Become a Fusion 360 Simulation Expert in 60 Minutes

SHEKAR SUB: Welcome to AU 2021. Today we are presenting the class, Become a Simulation Expert in 60 Minutes, we will start with a brief introduction.

My name is Shekar Sub. I work on the fusion simulation and generative design. I've been at Autodesk for 24 years. I have presented at AU many times. And this is the fifth or the sixth time. I cannot remember.

In my previous life, I got my bachelor's, master's, and doctorate in mechanical engineering. I also was lucky to be a co-author of the book Mastering Autodesk Inventor. In my spare time, I volunteer for FIRST Robotics. And to balance the sedentary lifestyle, I do walking, yoga, and tennis. Hugh, could you please introduce yourself?

HUGH HENDERSON: Hi, everyone. I'm Hugh Henderson. I'm a QA engineer at Autodesk. I've been working for Autodesk for 19 years now. And what I work on is fusion simulation. In the past I've worked on inventor simulation.

And in my previous life I was a design engineer, designing fixtures for the IGT, the Industrial Gas Turbine industry. And I have a degree from the University of Illinois at Urbana-Champaign in mechanical engineering. I'm familiar with FEA. I took some classes there for FEA. It's really near and dear to my heart.

And also thermodynamics, I had a focus as a minor. And over to you, Shekar.

SHEKAR SUB: Yeah, I'm very excited to present with you, Henderson today. And a warm welcome to everyone again.

Let me summarize the key learnings of this class. You'll be able to set up a fusion simulation analysis with loads, constraints, and contacts. You'll be able to learn to interpret the simulation results.

We will be presenting tips and tricks throughout the presentation that's been gathered over the course of time from community forums and other discussions we had with customers. And we will share how to avoid pitfalls when using fusion simulation.

Here are the contents that we'll be going through today. We will start with simplification. And then we will cover what are the different types of studies that are available in fusion simulation?

To set up the simulation, you need to have a good understanding of materials, constraints, nodes, and contacts. We will see how much of a knowledge is needed about machine. We will cover pre-check and [? solve. ?] The pre-check command is very helpful. And then we will talk about results.

Fusion simulation is based on industry acclaimed Nastram, Explicit Solvers. And you will be delighted that Moldflow and CFD have been added to it. Not only that, you can send fusion simulation setups to Ansys, and then do the solve there. It is multithreaded for meshing and multiplatform. Works on Windows and Mac.

Here are the simulation steps that we are focusing on. Though there are other sub steps that you have to take when it comes to, for example, simplify, so you start. You are done with your design and what's next before you do manufacturing.

You have to create studies. You have to choose which type of studies do you want for your simulation? Then simplify the geometry, if needed. This could be an optional steps. Then we will talk about how to apply constraints and loads. And then doing the solve, how do you get results? And how to interpret them?

Let's talk briefly about simplification. In Fusion 360 we have the Simplify workspace. A workspace is just a set of commands. Think of this as an additional workspace to the design workspace where you can create multiple variants of the design workspace model. You can create what if scenarios.

Suppose you had a washer and you want to look at three different sizes of the washer, those could be three different what if scenarios. How do you simplify the simulation models? We will cover the commands in a demo. And this is the workspace where you also remove unneeded geometry that's not needed for simulation.

What are the advantages? You get reduced element count. And machine is much faster. In the demo case, we will see that it's at least 6X faster.

In the Simplify workspace, one should be aware of simulation models. So the base model is what the drawing is based on. Let's say you have this base model. You can create two simulation models for this.

One, I removed all the fillets. Fillets are some of these external fillets, on which there are no loads or constraints. They cause problems. Or I can keep the fillets, because I want to understand the stresses on the fillets. And I can remove the text from the second model and do a different analysis.

So here I have one base model and two simulation models. And on each of these simulation models I can do different types of analysis. What do you remove? Any face or edge which is a small feature, any entity that's a small feature you can remove if you are not interested in the stresses or the displacements on those faces.

[INAUDIBLE] there are no loads or constraints on those faces. So these are mostly fillet or chamfer faces with no load or constraint. Also, they should have no impact on the overall stiffness. Otherwise you might get incorrect results.

