Stress testing a Wing and Airframe?

Homebuilt Aircraft & Kit Plane Forum

Help Support Homebuilt Aircraft & Kit Plane Forum:

Armilite

Well-Known Member
Joined
Sep 5, 2011
Messages
3,517
Location
AMES, IA USA
I have a copy of LISA FEA software. There is a free version that is limited to 1300 nodes. It has been a while since I have used it, but I recall using more than 1300 nodes.

It seems to be OK on simple models but, as already pointed out, creating a good model is the problem. With simple structures, like cantilever beam, the results are what you would expect from theory.

With LISA you can create a structure out of "elements", and an element can be defined as a tube. This would make it easier to model a fuselage, instead of defining a gazillion node mesh.
==========================================================

Thanks, I'll check out this Lisa FEA Software. Basically, most of these Tube and Fabric Ultralight Airframes I'm talking about, use only about 4-5 different OD Size Tubes in either 6061/4130 Tubes in .058" to .065". They use only about 2-3 Size's of OD Bolts and Nuts. Tube Connectors are basically the same used on all Tube Airframes. The Grade of the Bolts and Nuts used are High Standard. Some of us do have Plans of different Airframes to get the Spec's from the Manufacture/Designer. It would be just nice to be able to compare say a Cabin Hoop in 6061 vs 4130 and at different Sizes and Wall Thickness. Also to compare Weight's of the different Parts.

Clound Based just means you can Share your Design with others in the Company or with others around the World if you give them Access. Windows 10 has it built in so you can access your Files at Home or at Work, or in a Motel room, your Car, etc., as long as you got a Internet Connection.

Only Files you Backup to your Cloud is Accessable, People still need your Password to get Access. I have about 9 Desktop Computers and 3 Laptops used in my house, Networked in my House and even one Desktop in the Garage. Eventually I hope to CNC my Lathe and Mill or just switch over to CNC Mill/Lathe Tools and will probably have a dedicated PC using Mach3 or Mach4 with them also.

Is this your Web Site? http://www.homemadetools.net/builder/Aerowerx I just signed on, haven't really looked at it yet. But looks like a good idea for Home made Tools. Another good Reason to be able to upload a CAD file here.
 

skier

Well-Known Member
Joined
Mar 5, 2008
Messages
1,079
Location
CT
Thanks, I'll check out this Lisa FEA Software. Basically, most of these Tube and Fabric Ultralight Airframes I'm talking about, use only about 4-5 different OD Size Tubes in either 6061/4130 Tubes in .058" to .065". They use only about 2-3 Size's of OD Bolts and Nuts. Tube Connectors are basically the same used on all Tube Airframes. The Grade of the Bolts and Nuts used are High Standard. Some of us do have Plans of different Airframes to get the Spec's from the Manufacture/Designer. It would be just nice to be able to compare say a Cabin Hoop in 6061 vs 4130 and at different Sizes and Wall Thickness. Also to compare Weight's of the different Parts.
As others on this thread have noted already, there really isn't a reason to use FEA on simple truss structures as are found on many ultralights. They're pretty easy calculations to run by hand if you spend the time to learn it. If you try to skip learning the fundamentals of stress analysis and jump right into FEA you're not likely to end up with correct answers. Further, you aren't likely to know that your solutions are incorrect. And, by the time you've learned the fundamentals of stress analysis well enough to understand if your FEA results are correct, you will realize that you can just do the calculations by hand in a fraction of the time it will take to make a model, mesh it, and set the boundary conditions.

The basics of structural analysis is pretty simple and generally boils down to a few concepts:
sum of the forces = 0
sum of the moments = 0
stress = e * epsilon
stress = Mc/I
stress = P/A
 

Aerowerx

Well-Known Member
Joined
Dec 1, 2011
Messages
5,754
Location
Marion, Ohio
Is this your Web Site? http://www.homemadetools.net/builder/Aerowerx I just signed on, haven't really looked at it yet. But looks like a good idea for Home made Tools. Another good Reason to be able to upload a CAD file here.
No, that is not my web site, but I do have a build log here in the member projects section. I have never seen it before.

That is my weight scale, though. I got the idea from a boat building web site, and improved on it.

I appears that they stole pictures from a post here on HBA.
 

