's the this the the wire set up. Now we want to set up the airflow.

So, we're going to have by default is air at 25 C, which is fine. I can put in an inlet. So, we're going to blow air on this side at 1 meter per second, 20 degrees Celsius. Then we're going to put in an outlet, so you have any let you always need an outlet.

You see if the otherwise air gets filled up there and it's got nowhere to go. Zero. And then the rest of the boundary is collected by this boundary condition thing. And we're going to set this to Free slip wall.

So, we don't have No, so the wall means there's zero flow, zero tangential velocity along the wall. Because I have fairly large mesh here, that's going to show up as some pretty large gradients. You know, one second if it cannot move, then it can move. So, we're going to let it slip by.

And typically, we should converge within the hundred iterations, but let's do 200 just to be safe. What we typically like to do is monitor the temperature because we care about what the temperature is on the wire. So, let's add a monitor point and we're going to use an expression.

So, we're going to monitor Max value of Variables, let's find temperature at Locator 3D, the wire. So, we're going to try to track the maximum temperature on the wire as we run the simulation. Then that's going to tell us when things stabilize.

So, let's go ahead and go back to the workbench now that the CFD part is set up. We can do a quick analysis and see how hot it gets with one meter per hour wind. I probably don't need to do HPC, but let's just run off four cores. The simulation has completed.

Let's go ahead and take a look at the results here. So, we have a temperature plot. Let's plot Eun insights. There's a surplus protein? Now, does that look bad or I don't think so. Gotcha. So far, point in time. And let's try that again. So, here starts now at 50 degrees two squared.

Now, is that current value? So, well, I'm going to have to or English Engineering, let's see. Okay, okay, okay, custom, that's what we want. We have Celsius here. Okay, so it goes up to 53 degrees Celsius. Let's change the units so it's not in scientific notation.

Okay, very similar patterns to our structural analysis, but because the convection is different, because we have force convection here, the little wire is much cooler. You can do some cross-section plots on the YZ plane and let's color it with pressure, still velocity, do temperature maybe.

You can see the heating effect of the wire and the air picking up the heat as it convects downwards. We can look at the velocity to show that this is a 26 mm per second maybe. I used a strange inlet velocity here. Oh, 1 inch, that's let's do 1 meter per second. We'll run this again.

So, always check your units. Let's look at our results again. Let's recognize that there's some new results. So, hopefully, this will show temperature and speed is just about 1 meter per second. Okay, let's take a look at the temperature again.

So, now it's only 37 degrees Celsius after the increased airflow. So, airflow obviously has a huge impact on it in addition to temperature. We can look at current density, just like before. See the same kind of concentration of currents near the corners.

And we can look at electric potential, so this shows you 0. So, 56 mV of voltage drop. That's also very similar to the electrical results from our ANSYS Mechanical Thermoelectric Analysis. So, that's an example of ANSYS CFX. Let's go ahead and do this now in ANSYS Ice Pack. I mean, ANSYS Fluent.

Let me close some of these things so that my computer can free up some memory. Ice Pack is a little bit different. We're going to have to duplicate our geometry because Ice Pack requires that we create an Ice Pack model for this for this analysis.

Ice Pack analysis requires us to specify the type of geometry it is, and I'm going to use a very simple model to represent it. This is our original geometry, and for Ice Pack, we need to go to ANSYS SpaceClaim and do some simplification.

Now, you can actually have the full CAD object, but it is much more efficient to model parts in Ice Pack as Ice Pack objects. This is because when it's all hexahedral elements, ANSYS knows exactly how to mesh it most efficiently.

For the airbox, I'm going to turn it into a bounding box because Ice Pack allows us to have overlapping models. So, that's all I need to do to prepare this model for Ice Pack. Now, I'm going to find an Ice Pack component system and start the analysis.

The way Ice Pack works is that it automatically wraps a cabinet around the largest part, which is our enclosure. And really, we just need the cabinet because our enclosure was meant to be our flow domain. So, I'm going to delete this part. So, now we just have a cabinet.

We have to define our wire again. So, we need to define a material. So, let's call it tungsten, and it's a solid. We gotta define our properties again, so let's pull this over. Okay, there we go.

So, the density is this value, and specific heat is 140, and thermal conductivity is 42. 05. So, that's tungsten. Tungsten will be these four components. You can see when I select it, it highlights a little bit to look at it a little bit better.

We can go to select a solid, highlight, so this way whenever I select something, it gets highlighted. So, these parts will all be tungsten, and it will be heated by joule heating. There's a constant joule heating or there's a variable version.

