Autodesk Simulation TV

Autodesk’s Sim Squad has recently released a new portion of their web site, called Simulation TV.  It’s a searchable list of simulation videos, ranging anywhere from what’s new in the latest release to tips and tricks in using each of the Autodesk Simulation products.  You can watch them from your PC, sure.  But they’re also viewable from your smartphone or tablet, so you can watch them on-the-go.

If you’re looking to learn more about Autodesk Simulation Mechanical, Multiphysics, or CFD, this is a great resource.  Check it out!

Autodesk Simulation TV

Check out the rest of their site at www.simsquad.com to see lots of other great information on Simulation tools, as well.

Autodesk Simulation CFD

Performing Computation Fluid Dynamics (CFD) calculations can be a pretty tricky task.  And there are a number of software products available to assist in these situations.  It’s been right about a year since Autodesk acquired Blue Ridge Numerics, adding their CFDesign product to Autodesk’s portfolio of simulation solutions, rebranded as Autodesk Simulation CFD.  The previous solution of Autodesk Simulation Professional (formerly Algor Professional) had some CFD capabilities, but its role was better-suited to stress analysis and non-linear dynamics.  Simulation CFD provides Autodesk customers a much better solution to CFD problems.

Here’s a brief list of some of the things that Autodesk Simulation CFD can do, depending on the version obtained:

• Static and dynamic fluid flow
• Laminar and turbulent flow
• Humidity and steam in air
• Heat-transfer analysis by conduction, convection, and radiation
• Solar/UV heating (even through transparent media, like windows)
• Mixing of two different fluids
• Subsonic and Supersonic fluid flow
• Compressible and incompressible fluids
• Rotational fan airflow
• Motion and rotation of components
• Tools for critical decisions and side-by-side comparison of different scenarios/models
• Automatic and interactive mesh tools

With all of these capabilities, there’s very little that the CFD products cannot do.  Keep in mind, there are three different levels of the product, for which you can compare the features of each here. And while these tools aren’t inexpensive, the information and monetary/time savings they provide make them well worth the cost.

Using 2D Models in Simulations

When most people think about simulation tools, the first thoughts typically jump straight into 3D models and fancy analyses. However, that doesn’t need to be the case for every scenario. In many cases, a much-simpler 2-dimensional simulation can be run to feel things out before jumping headlong into a full simulation of a part of assembly. Below, we’ll take a brief look at some of the things we can do with 2D simulations. And hopefully you’ll be able to see how even these simple cases can speed up your initial design stage and save both time and money.

First, let’s talk about some limitations of 2D in simulation. The largest drawback to 2D is that it doesn’t describe anything with a sense of depth. So, when we run a 2D simulation, we need to understand the assumption that the cross-section we’re viewing has to represent something constant. If your cross-section changes as you move through the part, these types of simulations may not provide any really meaningful results for you. However, Autodesk Simulation products can still use these simulations for items that may be round, using axisymmetrical constraints or information.

So, what types of simulations can be done in 2d? Well, we can certainly do simple static stress analyses. But we can also perform dynamic simulations, which solve for motion, large/permanent deformation of components, and heat transfer. Each of these analyses can assist with your choice of materials and basic cross-sectional design of components. Even friction can be taken into account with these types of analyses.  If you have designs that you’d like to test that have been modeled in AutoCAD or other 2D formats, you can import those items into Autodesk Simulation products to test them.

Shown below is a simple example that demonstrates how simulation can assist with checking a seal on an o-ring. As the flange pushes the o-ring into the groove, we can see how the o-ring compresses to seal between the two components.

 

I recently had someone ask about information regarding Autodesk’s Inventor Simulation package, and, in particular, for documentation showing that the results obtained from this system were accurate enough for real-world use.  Before I just give out the link, I want to mention a couple of things about this topic.

First, how an I ensure that my models are going to provide accurate results?  Using an appropriate mesh size and element type are a great start to any simulation.  But another way to get good results is by using the most-appropriate types of constraints to simulate how the model would be mounted, welded, or bolted down in a real-life scenario.  If necessary, certain higher-end programs (like Autodesk Simulation Mechanical) can even involve sliding components that take into account friction, variable loading conditions, and even different coordinate systems (like cylindrical or spherical X, Y, and Z components) to provide the correct loading for your models.  Make sure you enable gravity, if desired.  And make sure you double-check everything before solving!

