Load Test Failure Aluminum Dragon

HomeBuiltAirplanes.com

Help Support HomeBuiltAirplanes.com:

proppastie

Well-Known Member
Log Member
Joined
Feb 19, 2012
Messages
4,167
Location
NJ
Check build log for pictures. As there is no substitute for experience, and it is practically impossible to obtain the experience short of a lifetime of work in the aerospace industry, I am not that upset......Its a retirement project, I have a excellent aircraft to fly when I want and am living the dream.

Possible Reasons for failure:

FEA model wrong...yes. Perhaps....

Inability to accuratly do a complicated "method of joints" by hand or spread sheet.....yes. However I have loaded Bruhn/Peery examples into the FEA (grape) and solved correctly .

The design is not properly executed so that the ribs do not flex and shift under load, which causes a condition/failure mode not properly modeled.

After spend the last 2 days playing with the model to remove excess constraints..... adding hinges, changing load conditions to accurately reflect the failure I understand a little bit more.

I am still not able to model the failure without adding additional constraints where I know the failure occurred. Currently the failure model shows PSI of 9K when the crippling failure of my angle is 18K.

Is it reasonable to change my allowable based on this test?....It does not seem correct to me.

As an additional data point....the original in wood has 2x the number of ribs...which in my hubris I only noticed about 2 weeks ago.
 

TFF

Well-Known Member
Joined
Apr 28, 2010
Messages
12,220
Location
Memphis, TN
Because you are building to a true minimum structure, your test fixtures may need to be as sofisticated as the structure you are building. There is nothing extra to help. You might have to cage the parts so there is no flex to test individual parts. Hard to tell if the engineering is right if there is testing problems. I am trying to encourage the engineering first. Proving is hard. Then you really know if it is too weak instead of a crutch right off. You may need more structure, let the testing direct you to true minimum.
 

pictsidhe

Well-Known Member
Joined
Jul 15, 2014
Messages
7,368
Location
North Carolina
Sheet it, even with paper, and it will be way stronger. You may have already allowed for that, though?
Might be worth investigating with mini tests on single pieces of angle. Simplify the test structure instead of elaborating the model to work out why theory isn't matching practice.
 

pictsidhe

Well-Known Member
Joined
Jul 15, 2014
Messages
7,368
Location
North Carolina
I just zoomed in on a pic. The angle doesn't look smoothly formed, it appears to have some edge ripples. That would trash the strength.
 

proppastie

Well-Known Member
Log Member
Joined
Feb 19, 2012
Messages
4,167
Location
NJ
not perfectly constructed, is an issue, I just do not know how much of an issue and how much the allowable is degraded. I seem to remember a pipe stress friend telling me when I stressed a medical device, "grooves and sharp corners 50 %"....
 

proppastie

Well-Known Member
Log Member
Joined
Feb 19, 2012
Messages
4,167
Location
NJ
Sheet it, even with paper, and it will be way stronger. You may have already allowed for that, though?
Might be worth investigating with mini tests on single pieces of angle. Simplify the test structure instead of elaborating the model to work out why theory isn't matching practice.
Will be fabric covered... where as fabric does offer strength I believe a "properly designed" structure should be able to withstand a limit test without it. FAA Glider Criteria does call out for testing with the fabric installed. Each of the 7 hinges support the spar, but the twisting moment gets stronger toward the root, notice the way the rib bent outboard.

Testing individual ribs will not be enough without being able to calculate that moment, and it is not the same as a simple spar moment in the Y (Z?) direction.
 

Aerowerx

Well-Known Member
Joined
Dec 1, 2011
Messages
5,355
Location
Marion, Ohio
Just curious...

I couldn't tell from your pictures, but are the joints 'pinned' or fixed? Wouldn't that would make a difference in how the forces are transferred from one piece to the next?
 

pictsidhe

Well-Known Member
Joined
Jul 15, 2014
Messages
7,368
Location
North Carolina
Fabric will add stiffness. Your failure was one of stiffness, not strength.
The joints are important on thin angle. Not loading through the shear centre adds bending. Loading angle through the shear centre is a PITA that few bother with. There may be a way to crimp the ends to allow it. Fixed joints will give you a headache trying to calculate/simulate.
How did you form the angle? Some of those longer bits look like they'd need roller forming, A brake would need using in stages, which is a PITA. I think I'd want a roller cutter and former if I was going to make that much angle.
Another thought, was your loading realistic? if you divided the load up into discrete lumps that would fit on a handy rib, those ribs wouldn't be loaded the same way as in flight, the front and rear of the aileron take some load in flight.
 

proppastie

Well-Known Member
Log Member
Joined
Feb 19, 2012
Messages
4,167
Location
NJ
Just curious...