If you are doing frequency analysis, you have to ensure that they don't alter the mass for frequency analysis. So what are some examples? Screws and bolts. Screws are tiny most of the time. Fasteners, you don't need them. You can remove them using the remote feature. And for bolts, you can use connectors where you can create bold or rigid connectors.

Lifting eyes or handles, name plates, these are all decorative. They do not add to the overall analysis or to the results, because they are only there for decorative purposes. You have to only look at the stresses, and the strains, and the displacement on things like the engine block, et cetera.

Panel switches or indicator lights are another example. And here you can use point masses. There could be some internal fillets that you should not remove because you want to understand the stress concentration there. So I would keep those internal fillets.

Symmetry is another way to leverage simplification. This is a brake rotor example. You can put constraints on this that will lead to excessive heat generation. Instead of that, you can just split it into 1/8 or 1/4, and put frictionless constraints on the yellow and the pink faces. And then use leverage symmetry.

Here is an example of a buckle where you can just study 1/4 of this and then do the analysis. The tip I want to share is, do symmetry changes in Simplify workspace, because you don't want them reflected in the design base model, which is in the Design workspace. Avoid modeling with symmetry if you are performing modal frequencies or structural buckling, because asymmetrical structures can have asymmetrical vibration modes, and which we want them to analyze when the structure is twisting.

Here are the Simplify commands, such as remove features, where it hunts down all the features in the model and then gives you a slider to remove them. Remove Faces, these are mostly for fillets which needs to be removed, where you select one face, and/or one or more faces, and it gives you all the tangent continuous faces that need to be removed.

We also have other commands, like you can say remove all except selected. Just a useful thing to remove small features. Replace with primitives. If you have complex parts, we can replace it with a simple box, or a spear, or a cylinder instead of analyzing all the complex geometry in the part you have.

So if you see the regularization failed error message, that indicates that you have a need for simplification. Let's look at a demo which will demonstrate simplification.

PRESENTER: Hi, everybody. I wanted to use another quick tip to touch on the new Simplify workspace, this time focusing on some of the incredible things it will do for you. Note some differences that you might find if you don't have Fusion Ultimate. Then compare solve times using this uber familiar engine bracket with some notable changes, text, chamfers, and fillets, on just about every face you can imagine.

So let's get to it. Once in the Simplify workspace, we can use the modeling tools you're familiar with when making your designs, things like extrude, revolve, and sweeps. But we'll focus on what's found under Modify.

The three found at the top of this list is where you'll probably spend most of your time, and they are part of Ultimate. Remove features. The first packs some real power. With this command enabled, all we need to do is select the body or bodies of interest. When I do, Fusion will start painting any and all of the features that will get wiped when you hit Delete.

Right now it's not much, but as I slide the filter to accept larger features, I'll be able to remove more and more. At some point it might go too far and we can back it off a bit. But further to the side here, we're provided with more options.

Maybe you want to avoid removal of all the fillets, or the holes, or chamfers. Just select the toggles from below. But when you get it highlighted the way you want, delete away.

Further to that, the remove faces command provides manual controls to get just the faces you want. And replace with primitives will do just like it sounds. Replace complex bodies with primitives to take their place. This works well for cylinders, boxes, and spheres.

But let's back it up a bit. Another goal I had was to illustrate why you might want to do this. So let's clone the setup we have, another Ultimate only feature, which will copy everything right down to the mesh. Then, with that Sim model two activated, roll back the timeline to make it fully featured again.

Now, we solve for both cases. We'll jump forward to where we've started comparing the results. And what I want you to note are the very minor differences between these two cases, starting with the max displacement. We're talking about a 2% difference between the simplified case on the left and the fully featured case on the right.

Similarly, the max stress values are within about 7% of one another. But the real alarming thing is not how close these results are. It's how much faster this simplified case ran. These two comparisons, drawn from the solver data window, show that the CPU time was about six times longer for the unsimplified version.

And understanding the reason for this is simple. When you compare element counts, we're talking about almost 38,000 elements versus 3,000. We hope you find using these simplification methods will be well worth the time invested in them. Cheers.