Second, how accurate is “accurate enough”?  You don’t need to have a super-tiny mesh size to get accuracy.  Once you get down to a certain size, there’s little advantage to going any smaller, and huge disadvantages to going smaller:  solve time and adequate rendering power.  A small mesh size means longer solve time.  If you’ve tested the model for results comparisons, and your simulation gets you within 5% of your expected results, is that good enough?  You need to ask yourself this question.  It’s best to start with larger mesh and work smaller, than to start tiny and find out it’s going to take three days to solve, cancel, remesh, and start solving again.

So, in short, how accurate your results are will depend largely on how you’ve set up your simulation.  But, using judgment and common sense, you can avoid time-wasting mistakes and get pretty darned close on the first try.

So, here’s the link to Autodesk’s whitepaper comparing some simple scenarios in both accepted real-world values and Inventor Simulation’s results.  I hope it provides some great information to satisfy any curiosity as to exactly how accurate simulations can really be.

Autodesk’s FEA Verification Whitepaper

Enjoy!

Meshing… Meshing… 1..2..3..

My last post mentioned going into the topic of meshing.  So, here we go…

Meshing is a huge topic in and of itself, when it comes to simulation products.  So this post may be a bit general or vague on some areas.  But, hopefully, I’ll cover all of the major points about meshing.

When you work up a new simulation, prep work makes all the difference.  And the most-prominent piece of this work is creating a good mesh for your part or assembly.  The image at the top of my blog shows a part that has been meshed.  The idea of a mesh is to approximate the shape of a part by breaking it down into small blocks, called elements.  In the image above, most of these elements are simple cube shapes.  But elements can be cubes, tetrahedra, wedges, thin rectangular blocks, or a combination of these.

But meshing requires a fine balance between accuracy and simulation time.  On one hand, the smaller we make the mesh size, the more accurate our model will be represented.  Curved surfaces or edges will be closer to reality, because the more small flat facets we put around a curved face, the smoother it begins to look.  (Again, we’re only approximating shapes using elements.)  And stresses will transfer more smoothly through the parts.  But if you make the mesh really small to get nice, smooth faces and edges, you create a LOT more elements.  Each element uses equations to calculate how stress moves from element to element (or, more correctly, node to node…  nodes are the points where corners of elements come together… the intersections of the lines of the mesh, if you will).  Each element that neighbors other elements, requires several equations to be solved.  So it stands to reason that an increase in elements will cause a very serious increase in equations to solve.  But let’s take this one step further…

Let’s say you have a cube…  you cut that cube in a 2x2x2 mesh, meaning you now have 8 elements.  If we even go to a 3x3x3 mesh, we now have 27 elements.  Each time you decrease the mesh size, you create an exponential increase in the number of elements, and also equations to solve.

So, simply taking an overall average mesh size of 1-inch, and making it 1/2-inch, will figuratively make the number of elements we have to solve for explode.  And our solve time goes up dramatically.  It’s common for simulations to take hours to solve.  But something as simple as decreashing the mesh size could take that time and make it days!  So, we need to keep in mind the balance of accuracy versus time to solve.  At some point, decreasing mesh size will only very-slightly increase accuracy (say, by an additional 0.01%).  If my answer is 100psi of stress or 101psi of stress, does it really matter that much?  If the answer is no, there’s no reason we need that additional accuracy.

Here are a few examples of mesh sizes:  one that’s too large to be accurate, one that’s a good size, and one that’s too tiny to be beneficial.

Mesh size that’s much too large to be accurate.

Mesh size that’s just about right for this part.

Mesh that’s too small to be beneficial. Larger mesh size is adequate.