I couldn't tell from your pictures, but are the joints 'pinned' or fixed? Wouldn't that would make a difference in how the forces are transferred from one piece to the next?
The joints are riveted, in the model they are calculated as beams welded at the centroid,...so fixed. and yes it does make a difference.
 

proppastie

Well-Known Member
Log Member
Joined
Feb 19, 2012
Messages
4,167
Location
NJ
Fabric will add stiffness. Your failure was one of stiffness, not strength.
The joints are important on thin angle. Not loading through the shear centre adds bending. Loading angle through the shear centre is a PITA that few bother with. There may be a way to crimp the ends to allow it. Fixed joints will give you a headache trying to calculate/simulate.
How did you form the angle? Some of those longer bits look like they'd need roller forming, A brake would need using in stages, which is a PITA. I think I'd want a roller cutter and former if I was going to make that much angle.
Another thought, was your loading realistic? if you divided the load up into discrete lumps that would fit on a handy rib, those ribs wouldn't be loaded the same way as in flight, the front and rear of the aileron take some load in flight.
There are pictures of the two bending braked in the build log..... The thin hardened material will bow no matter how you bend it, but you rivet it together in a fixture straight. I believe the minor assembly stress is not significant. The fabric strength is not figured in the design. The assembly is "in theory" designed to survive ultimate and tested to limit, so properly designed it should survive. The buckling is calculated using Gerard method Bruhn C7.16 Figure C7.7 for the formula. 1973 addition. The distribution is according to FAA Glider Criteria page 32 which is a triangular but in practical terms it is a uniform distribution to 1/3 the distance of the rib. The loading is 9 psf limit and designed currently to 13.3 psf,....I believe I originally designed it to 20 or 27 psf about a year ago and can not remember. Surly there was problems with the original design.
 

mcrae0104

Armchair Mafia Conspirator
HBA Supporter
Log Member
Joined
Oct 27, 2009
Messages
3,139
Location
BDU, BJC
Your failure was one of stiffness, not strength.
No, this is buckling. The definition of stiffness is commonly misunderstood, but the problem here is not stiffness. Stiffness has to do with the amount of deflection you get relative to loading.

Not loading through the shear centre adds bending.
Not loading through the shear center adds torsion, not bending. I don't mean to be too picky, but let's keep the terms straight.

PropP, are you loading the FEA model only at the joints, or distributing the load along the length of the top and bottom chords?
 

mcrae0104

Armchair Mafia Conspirator
HBA Supporter
Log Member
Joined
Oct 27, 2009
Messages
3,139
Location
BDU, BJC
The fabric may have held the angles and prevented buckling for the ultimate test load.
Yes, and it may have allowed the load to be distributed more evenly over the test article. It looks like the failure occurred right at the edge of the plywood.
 

proppastie

Well-Known Member
Log Member
Joined
Feb 19, 2012
Messages
4,167
Location
NJ
No, this is buckling. The definition of stiffness is commonly misunderstood, but the problem here is not stiffness. Stiffness has to do with the amount of deflection you get relative to loading.



Not loading through the shear center adds torsion, not bending. I don't mean to be too picky, but let's keep the terms straight.

PropP, are you loading the FEA model only at the joints, or distributing the load along the length of the top and bottom chords?
The test article was load with jugs of water and sandbags in boxes from the hinge line to the 1/3 distance. In practical terms it was a uniform loading from the spar to the 1/3 point over the rib and spread out inboard/outboard according to the size of the container. Build log show water jugs and box of sand. It failed just aft of a vertical support, the load was way forward of that point. That vertical support is at 50% rib cord, the weight sat from 0-33% uniform load.

The 3d model was loaded as either a point load at 1/3 on the top of each rib or triangular loading (result should be the same) for each rib.
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
6,289
Location
Saline Michigan
Proppastie,

Several things stood out to me. It appears that you anchored the hinges and the actuator, and then proceeded to load up the structure starting at the end opposite the actuator. Is this true? Next, the loads look big to me, like you started at the limit loads you were expecting. Is this true? If you are not extremely careful, you can inadvertently overlaod the part and get NO useful information from the test...

First piece of advice on this is to start small and work up on load. Second piece is to avoid all impact in loading the device. Third piece is to add weights in small increments.