Armilite

Well-Known Member
Joined
Sep 5, 2011
Messages
3,517
Location
AMES, IA USA
If your weld leaves a piece of 6061 T0 behind, you're using the wrong filler rod! I've read that welds using 6061 rod have cracking problems. I think on the Lincoln welding site. Ah, here it is:
http://www.lincolnelectric.com/en-us/support/welding-how-to/Pages/aluminum-design-mistakes-detail.aspx
They want you to use rod with extra silicon or magnesium. They also say it may be better to start with T4 and bake for an hour at 400 F.
---------------------------------------------------------
Grossing $1.6 million isn't going to leave much left over without a decent profit margin.
===================================================================

Well the Retail Cost of the Hirth 50hp Engine used is posted, and they get a Price Break from Hirth for a Profit Matgin, probably 15%.
F-23 W/Electric-Start & Slide carburetor $4959.00
F-23 W/Electric-Start & Diaphragm carburetor $5232.00
F-23 W/Electric Start & Electronic fuel injection $6589.00
Belt reduction drive $898.00
Optional oil injection $319.00

You can go on ebay and Buy a NEW Kawasaki/Suzki 440 Sled motor for a $1000 on Average, you still need to develope a Tuned Pipe and a Belt Drive or Buy an existing Belt Drive, but you could probably cut almost $3000 off the Hirth Setup. An Arctic Cat 440 Sled motor with the right Designed Tuned Pipe for 6500rpms can make 50hp at 6500rpms. It's probably also lighter than the Hirth by a few pounds.
eBay item number:371404523420 New Carb $300.

I agree, there is probably a low Profit Margin on most New Part 103 Ultralight's, probably in the $5000+ range. Just $5000 x 150 SOLD = $750,000, $5000 x 100 SOLD = $500,000, $5000 x 75 SOLD = $375,000, $5000 x 50 SOLD = $250,000. The T-Bird Factory when West of Des Moines was, Husband & Wife out front doing Sales/Parts and usually only (1) guy in the shop, 2 at most, when I visted on different ocassions.

Now all you need is Basic Hand Tools($200), a JD2 Model 3 Tube Bender and (3) Dies ($1500), a Press Brake($350), Cut Off Chop Saw($90) a cheap Mig Welder($450) basically less than probably $3000 in Tools/PC/Software, and some Office Space, and Space to make parts and Assemble them, 40ft x 60ft would be more than enough. I looked at a 40ft x 40ft Brick Building, with a 40ft x 60ft Morton Building attached to it all for $50,000, 6 months ago with enough land for a Grass Air Strip behind it. Building had a Hyd Hoist and Large Air Compressor, also had an upstairs and a Bathroom. I have seen cheaper and bigger Buildings for Sale.

All Sales depend on your Marketing Skills. Most of these Small Manufactures don't even have a Display Model or Show Room, if someone showed up to look at one.

Basically what I'm trying to say is, it doesn't take that much to start one up, you could do it out of your 24ft x 24ft garage to start and even with low profit margins you can still make some decent money! Today with some Shops having CNC Water Jets, CNC Lasers, CNC Plasma, CNC Tubing Cutters, it's probably cheaper and faster to farm some parts out if there close to you. Like my local small shop I can order 200 small brackets like the Red one above and have them Water Jet cut the next day. Many of these Parts are just made from just Tubing, or Extrusions, and Flat Stock. Like I said, there are only so many Parts you have to make, and usually only 4-5 Types. For making Ultralight's and many Kitplanes a 3ft x 3ft CNC Laser/Plasma/Water Jet cutter would be plenty big. Lot's of good used machines out there if you look. But even Parts for only 150 Airframes would be very low Volume, and could be done manually.
 

Armilite

Well-Known Member
Joined
Sep 5, 2011
Messages
3,517
Location
AMES, IA USA
No, that is not my web site, but I do have a build log here in the member projects section. I have never seen it before.

That is my weight scale, though. I got the idea from a boat building web site, and improved on it.

I appears that they stole pictures from a post here on HBA.
==========================================================

Ok, I'll have to check it out later. Always nice to find ways to make Tools that can make the job, easier, hopefully Cheaper to do, and Faster!
 

Armilite

Well-Known Member
Joined
Sep 5, 2011
Messages
3,517
Location
AMES, IA USA
As others on this thread have noted already, there really isn't a reason to use FEA on simple truss structures as are found on many ultralights. They're pretty easy calculations to run by hand if you spend the time to learn it. If you try to skip learning the fundamentals of stress analysis and jump right into FEA you're not likely to end up with correct answers. Further, you aren't likely to know that your solutions are incorrect. And, by the time you've learned the fundamentals of stress analysis well enough to understand if your FEA results are correct, you will realize that you can just do the calculations by hand in a fraction of the time it will take to make a model, mesh it, and set the boundary conditions.