SHEKAR SUB: As you can see, there was only 2% difference in the displacement, but we got a 6X speed up with reduced number of nodes and elements in the mesh.

Another tip I want to share is, often there is complex geometry where you can have several fillets coming across at an intersection. And sometimes if you try to use the remove face command, it might not work.

One tip we want to share is, you can add spear geometry at the intersection of the fillet faces and then use the remove faces command. And then the remove faces command will work.

With that, I want to pass the baton to Hugh to talk about studies.

HUGH HENDERSON: Thank you, Shekar. So as Shekar mentioned, the Sim workspace is integrated into Fusion. So the first step is to enter simulation. And we use the workspace switcher, shown there on the right. And what we do is we make a copy of the design model so that sim can work without affecting the original design.

And the other features that we have is not only the results, but we can compare results. And the workspace is similar to generative design, where we tried to make all the tools familiar. Next, the UI is much like the rest of Fusion, where there's an order from left to right, where you start at the left part of the ribbon, and work your way over as you progress into the setup.

And we have its own browser representation for the studies, and the loads and constraints. It's an easy way to access and keep track of where you are in your setup. And additionally, in our marking menu, if you right click in the canvas, we offer simulation only commands that are commonly used in not only the radial menu, but also in the dropdown. Next slide.

So you have to choose a study type. That's the first thing you're presented with when you enter sim is the new study dialogue. And to help you decide which study type I'm going to use, at the very bottom left there's a Help Me Choose a Study Type text. And you can click on that. And it will ask you questions, like what do you want to-- what are you interested in?

So, for example, the first one, will it break? Oh, you need static stress analysis. Or will it vibrate apart? Then you will choose modal frequencies, et cetera.

So we do give you some assistance here. And the other really cool thing about these studies is that you can kind of reuse work that you've done before by cloning a study. And then you just tweak it, make changes. And then you don't have to redefine it from zero.

You can also delete them, et cetera. And what's exciting now is previously we had only eight study types. But we've added two new physics types. We have Ecooling and plastic injection molding.

And this is the one where, Shekar mentioned, we have CFD technology and the Moldflow technology. It's plastic injection molding, how the liquid will solidify into the shape of the mold. It uses rheology. So those are really exciting new techs that we have.

So I'm just going to go through a couple of the main study types that you will probably be using. And static stress is probably good for 90% of the problems you have, because most of the time you're not going to have large deformations or go into the bending zone where it's going to permanently deform. So static stress is probably your go to study.

Next is modal. You want to make sure that you don't have any resonant impulses. That's going to really interfere with the design and make it fail that way. And then thermal, pretty self-explanatory. Will it overheat?

And then finally, shape optimization, where we just want to lightweight it. Say that you build some bracket, and it has way more material than you need for the strength required. We'll just kind of melt away the parts, the material that doesn't carry any load, that's not necessary.

And this one you can only solve on the cloud. And yeah, the tip is to use a fine mesh size, because that will give you the resolution of how smooth the outcome will be. And over to you, Shekar, for materials.

SHEKAR SUB: Yeah, thank you. So let's talk about physical materials. We are not interested in the appearance materials. These are normally assigned to bodies. Materials play an important role.

When you create a study, the materials in the design are carried through to simulation. You do not have to do any changes for materials. You can go to the next step of constraints, applying constraints, loads, et cetera. However, in the study materials dialogue, you can override the material.

In this case, there are two bodies. I have steel in the Design workspace. And I can override them to aluminum. And they're nonlinear materials, such as titanium, high strength alloy.

We have several material libraries to select from. The Fusion 360 material library has at least 10 plus categories with hundreds of materials. The Fusion 360 non-linear material library has 12 materials.

Fusion 360 [INAUDIBLE] material library has around 11 plus. The question is, what happens if I have some special material that's not in Fusion? Well, no problem. You can create a custom material and input the properties.

So in terms of some of the tips, we warn you if the material cannot be used for solve. So let's say I use air as my material and apply it to a bracket to do the solve. Obviously, that cannot be solved. You need steel, aluminum, copper, whatever, right? So that will be [? blank. ?]

