Product Development Notebook » chris loughnane

FEA: Saint-Venant’s principle

chris loughnane — Wed, 25 Jan 2012 04:52:58 +0000

If you’ve done FEA, you know that you can get some funky results near boundary conditions.

Fixing a face such that absolutely zero motion is allowed, for example, can create high stress concentrations that may not be real. However, thanks to Saint-Venant, we don’t usually have to care.

Because…

“… the difference between the effects of two different but statically equivalent load becomes very small at sufficiently large distances from load.”

So the high stresses caused by the load I’m applying at a point don’t affect what’s happening elsewhere, and so I can still have faith in my results.

It’s this principle that enables you to solve a model whilst only cranking up the mesh density (to get convergence) in areas of interest; If Saint-Venant’s principle weren’t true, your model would be pitch black with mesh lines.

frequently held false beliefs in product design

chris loughnane — Thu, 19 Jan 2012 22:22:57 +0000

I recently posed the following question on Quora

What are the most frequent false beliefs in mechanical engineering and product design?

I find the answers compelling and hope you do as well.

Design doesn’t depend on context
The designer knows more about the user than the user does
The idea is the hard part
The “light-bulb!” theory of idea generation
The “lone genius” fallacy
It will work the first time
Everything you learned in school is important
Nothing you learned in school is important
If it looks good in CAD, it is good.
All manufacturers are the same
Electical > Mechanical
Gut > Math
Once a prototype works, getting to production is easy
Prototyping is easy

And some of my own…

Prototyping is unneccessary
Prototyping isn’t worth it.
That anything is truly off the table. (On more occasions than you might think, one person at the client will claim a certain feature or date is sacrosanct, only to be overrun by someone else at the client in order that the product needs to have this other feature or hit some other date). Use “off the table” with discretion.
That decisions should be made NOW (making decisions as late as possible (although not later) means you make them at the point when you have the most knowledge about the project.)
That Finite Element Analysis doesn’t need to be coupled with physical testing

Add yours in the comments.

video: quantum levitation

chris loughnane — Sat, 14 Jan 2012 18:27:46 +0000

Came across this today. They even tease you with just enough of a physics explanation to get you curious.

How it works

According to the website, the ceramic, once brought down to a low enough temperature, becomes a superconductor (more like a super-duper conductor as they claim zero energy loss).

Now that it is a superconductor, the magnetic field attempts to avoid it, so where it would usually penetrate the disc, it now goes around it for the most part. In some places it penetrates the disc where the superconductivity is weak. These weakened locations are discrete (i.e. grain boundaries), so they act to constrain the disc from moving.

Awesome.

why I’m studying for the patent agent exam

chris loughnane — Sat, 07 Jan 2012 22:15:06 +0000

Someone asked me the other day why I was studying for the patent agent exam.

It’s a good question. Here, in no particular order, are my reasons

As a design consultant, I regularly am exposed to the IP of my clients and their competitors. Understanding where the IP “white space” of a particular field is a critical input to designing a marketable solution.
Having the skills to be able to draft and prosecute (legal-speak for “process”) a patent application means that if I have (or a friend has) an idea that might be patentable, I don’t have to cough up $4,000 – $10,000 to have someone else check over it to make sure it is properly constructed (the importance of which cannot be overstated, a poorly written patent is worthless). Having a really strong patent can be the difference between getting your startup funded or… not.
It’s low-hanging fruit. If you a bachelor’s degree in a “recognized technical subject” (which includes pretty much any engineering degree… full list on page 4 of USPTOs General Requirements Bulletin), all you have to do is pass an exam.
I like to broaden my skills. CAD, tolerance analysis, mechanism design etc. are still my bread-and-butter, but the uncertainty of what lies a decade or so down the road has me thinking it’s a good investment to diversify my most valuable asset: my ability to work.
If I ever become so-inclined (I am not), once I am a registered practitioner, the only thing that stands between me and a career as a patent attorney is law school. Can’t be any tougher than engineering school.
It’s a credential. Some companies will pay you more just because you have a credential (which I think is silly and lazy of whoever is evaluating talent and performance, but I digress), so that can be nice.

Those are my reasons. Bottom line, I’m convinced the cost (both money and time) is less than the benefits.

What do you think? Am I viewing things through rose-colored glasses?