The basics of structural analysis is pretty simple and generally boils down to a few concepts:
sum of the forces = 0
sum of the moments = 0
stress = e * epsilon
stress = Mc/I
stress = P/A
==================================================================

Without trying to make this to Technical here, Basic Common Sense goes along way in this World, my Point is, with this FEA Software you could upload say and original Manufactures Bracket lets say it's 3/16" 6061 and then do a Test with different Pounds of Force and it Color Codes the part to see the Weak Spots and if enough Force is used eventually Breaks, or all RED in the Software. Like the Simple Red 944 Porsche Tow Bracket is say 3/16" Steel. It's bolted to the Car Frame with 3/8" Bolts, and is used to Tow the car. If the 944 Car weighs say 2400lbs, is 3/16" Strong enough. We know that by going to 1/4" it is stronger, but is it really needed, could even 1/8" thick Steel be enough. Most things are Designed to a 3x Factor account of Law Suites today, not because it's really needed for the Design.
 

Topaz

Super Moderator
Staff member
Log Member
Joined
Jul 29, 2005
Messages
14,310
Location
Orange County, California
==================================================================

Without trying to make this to Technical here, Basic Common Sense goes along way in this World, my Point is, with this FEA Software you could upload say and original Manufactures Bracket lets say it's 3/16" 6061 and then do a Test with different Pounds of Force and it Color Codes the part to see the Weak Spots and if enough Force is used eventually Breaks, or all RED in the Software. Like the Simple Red 944 Porsche Tow Bracket is say 3/16" Steel. It's bolted to the Car Frame with 3/8" Bolts, and is used to Tow the car. If the 944 Car weighs say 2400lbs, is 3/16" Strong enough. We know that by going to 1/4" it is stronger, but is it really needed, could even 1/8" thick Steel be enough. Most things are Designed to a 3x Factor account of Law Suites today, not because it's really needed for the Design.
Unfortunately, the real-world of FEA software is not that simple. I certainly appreciate what you want to do, but the fact of the matter - said several times above - is that you have to really know what you're doing before structural analysis software will output results even remotely in the ballpark of the real world. In the hands of someone properly trained, it can drastically reduce design times. But you have to know what you're doing and, as other have so clearly said, if you're at that point, a simple tube-truss is child's play for doing it by hand.
 

Armilite

Well-Known Member
Joined
Sep 5, 2011
Messages
3,517
Location
AMES, IA USA
FreeCAD FEM Module https://www.freecadweb.org/wiki/index.php?title=FEM_Module
Biggest problem with FreeCAD is that it's advancing so fast that it's nearly impossible to keep up.
==========================================================
Thanks I'll check out FreeCad. Didn't have much luck with Lisa with only 1200 nodes in Demo mode I couldn't get it to do anything with even a simple bracket. Got it to open some Stl files, but had trouble with Step files. In Demo Version I would rather see a Full Functioning Program with so many Day Time limit rather than a crippled version that won't really let me see what it can do.

Off Grabcad an example of a part.
 

Attachments

Matt G.

Well-Known Member
Joined
Nov 16, 2011
Messages
1,253
Location
Kansas, USA
That image you've posted is a perfect example of what I'm talking about- the two holes at the top appear to have 'fixed' constraints applied to all of the nodes inside of them (not realistic) and the load is applied unrealistically to an entire planar surface (note the cluster of force vectors under each of the lugs). Then there's the Von Mises plot, which is used mostly by people who don't know what any of the stress output options mean, so they use the default. Maximum principal stress would be a better choice. A pretty picture that is unlikely to provide any useful information.
 

Aesquire

Well-Known Member
Joined
Jul 28, 2014
Messages
2,793
Location
Rochester, NY, USA
I haven't used finite element analysis in decades, and it was part of the suite that was packaged with our dedicated CAD/CAM computer system in the machine shop. Remember track balls?

It was great, but for simple stuff like a warren truss, hand calculation was faster and less confusing than the software. I'm sure the current software is wonderful, and the hardware is a quantum leap. We used to brag we had better computers than NASA used on Apollo. ( true, but today, so is my phone ) Then again, I still have a slipstick in the drawer. It works when the batteries die and the power is out.

One quibble. $5000 profit? That's for the year, right?

True, the required tools are pretty basic. I've seen plenty of home workshops that can easily make all the bits for a Quicksilver..... excepting the plastic saddles, which are easier to just buy from Leading Edge Airfoils. ( and today by the time your 3d printer makes you a complete set, the Post Office already delivered the ones I ordered..... And, yeah, I've made those from delrin with a lathe & mill, until they became commercially available. )

I don't remember anyone back in the 1980's etc. doing CAD or FEA on their airframes.