So, we're going to use the variable version and take my resistivity and put that into here and have it no temperature difference, no temperature variation. So, basically, the resistivity is a constant resistivity. Now, we want to add the terminals. The terminals are these little battery signs.

So, we are going to move this to here. So, this assigns a boundary condition here. And we're going to expand that. It's not going to add any power but it will add some current to it, 5 amps, and no voltage. And let's add another little battery here. And we're going to move this to here.

And I just click on the surface over and over again until it gets me the surface we want. Resize it so that it's the right size. And edit the the the voltage to zero. So, exactly the same setup as before. We have 5 amps coming in here, zero volts over here.

We want to adjust our cabinet so that the maximum side of Z is an opening and the minimum side is also an opening. We can do a grill too, but I will set an opening instead. The property for this will be in the Z axis, minus one meter per second wind, so it's going to be going that way.

And on the other side, it will specify ambient static pressure. Okay, so you can see one of the benefits of Ice Pack is it does all of the meshing setup automatically. We can go through this quick quick problem setup wizard. So, this will be an inlet, outlet, force convection.

I think we had it set at turbulence, so I'm going to set the same thing. We're not going to we're going to ignore heating due to radiation for now, no solar radiation, steady state under solution. Typically, we want to also let's run a couple of hundred iterations. And for parallel, we can do this.

We can use let's say four cores, the same as our other model. And typically, we want to monitor the temperature. So, I can select the part I want and create a monitor point. And it puts it in the middle, which is fine. And it creates a monitor point up here somewhere here.

And we're monitoring the temperature, which is exactly what we want. So, let's go ahead and run the simulation now. So, Ice Pack automatically meshes this model. It's going to be a pretty coarse mesh.

We could definitely tighten it up, but Ice Pack has some basic settings to make sure there's three elements through the thickness across the different parts. So, it's not bad. It generally generates a mesh that's good enough to capture the temperature gradient across the different parts.

Obviously, a problem here. I didn't let it run long enough to converge. So, Ice Pack has a couple of different solvers. I probably should pick another solver. So, I'll show you how I have to go about fixing that. Let's go to 500 iterations.

And this time when I do solve, I'm going to use the coupled pressure-velocity formulations, probably should be on default. This makes it similar to the CFX solver that we just used previously. Let's go ahead and run this again. So, amazing. Right?

How fast a different solver converges, rather than running 150 iterations, 200 iterations here, it converges in 40 iterations. Shows you the power of the pressure-velocity coupling. Let's take a look at the temperature on our wire and see how much it's going to change. Let's see and display.

So, 33 degrees versus what was our CFD post result. So, 33 degrees. We could, we can also plot electric potential. So, 56 57 millivolts of potential drop. We can locate electric current density. The coarse mesh really didn't pick up the kind of the the corner high current areas.

So, this is the area where more detailed analysis may have helped. And joule heating density here shows you how much a heat has been generated in the middle of the wire, which makes a lot of sense. So, let's take a look at this and just to compare the temperature.

Okay, 37 degrees versus what's the temperature here. 33. I believe. Okay, so, oh, I know. No, we did we did define tungsten then. Okay, so it's probably a meshing issue. And if we're going to do a little bit of a test here and we're going to do a test here and we're going to do a lot of measuring.

So, if we look at part of the wire here and we showed the mesh, you notice how much coarser it is. And so, we can go in and set some settings on the mesh. You know, we can do local meshing and it object parameter select the wire and say you define how many counts and and other things.

But the idea of Ice Packers really is to model very complex assembly structures and those are the things that can be done here in this patch data point and we can do some work with the plate and and other things.

But the idea of Ice Packers really is to model very complex assembly structures and these are the things that we have to do and these are very quickly and you can see it does make sure there's always only two elements across the thickness.

Okay, so, I could really step up the mesh quite a bit here, even if I pick some global settings, like element in gap minimum element on edge, that should be three. That's a line. Generate this was only a 15,000 know that mesh here.

So, I'll need to there's a course and normal one there's a lot of other mesh controls that we can use to tighten up that mesh.

But the idea here is I wanted to show you what we do to in simulation to generate a dual dual heat analysis in essence, mechanical using conduction-based thermoelectric, we can do in our ssd product, a full detail CFD analysis that gives you a lot of control for meshing and kind of boundary layer modeling, advanced turbulence models, and for very large electronics cooling of the applications where you may have hundreds if not thousands of components.

Ice Pack lets you set up these models super fast. So, three different solutions, all can do dual heating. Hopefully, this was helpful. And if you liked this video, please subscribe to our YouTube channel. And if you have any questions, please feel free to leave a comment below.

And if you have any questions, visit us at ozeninc.com. Thank you, and have a great day.