If you’re interested in studying for the exam yourself, check out this study aid I put together. It’s better than anything I could find

No related posts.

badass trajpar equations

chris loughnane — Mon, 02 Jan 2012 14:00:51 +0000

if you never use pro|e, stop reading lest you get a serious case of CAD envy.

If you do use pro|e, yet of never heard of trajpar (short for “trajectory parameter”), it is a parameter that varies from 0 to 1 along the path of a sweep (0 at the start, 1 at the end). Combining this relations gives you an amazing level of control over your model.

For example, if you want a pure, trigonometrically pure transition from one value to another (comes in very handy when you get into high-end surfacing).

The reason I bring this up is that I’ve recently had occasion to some surfacing that need be both scalable and smooth. Under such circumstances, I have found the following to be immensely helpful. (pdf here)

If you have any questions about trajpar, or if you think you may be able to use this in a model but aren’t sure how, hit up the comments. I could talk about trajpar for hours.

this has always bugged me.

chris loughnane — Mon, 02 Jan 2012 00:26:51 +0000

In fact, 1 fl oz of water weighs 1.04 oz. (Full math at bowl of plenty)

design we love

chris loughnane — Sun, 01 Jan 2012 19:36:26 +0000

I’m loving the design we love Board on Quora.

This teaspoon in particular caught me eye.

my sporadic obsessions

chris loughnane — Sat, 24 Dec 2011 17:46:42 +0000

This one struck a chord. Via stuff no one told me.

No related posts.

some notes on nonlinear fea with contact

chris loughnane — Mon, 19 Dec 2011 14:29:18 +0000

So I haven’t had an opportunity to be much cool FEA lately.

Still, I’m pretty regularly in reviews where we are reviewing setup, discussing solving approaches, analyzing results etc. It seems lately that most of these analysis involve a fair amount of contact (the presence of which, I have mentioned before, is an indicator that you may want to be use a nonlinear FEA approach), and I have taken notes.

Two of the notes that jump out:

Friction is a mf

There are several mathematical models that can account for it, but friction seems to regularly cause mathematical chaos (i.e. asymmetric matrices due to the non-associations of Coulomb’s law… blah). If you can’t converge to a solution, consider lower your friction down to 0.0x

Augmented Lagrange > Pure Penalty

Both are methods for solving contact problems. The problem they are trying to solve is that while in real life contacting parts don’t “invade” each other, in FEA models individual nodes can encroach.

The Pure Penalty method deals with this by applying a stiffness such that the further a node tries to encroach, the greater the resistance (like a spring). The problem with this is, to be perfectly accurate, you would have no encroachment, which would necessitate infinite stiffness. This displacement and stiffness result in a contact force. Numbers like that can make the maths difficult to solve.

The Lagrange method solves for force directly, treating the force as a separate D.O.F (think Free-Body-Diagram). This theoretically means there is no encroachment of elements, but because is so close to positive and negative numbers, this can become unstable. To account for this, a “maximum allowable penetration” parameter is used.

The augmented Lagrange method is a sort of combination. It uses the Pure Penalty method in the beginning, but as the stiffness becomes larger and larger (i.e. more and more difficult to solve) some of that Lagrangian math gets loaded on there to take off some of the burden. You can read more about the math here, but suffice it to say that, for moderately complex geometry, the augmented lagrange method seems a good middle ground.

Bottom line: I would use augmented Lagrange.

That’s all for now.

protect IP PSA

chris loughnane — Tue, 13 Dec 2011 17:08:47 +0000

I don’t usually (and don’t plan to often) use this blog as a medium for politics, but I feel strongly enough about the Protect IP act to share this.

Essentially the bill would allow the entertainment industry to censor the internet; destructive in itself but perhaps setting an even more dangerous precedent.

From fightforthefuture.org

It’ll give the government new powers to block Americans’ access websites that corporations don’t like. The bill would criminalize posting all sorts of standard web content — music playing in the background of videos, footage of people dancing, kids playing video games, and posting video of people playing cover songs.

This legislation will stifle free speech and innovation, and even threaten popular web services like Twitter, YouTube, and Facebook.

The bill was just introduced: We need to act now to let our lawmakers know just how terrible it is. Will you fill out the form above to ask your lawmakers to oppose the legislation?

This video explains the situation well. If you feel as strongly as I do, contact your congressman.

PROTECT IP / SOPA Breaks The Internet from Fight for the Future on Vimeo.