Often the first load testing was done when somebody got curious, like the Easy Riser, ( once by the "factory" long after sales started, using an old, clapped out trainer they traded for a brand new kit. And again by a dealer, who was hanging engines & other parts on them who wanted to be sure he was doing it right. I was there for that one. ) or some fellow tried to register it in England and they demanded a test to approve it. Not all passed.

Usually in the bad old days, people just copied the one they saw at an airshow and would hang off the frame like a jungle gym, after building it, to show off "how strong it was". ( They were the guys who went to airshows with a notepad & tape measure, and were a "factory" 4 weeks later )

And, after years making parts in job shops for the Lockheed Redacted and the Sikorsky Classified, ( And occasionally reading the entire pallet of documentation attached to each part ) I became a big fan of breaking stuff.

Computers only verify what you know on stuff very close to what you already did the hard way. If there's something you didn't know, neither will the computer.

Finally, take the burrs off, put a radius on that, and pile on the sandbags. It's the only way to be sure.;)
 

Jay Kempf

Curmudgeon in Training (CIT)
Lifetime Supporter
Joined
Apr 13, 2009
Messages
4,302
Location
Warren, VT USA
The commercial CAD packages now have built-in FEA, but they are not particularly useful for analyzing aircraft structure. You have minimal control over mesh and connections between various pieces of structure. You can make pretty pictures, but they are likely to be garbage.
Even the less than adequate CAD integrated FEA is pretty useful now. For just general hot spots in a mesh most do fairly well. You need some advanced stuff for doing first to fail laminate schedule stuff.

Generally the OP is lumping way too much stuff together. If you have limited tools you have to cut the problem down to size and learn what individual parts are doing. Can be done. But you have to understand first how to apply aero loads and then how to react them at the part level. Not hard but tedious in a gratifying sort of way.
 

pictsidhe

Well-Known Member
Joined
Jul 15, 2014
Messages
8,812
Location
North Carolina
I learned how to analyse structures by hand and by FEA at university nearly 30 years ago, I've not bothered with FEA since. As others have said, if you can't do it by hand, you shouldn't even bother trying FEA.
 

ScaleBirdsScott

Well-Known Member
Joined
Feb 10, 2015
Messages
1,281
Location
Uncasville, CT
Interesting stuff on FEA from people who've done it.

I think there's a lot that can be gleaned by a designer for rough passes using what's out there today. But I also will concur that there are depths to it way beyond that level that will yield good engineering results. I've had people explain it to me and how to get basic results, enough to say whether further analysis is really warranted in a specific area. Of course I'm not an engineer, it's those things it starts doing at the extremes that go beyond the nuance of my understanding; but the truth is today, with a 2015+ CAD package with built-in simulation, there's very little that can't be taught to a CAD monkey to do the work on those systems despite not knowing the long-form math. It's about knowing the concepts and principals and how things need to be setup. I leave it to engineers to look at the results, but I'm definitely able to set up and run the job and get what come out as valid results, and even I can generally look it over and make sense of it on some level.

Anyway for those who are still thinking about FEA as it was in 1995, or even 2015, Just look at how far things are progressing now. I'm pretty-sure that in a few years we'll have simulation software that can largely replace the current 'static' FEA for most design purposes. Already there's real-time simulation engines that can manage material deformation and how things will crumple and distribute loads, and that's when trying to calculate cars crashing into eachother at 200mph. Is it 'accurate'? no, not really. Is it unable to become accurate? I think it's simply a matter of applying the right math and programming at this point. Clearly we have the computational horsepower to spare. (and to think we're burning major portions of it on cryptocurrency)

I get the distinct impression from the way things are gong with this stuff that in a few years, a simulator that can just do real-time load calcs based on taking a model of your object and applying loads in an intuitive way as they would in real practice will be available, and it'll mean the average person can do credible simulations. Does that exist in some lab somewhere, right now? I would put money on it myself. But I'm thinking in terms of 'download this today and be analyzing things in an 'intuitive' way by dinnertime.'

That won't make everyone an engineer. But it will be such that the average guy using a 3D printer will be able to determine if they need to add more layers or try ABS vs PLA with a simple virtual model and a virtual hammer. Furniture designers will be able to figure out if a 300lb music critic dancing in his seat will break the legs, and kids will use it to see what happens when you smash a schoolbus with two dumptrucks, and then make a living out of streaming it on Twitch. (well, they're already doing the latter)

But we'll be able to take stuff like that and see if our virtual wings will tear off at 11G if we use 2024 vs 6061.