Let's talk about each:

Starting small and working up means that you apply the well distributed light load to the whole thing and look for issues. Then you work up. It seems tedious, but it gives you an opportunity to see where deflections are bigger than anticipated and maybe a chance to detect parts deflecting out of plane. When you do get to a failure, you have upper and lower bounds on what fraction of limit load you were at when failure occurred. You were between 60 and 80% or between 62.5% and 75%. This allows you to beef up appropriately;

Avoiding impact is greatly underestimated. In Mechanics of Materials by Timoshenko, he takes us through setting a load on a column, where the load touches the column at zero load and then is instantaneously released. Peak load is 2.0 times basic load. While we may not be able to instantaneously let go of something, we can easily let it drop a short distance and get even higher load factors. Even very good setting can easily result in momentary loads of 1.5 times static. Support the part broadly until all of the dead loads are applied, then either remove supports or pick up the part by another means. In this case, I would anchor the hinges and support the aileron on the table, then for each load increment, use a bottle jack to lift up on the actuator until the aileron comes free of the table everywhere. Photograph and measure, then unload the jack and do the next load increment.

Small increments keeps all of the impact loads, local loads on the parts are kept reasonable and show you when something bends out of shape from them, and in general gets you as much information about how much you need to beef it up, and where.

There is one other way to do all of this, and that is a whiffle tree. Getting the loads spread out is the big trick.

Billski
 

wsimpso1

Super Moderator
Staff member
Log Member
Joined
Oct 18, 2003
Messages
6,289
Location
Saline Michigan
Will be fabric covered... where as fabric does offer strength I believe a "properly designed" structure should be able to withstand a limit test without it. FAA Glider Criteria does call out for testing with the fabric installed. Each of the 7 hinges support the spar, but the twisting moment gets stronger toward the root, notice the way the rib bent outboard.

Testing individual ribs will not be enough without being able to calculate that moment, and it is not the same as a simple spar moment in the Y (Z?) direction.

Hm, elastic stability appears to be your enemy on this part. In almost all fabric covered wings, there are diagonal braces between ribs to stabilize them, and then the fabric skin is applied too. These all keep the ribs from bending out of plane and collapsing. I do suspect that if you skinned the part, it would have done better.

Your observation about the surface accumulating load from tip to root also points the way to smart strengthening. The diagonal ribs could have large spacing at the tip and work with smaller and smaller spacing toward the root, tailoring the torsional stiffness to match the torsional load on the part. I am still concerned that a lack of rib stabilizing made your failure occur early. I would think about adding diagonal braces between ribs (Dacron lacing) and then a fabric skin, then do the job on test procedure per my other note on here.

Billski
 

proppastie

Well-Known Member
Log Member
Joined
Feb 19, 2012
Messages
4,167
Location
NJ
Proppastie,

Several things stood out to me. It appears that you anchored the hinges and the actuator, and then proceeded to load up the structure starting at the end opposite the actuator. Is this true? Next, the loads look big to me, like you started at the limit loads you were expecting. Is this true?
Billski
Yes the loads were the limit loads....I guess a big mistake.



Proppastie,


There is one other way to do all of this, and that is a whiffle tree. Getting the loads spread out is the big trick.

Billski
That would be nice....increments of sand sound easier.....be nice to have strain gauges too.

I think you have pointed out a big mistake which should help all of us down the road.
 

proppastie

Well-Known Member
Log Member
Joined
Feb 19, 2012
Messages
4,167
Location
NJ
Hm, elastic stability appears to be your enemy on this part. In almost all fabric covered wings, there are diagonal braces between ribs to stabilize them, and then the fabric skin is applied too. These all keep the ribs from bending out of plane and collapsing. I do suspect that if you skinned the part, it would have done better.

Your observation about the surface accumulating load from tip to root also points the way to smart strengthening. The diagonal ribs could have large spacing at the tip and work with smaller and smaller spacing toward the root, tailoring the torsional stiffness to match the torsional load on the part. I am still concerned that a lack of rib stabilizing made your failure occur early. I would think about adding diagonal braces between ribs (Dacron lacing) and then a fabric skin, then do the job on test procedure per my other note on here.

Billski
Looking at span-wise bracing in the 3d FEA model. Your comments point to stability problems the FEA did not calculate....Which makes lots of sense as per possible failure causes on OP

To any future designers....I believe it is so important to do these tests....it is too easy to make a mistake. You do not know what you do not know.
 
2
Group Builder
Top