Sometimes some material property might be missing. Let's say Young's modulus is missing. Then we will throw a warning. Let's say a huge [? chunk ?] is missing. Then we throw a warning, because we want to make sure that your safety [INAUDIBLE] calculation, if it is based on [? yield strength, ?] is reflected in the results.

And if you apply a linear material for non-linear solve, then the linear material does not have nonlinear properties. So how will we solve, right? So we throw a warning.

So in terms of physical materials, a quick overview is isotropic materials are where the thickness is constant. The Young's modulus is constant. If you want a custom material, like I mentioned before, if you follow this link, it will give you steps on how to create, for example, steel material A516M grade 65 smart in Fusion 360.

One of the customers was asking about this. And we had to suggest that he had to create a custom material. Some materials such as wood are orthotropic. In isotropic, all the properties are the same in all the directions. But in orthotropic the radial and the axle directions have different properties. So that's not supported.

Same with visco-elastic if you are trying to simulate form or shoe insoles, for example. The tip here is if you want to change the properties on multiple materials, then basically you have to access the study materials command. And then when you hit the right mouse button on a material, all the components that have the same material are automatically preselected so that you can change the extent of the ultimate tensile strength that you want to use for safety factor calculations.

We have advanced materials too. You can define a stress strength which goes beyond the [? heal ?] point for nonlinear analysis. You have to enter the values, or you can use a CSV file from a site such as MATLAB, and use it.

In this broad dialogue, you can always use Control to add rows, or Shift to select a bunch of rows and change the properties, delete them, et cetera. The nonlinear material properties that we support are elasticity, elasticity, and elasticity.

And when you go to the Create Nonlinear Material, or use a nonlinear material, you can see the properties. All the properties in the libraries cannot be edited. You have to create a custom material. You can always copy from the existing libraries.

And then we have hyper elastic materials, such as car tires or fluid seals. And these have to be highly flexible under high loads. These properties are also provided in the Fusion 360 simulation material library.

Then we have temperature dependent materials, where the properties can vary at different temperature points. And you have to enter a graph for that. And we provide tools to help you out with that, including additive materials.

Next, once you have figured out the materials, you don't have to do anything. If you do the pre-check now, it is going to say, hey, you don't have any constraints. What are constraints?

The goal of the constraint is to limit the translation and the rotational degrees of motion. You need at least a few, otherwise your part will be floating in space. You can apply them on faces, edges, and vertices.

And here are some of the constraints. For those of you who attended the class previously, you might see the remote constraint, which is kind of new for this year's. You can also completely not apply any constraints and use a checkbox, like remove rigid body molds.

This is like an auto option where the solver will apply the needed constraints, the acceleration loads to keep the model statically stable and do the analysis for you. But there are certain [? gauges ?] just to check this option. And I would suggest you to check help on this before checking this option.

Here are the four constraint types that I would like to bring attention to. Fixed constraint is the most useful one. Let's say you have a bracket attached to [INAUDIBLE]. You want to harness the degrees of freedom in the X, Y, and Z direction.

Then you can just check this Ux, Uy, Uz boxes in the UI. And that will fix the bracket without any moment. You can unselect to unfix in the dialog box.

Pin constraints are used on cylindrical faces. You basically have three directions to worry here. You can arrest motion in the radial, axial, or the tangential directions.

Frictionless constraints. We just saw it in the symmetry slide, how you can apply frictional constraints where the motion is arrested in the normal to the face direction. And here, there is no movement normal to surface. It is only in the tangential direction.

Then we have a remote constraint. Think of remote constraint as you have to specify an anchor point in thin air. And from that point, it will keep the entity, the face, fixed. It is similar to a fixed constraint, except that the anchor point is in thin air. Again, you have Ux, Uy, Uz, and Rx, Ry, Rz, which is the translation and the rotational degrees of freedom.

Then we also have idealization types. What are idealization types? You don't need to model all the geometry. You can specify a point mass.

Let's say you have some kind of coiled wire on the back of a truck. You can use an auto point mass. Because it's very difficult to model a coil on the back of a truck. Or you have a car on a bridge. You can use an auto point mass. There we just create some graphics. And the geometry is not created.