Meanwhile engineers who learn the latest stuff will be hopefully using it for loftier goals, like analyzing Mars landing vehicles and structures.
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
7,954
Location
Saline Michigan
If your weld leaves a piece of 6061 T0 behind, you're using the wrong filler rod! I've read that welds using 6061 rod have cracking problems. I think on the Lincoln welding site. Ah, here it is:
http://www.lincolnelectric.com/en-us/support/welding-how-to/Pages/aluminum-design-mistakes-detail.aspx
They want you to use rod with extra silicon or magnesium. They also say it may be better to start with T4 and bake for an hour at 400F.
Actually, if you use anything but a 6061 rod, the weld nugget won't be 6061 anymore. And no matter what rod you use, the metal near the weld will be close to 6061-O right after welding. Let it sit around for weeks to months and it will go to near a T3 or T4 state, but without a very specific heat treat cycle, it can not get back to T6 level.

Billski
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
7,954
Location
Saline Michigan
First, this ain't testing, this is analysis and simulation. If you do a good job on design and then on analysis and simulation, when you run tests on actual parts under actual load cases, there should be no surprises. Let me tell you, there are still surprises, sometimes big ones. Which can show you how how bad some folks are at any or all of design, analysis, simulation, and testing.

One of the things both professors I learned FEA from emphasized was "how do you know your analysis is any good?"

The design folks should have some idea of what stresses, deflections, and Eigen modes they should have, at least at some places or under some loads. Simply compare the apriori estimates to the ones that come out of the FEA. We frequently have a base product that we know worked OK in the field, and so we did the same type of analysis on the old and the new and compared them. Of critical importance in using this method is using the same type of boundary conditions, as they can make huge differences. Some load cases can not be checked by any reasonable method, but you can usually do a case that is easily run on a hand calc that is not far off actual loads... Another way is to run upper and lower bounds and checking that your stresses are between them.

Billski
 

flyboy2160

Well-Known Member
Joined
May 25, 2014
Messages
363
Location
california, USA
...
That won't make everyone an engineer. But it will be such that the average guy using a 3D printer will be able to determine if they need to add more layers or try ABS vs PLA with a simple virtual model and a virtual hammer...
Right, it won't make everyone an engineer, which means "the average guy" can't really know whether the " just push the button and have it done by dinnertime" answer is anywhere near being correct.

You must always do some type of closed-form hand calculation to have some idea if the computer "push button" answer is correct. As one of my first instructors said: The most important calculation you do is that first order one without a computer to get the answer to one significant figure in the right order of magnitude. I want to know if a rocket going to the moon is going to weigh 6 tons, or 60 tons, or 600 tons or 6000 tons, or....

I've even seen real engineers fall into the "Hand calcs? We don't need no stinking hand calcs." fallacy. That is a sure way to get people killed and to lose lots of money.

There are many engineer wannabes and engineering denigrators here who just don't want to put the time and mental effort into learning engineering. Pushing that computer button isn't a substitute for learning the principles.
 
Last edited:

pictsidhe

Well-Known Member
Joined
Jul 15, 2014
Messages
8,812
Location
North Carolina
A custom built analog computer will often still produce the quickest and most accurate results. Build a model, plaster it with strain gauges and load it up. Accuracy will depend on how accurate your model is. Obviously, testing the actual object will give the most accurate results! Computer simulations often make assumptions. It's the accuracy of those assumptions that can make or break the simulation. Having some idea of what are valid assumptions is where the skill lies. your computer can get that very, very wrong.
If I'm designing something, I like to get the answer by two separate methods. Usually, one is accurate (hopefully), the other is a sanity check.
 

narfi

Well-Known Member
Log Member
Joined
Feb 23, 2016
Messages
1,035
Location
Alaska
Isn't that a part of common sense that we learned in grade school called 'estimating'?
You know before doing a problem approximately what the answer will be, you just don't know to the exact number......

To simplify, you can know that 105/10 is going to be slightly more than 10.
You know this because you have studied division and you know that 100/10 is 10, so 105/10 must be just slightly more.

You couldn't 'estimate' that answer if you hadn't studied division and understood how it worked and what to expect when you sat down with pen and paper or a calculator to do it.
As you got older and deeper into math you were able to get the right answer in your head even without a calculator. You know that 100/10 is 10, and you know that 5/10 is 1/2, so you know in your head that 105/10 = 10.5

I imagine this is much the same way. Studying the basics should allow you to somewhat 'estimate' an answer that you can compare with an actual calculation.

My problem is I haven't studied any of this, so I not only can not estimate an answer, but I don't even know how to enter it into the calculator to compare with my estimation.
 

BBerson

Light Plane Philosopher
HBA Supporter
Joined
Dec 16, 2007
Messages
14,955
Location
Port Townsend WA
My dad taught me to estimate with 10 like that. I am not sure if it was ever taught in school.
 
Top