Both Autodesk Inventor and Autodesk Simulation products have what’s called a “convergence” utility.  The way that it works is it takes your simulation and runs it twice, once at the normal mesh size and one at a slightly-reduced mesh size, and compares the two results.  If the difference in results is small (you set what “small” means), the program will stop there.  If the difference is too large (say, over a 10% difference in the results), the mesh size steps down again and the simulation runs once more.  Again, it compares results from the last two runs to check for difference in stress.  It continues this process until the results “converge” toward a specific value, or until it runs out of the number of tries that it is allowed to run before giving up.  (Any more than around 4-6 mesh reductions and it probably won’t converge anyway.  Plus, this means it would need to solve repeatedly for each time it reduces mesh size, significantly increasing solve times.)  If the stresses do converge on a value, it means our mesh size is small enough to be reasonably accurate.  So, it’s best to start with a larger mesh size than you’d think is appropriate and work smaller from there.  If you start with a tiny mesh size, you may just be wasting time waiting for the program to solve.

To finish, a good mesh size for general stress analysis would be one where you have at least 2-3 layers of elements through the thickness of the material.  Again, start a little bigger, if you think you’ll end up with too many elements.  You can always test a second run with a smaller mesh size, or use the convergence utility to see how accurate your results are.  Otherwise, you could be waiting for hours or days for results that aren’t even useful anyway.  And every FEA package should have a way to refine the mesh in specific areas to increase accuracy only where you need it most.

Hopefully the idea of mesh size makes a bit more sense now.  Particularly the idea of how to set a mesh size appropriate for a simulation.  Feel free to post any questions you may have on this topic, and I’ll be happy to provide an answer, as best I can.

Poll: Do you use simulation..?

Just tossing out a quick poll to see who uses simulation in their design workflow.  I plan to leave this poll open indefinitely.  Please, only vote once.

Introduction

Welcome to the very first post on my blog!  I’ve never had a blog before, so this is going to be an interesting experience.  I hope that I can provide some great information, as well as provide a place for everyone to learn and grow.

The purpose of this blog is to post information regarding 3D modeling and simulation products.  (To know more about my background, click the “About” link on the menu above.)  Simulation software has tended to be pushed to the back of the line, especially with today’s economy, in that it’s extra money, training, and time being spent on software, rather than producing any tangible goods.  But that trend seems to be changing.  Companies want to streamline processes, cut costs, and bring products to market faster.  And THAT is where simulation software really comes into play.

So, why simulate, when you can physically test?  How does a simulation product do all of these magical things?

Simulation software can assist with designs, helping to work out some of the details before any material is shaped, formed, or cut.  Using the 3D model produced in CAD, a company can discover weak points or over-engineering, or even test multiple design possibilities, to figure out the best course of action.  This means fewer physical prototypes, resulting in a shorter design cycle.  Of course, the fewer prototypes, mistakes, and less testing that needs to be done also means cutting costs, in terms of both material used and man-hours spent doing the work.

Sim software can run just about any scenario imaginable.  The main product that I use, Autodesk Inventor Professional, includes a simple and easy basic environment to do static stress analysis, modal analysis (discovering natural frequencies of an object or assembly), and kinematic analysis.  It’s a great way to start off in the world of simulation.

But what else can sim software do?  If you move up into a more-comprehensive Finite Element Analysis (FEA) software, such as Autodesk Simulation (formerly Algor), you get so many more possibilities.  Here’s just a short list:

• Static Stress Analysis
• Dynamic Stress Analysis
• Modal Analysis
• Non-linear Material Models
• Permanent Deformation of Material
• Heat-Transfer Analysis
• Fluid Flow Analysis
• Electrostatic Analysis
• And more…

Of course, as capability and performance goes up, so does complexity of the software.  But the ability to simulate your product is out there.  And nearly any company can benefit from using it.  I whole-heartedly believe that.

Do you need to be an engineer to understand and use the software?  Not necessarily.  Inventor Pro’s sim environment is pretty simple and straight-forward.  Once you try it out, you’ll see how simple it can be.  But, a full-fledged FEA system is something that, for the most part, an engineer will be required to run.  It will have lots of options and settings, different solvers, and require input of various kinds to run properly.  But with proper training, it’s certainly possible for any technically-inclined person to learn.  And if you only need a one-time simulation, there are always consulting companies ready and willing to help you out.

So that’s it for my first post!  I don’t want to go overboard with it.  I still have lots of material to write, and intend to go over all sorts of info, from what FEA is all the way to understanding results.  But those will have to wait for another day…