You can create manual point mass also, which we'll cover a little bit later. You can create a bolt connector, where you have two plates and you want to make sure they are fastened together. Here you have the threaded or bitmap option.

Again, these are graphics only and no geometry. Rigid body connector is another connector type where you have to specify the start point, or the start entities and the end entities. The start point is about text here, and the end entities are like edges.

In this case, you have to be careful. Rotation is allowed. And you have to still arrest rotation by selecting more number of vertices on the triangle here with the edges or faces of the blocks. And there is no movement normal to surface for rigid body connectors.

This question comes up whether we have string connectors, or shelves, or BIM idealizations. And that we do not support today.

Let's briefly talk about loads. Loads are the ones that causes stresses and displacements. If you had no loads, there is no stress, nor displacement. So these are forces that act in real life. Is someone pushing the block? Or is the block being pulled? And this could be gravitational force.

In this case, here are the different load types that are supported. And the goal is to specify the magnitude, direction, and units so that you get meaningful results.

Again, we will cover four important load types. The force is the one which is used widely. You can specify a force in any direction. You can also specify a limit target, which basically means that you don't want to apply the force on the entire face. But you want to apply only on a subset of the face.

Then you have force per entity, where you can specify if you have 100 Newton force on two faces, if this option is checked, then we get 100 Newtons on each of the faces. If it is unchecked, then you get a total of 100 Newtons on the two faces. So obviously, this is for multiple entities.

Pressure is always normal. Imagine fluid in a vessel, and you want to represent that with the pressure. Again, multiple entities are selected. You can also apply a torque to the model. So you select basis, and you select an axis. The torque acts along the axis. And we always go to the center right of faces if faces are selected.

It does not matter where you apply the moment. And again, multiple entities are supported. Bearing nodes are used if you have bearings in your model, between a shaft in a hole.

Here, as shown, if you have a full face, then the bearing load acts on 180 degrees. And you can see the parabolic distribution, where it is high at the top, and fades as it goes to the ends.

What are load cases? If you are doing an analysis, let's say you are doing the analysis of the motorbike maneuvering a corner. So then, if you want to analyze the study with just the weight of the rider, that could be load case one. If you are analyzing, accelerating, and braking as you turn right or left, that could be load case two.

And if you want to just study the cornering forces, you can put those loads and constraints in the cornering load case. So load cases are just a way to group loads and constraints and do the analysis so that when you do generate results, you can view the results by load case.

Things which are not unique to your load case are such as materials. Once you specify a material, it's for the entire study. You cannot specify a material or a contact just for a load case.

You can always double click in the browser to activate the load case. And you can view results by load case, like I mentioned earlier. Point masses we covered. So if you have a geometry in your study, you can approximate it with a point mass. And we just create the graphics, as shown here.

It not only reduces the file size, but also the element count and processing time. Selected solid bodies will be suppressed. If you don't want the selected solid bodies to be suppressed, then you can just use point mass manual.

This just means that there is some point mass which I am not doing any geometry. I don't want to suppress any component or body. I just want to represent the point mass. And that can be represented by the manual point mass.

We just specify the faces it's connected to. And then in the UI, you can always see which direction corresponds to which of these arrows here by changing the input field value while you are dragging these arrows. Next, Hugh, could you please cover contacts?

HUGH HENDERSON: Yes, thank you. Thank you, Shekar. So contacts. So why do we even need contacts?

Well, usually you don't have just one body that you're simulating. You're going to have like two or more. And so how does the load get transferred from one body to the next? It's via contacts.

And it specifies how they're connected. And another thing to note is that it has no relation to joints in an assembly. So if you went and created a bunch of joints in design, and you come into simulation, it's not going to know that this one is going to twist around the other one, for example. So you kind of have to ensure that you set up the contacts according to how the mechanism behaves.

So here we see the contacts manager, where the contact type is bonded. And that's our default contact type. Cause usually two bodies are glued together. And the way that we figure this out that they should be glued together is that we look at all the bodies in the model. We look at which faces are touching.

And so if face A touches face B, then hey, we need a contact there. So we build a contact automatically. And then you can go and change it to a different type if needed.

And here are the contact types. So bonded we already talked about. Separation. So Shekar had the example where those two plates with the bolt connecting through them.

So you have to know, hey, if I leave this as bonded, the force is going to go through that contact, and not even care that the bolt is there for most of the load. And so what you have to do is change the type to this separation. And so the separation will really allow the place to separate so the load will go through that bolt connector.

And you have associated degrees of freedom with all these contacts. So usually there's like a normal direction and a tangential. So that's how these differentiate each other. What is allowed to be free, either normal or tangential?

And sliding is the most interesting one, because it's the one that allows a friction coefficient to be taken into account. And then we have rough also as a common used one. And you'll be able to see this table in the online help. I'm not going to go through all these. But kind of like what I mentioned before, it's all about the degrees of freedom.

And then meshing. Over to you, Shekar.

SHEKAR SUB: Thank you, Hugh. So with mesh settings, I'll briefly cover some of the dials that are important. You don't need to do anything. But if there are solver errors, then you've got to look at the mesh settings.

One is the average element size. You need to make sure that you are using the right size. There is element order. You can use parabolic or linear.

We use the tetrahedron. So we use the four sided node. And it is linear. But if you use parabolic, then we also add [INAUDIBLE] side nodes here. And you can make the edges curved to make sure that more nodes are covered. And it's a different analysis type. It's the same study, but different method.

Here is a good example. On the left you see that the coarse mesh is giving you 16,000 PSI. But the fine mesh is giving you 32,000 max PSI. So it just shows that same load case, but two very different results. So be aware of your mesh size.

Then there are things like aspect ratio. Aspect ratio is the ratio of the two edges, and maximum and the minimum edge. And you need to make sure that they are equal to one, otherwise you'll run into stress concentration. And things such as turn angle, I would use a lower one to get a smooth circle versus jagged, like shown on the left.

How do I make sure that the mesh settings can be controlled? We can't use one mesh size for all the entities in the model. So you can use local mesh controls, where you can go and select the adjacent faces, and then refine the mesh, and make the mesh density fine or coarse depending on the need.

The results are adaptive mesh refinement. And this is more like an auto mode. And we'll just see a demo which will talk about these dials, which summarizes.

In the demo, you will see that if you do no AMR, then mesh size is constrained. And if you do AMR, you get a fine mesh size on one end, and coarse on the other. So it automatically knows when to do a fine mesh and when to do a coarse mesh. So let's see what the demo has to say about AMR.

So in this demonstration I will use a long BIM with a square cross-section. We go to Simulation, and Create Studies. We will create two studies, one without adaptive mesh refinement.

Here when you edit, the load value you can see there I'm applying the load like a cantilever. And it is fixed at one end. And then when I look at the adaptive mesh refinement, this is set to numb. That's my study one.

And in study two, I'm going to clone study one. And then I'm going to say study two, use adaptive mesh refinement.

And initially I was thinking of using a low value. But I'm going to set it to high. And there are four different results that you could select.

What it does is basically it does passes on the model. Pass one, it uses a certain mesh size. Pass two, it refines the mesh in some regions using finer meshes. It keeps on doing it till it converges.

And here you can see that when I get the results, I am ready to compare. I used one [INAUDIBLE] stress, but I can show you the difference clearly by using the first principles stress because of the colors.

On the left I have no AMR. On the right, I have AMR. And you can see that the-- look at the number of nodes and the number of elements. With AMR, it's more. Because it did more finer mesh in certain areas. Not only that, if you look at on the right, the mesh is not constant throughout.

Yeah, I'll stop the demo here. You get the idea, that AMR does automatically refine the mesh in regions where needed. And next we will go to pre-check and solve, Hugh.

HUGH HENDERSON: So pre-check. What this does is it ensures that the solver can be able to run the setup successfully. Oftentimes I make a mistake. Maybe I'll forget to add a constraint. Or I haven't set the material to something that is available that the solver needs to use. As Shekar mentioned, it has to have stiffness, et cetera.

And you could either invoke the pre-check automatic, or manually, just to see if you're OK. But when you press Solve, we'll do a pre-check implicitly and let you know, hey, if there's nothing wrong, we'll just let you send it.

But if there's something wrong, we'll stop you and say, hey, you should think about this. It's a warning. Or if it's serious, like oh, you don't have any loads or constraints, then we'll block you from solving.

And then finally, if everything looks good, you get the green light. And the pre-check passes. So next, the solve dialog, we can either solve on the cloud, or locally.

Locally, there's only four study types that are able to solve locally. The very basic ones, linear, static, modal, thermal, thermal stress. Now, the rest have to be cloud solved.

And the best thing-- I love cloud solved, because it's what we call asynchronous or modal, where you could keep working on the design, and have it open. And the solve happens up on the cloud. So that means it doesn't lock up your machine. You're free to work on other designs.

And with local, you have to wait until the solve has completed before we give you control of Fusion again. So that's why I love cloud solve. And it does cost cloud credits. So we have a dialog that explains what the costs are.

You can use a different account if you have those available to say, hey, for generative studies I want to use this account. For sim solves, I'll use that other account. So we have some good flexibility on how you use your cloud credits.

And next, the results. As Shekar mentioned, this adaptive mesh refinement, we can define how the convergence criteria is specified, whether we want to make sure the displacement is going to be stable, or the stress is going to be stable. And then it's different for different physics types. Say, for thermal, you're not going to do displacements. You're going to want to converge on either the heat flux, so temperature. And it's pretty easy to use, self-explanatory here.

And then the results. Once you do this adaptive mesh refinement, you can actually see, did it converge? Instead of like looking at the elements, or all the colors, what you could do is bring up this-- it's called a convergence plot. And it will have a graphical representation of whether it converges, or converged, or if it diverged.

So here, you could see that it stopped. It's well below the target rate. So this one converged. Didn't even have to go into more than just one or two solution steps.

It could be that it's a tricky problem. And it diverges. If you have a stress concentration or something, it will never reach that target of the percentage difference between the previous result and the current result. So you want to make sure that as you refine the mesh, it doesn't change the answer.

So our result types you could see in the table here. I'm not going to go through all of them. But yeah, the unique one is in sheet optimization. It's the load pack criticality. And buckling has a different one. It's a critical load factor.

And then the most important one that we let you know about is default. As soon as you get your stress result, we'll let you know the safety factor. And the safety factor, of course, below one means it's not safe. It's not even going to hold the working load.

So usually you want to design to a safety factor of maybe around 2 and 1/2 to be safe. Some industries, they do a lot of testing in the aerospace industry. So perhaps maybe it's down to 1 and 1/2 because they've tested it so much they know that 1 and 1/2 is actually safe. So yeah, there's the definition of the safety factor, your material strength divided by the actual stress, the highest stress.

And there's another result. If you have contacts, we could give you the contact pressure result. And to make it even more clear for the safety factor, we will recommend, according to the safety factor, here's what you might want to do next.

So if it's insufficient, oh, you better get a stronger material, or go to nominee or to see why it's insufficient. Maybe it'll strain harden if that's what you want to do.

The other one is marginal and sufficient to access it. So the cool thing about the excessive one is we'll tell you, hey, you could go to shape optimization. Reduce the weight. You don't need all this extra material. It'll save you on cost and performance, et cetera.

Then finally, you might want to share your results with others. It's almost guaranteed you will. So we have an easy way to do this. You just press one button. And then a report is generated.

And so it's a summary of all the setup, the results. And it even has images of what all these results look like in the deformed body.

And then finally, instead of just a HTML report or a Word doc report kind of thing, PDF, you can actually share the model with others by just inviting them to your hub. And the results are also shared.

So I'm user A. I generate results. I say, hey, Shekar, go look at this design that I created, and look at the results. He will be able to see the results. And not only that, we can view them in a web browser.

And then finally, the Compare workspace, Shekar showed you some of that. That's the best way to really compare side by side the different results, instead of having to flip back and forth a bunch of times.

SHEKAR SUB: Thank you for attending this class. And hopefully you can use some of the tips and tricks shared in this class for your design to manufacturing workflow with simulation being in the middle, and be able to repeat the workflow till you get an optimized part that you can manufacture. Thank you for attending.