Collaboration?

[Speedboat100]

Speaking from the perspective of a mechanical engineer (but not the kind who has done much practical work with structures), one oft-overlooked part of any new design is testing, testing, testing. Good test engineers are worth just as much as good structural engineers.

For instance, my background is autonomous flying robotics (aka drones). I can-- and have-- written a lot of autopilot code but I don't really have the easy ability to test it. (It's hard to test flight code from a city center, or when the season is inclement, or when I don't have access to the right kind of aircraft.) Some of my happiest days were writing autopilot code that others were then able to go fly that very minute. I had a tight collaboration with my testers.

But a good tester isn't just a crash-test dummy. Test engineering is not done seat-of-the-pants any more than structural engineering is. It is, however, much more experience driven than theory driven. So it can-- and can only-- be learned outside of formal classroom environments.

The upshot is that if you can figure out how to offer excellent test skills to a designer, the designer might be able to offer you excellent design skills in return. If this sounds useful to you, my advice might be to develop and market the kind of test skills that we're looking for. Then its win-win for all involved.

I agree 100 %...you have to work on it.

I was just wondering as I have a structural innovations that would it possible to make them into ½ scale and run loads accordingly in relation to the scale.

Can the test results from exactly similar structure be valid also in 1:1 scale ?

The benefit of this would be less work in the actual testing...also I find it very difficult to find several hundreds of kg:s of weights to actually make 1:1 testing for a wing .

To put is shortly...could I assume the real plane is safe if the ½ scale model of it has been tested to + 10 G and -5 Gs ?

 

[drgondog]

As I read through the posts I couldn't help but sympathize with your dilemma. First, to introduce myself in a fashion - I have BS and MS Aero. and 8 years industry (Bell, Vought, co-op Lockheed) experience learning how to apply the degree analytics to real airframes. Core degree studies were Structures, Aerodynamics and Performance. That said, I requested (and got) my first assignment in Airframe Design because I wanted core experience in working with several key groups - Mfg/Prod, Structures - to synthesize the analytics to match real world processes.

If I understand what you want to do - is design, build, test and fly your creation? If my understanding is correct I see interesting challenges that extend past 'math checking'..

I assume that you have done the preliminary design work to propose, lay out and place the engine, fuel tank, cockpit, airfoil type, wing plan form, empennage (conventional), landing gear basic, cockpit enclosure, battery, etc. At this stage you assume that you will gather and solve external loads for design of internal structure and arrangement?

You might assume an elliptical loading for pressure distribution spanwise on the wing, table look up for section properties and Aspect Ratio deviation from the 2-D theoretical lift and drag, to arrive at preliminary spanwise load distribution for later spar design.

External aerodynamic loading (minimum) considerations are AoA (i.e. dive pullout), lateral loads imposed by yaw input, and gust loads to design to, as well as landing loads - imposed on wing, empennage and fuselage. Later you will re-iterate on aileron and flap loads, as well as theoretical stall characteristics. Empennage and wing sizing, wash out and aileron sizing are subject to modification during both the low speed and the stability and control analysis part of your design.

The loads imposed on empennage are necessary for preliminary sizing of longerons and shear panels and bulk head design all the way through the cockpit area to the engine mount bulkhead (IMO). Longeron sizing should consider both engine thrust and prop torque loads as well as bending and torsion lads from empennage.

So, at this stage weight distribution projections are critical to locate center of gravity for empty and normal and max gross weight conditions - which will be essential for location of the wing 1/4 chord and positioning for the horizontal stabilizer and subsequent stability and control analytics.

So far, what you have is an envelope with weight items tabulated from a datum point, maybe spinner/fuselage interface. You have a notion of wood, sheet metal fabrication, rivet/weld/bond methods for assembly based on your skills and experience. If fabrication choice is sheet metal and rivets you are now thinking about assembly processes including jigs and forms.

As you begin on detail design, you are simultaneously thinking how that part/section will be fabricated and integrated.

Based on what I read above, you don't have an actual way forward to a.) gather the aero loads or b.) perform structural analysis on any part of the airframe - either static or dynamic, or c.) perform stability and control analysis to locate and size empennage relative to the cg and wing/airfoil section characteristics.

That is a problem.

The Wright brothers and early pioneers didn't have degrees or the training so as a practical matter, neither do you. That said, I would think that placing your butt in the cockpit should make you a little cautious about proceeding. If I seem to lecture I apologize - there are guys on this forum that can offer more than I have above.

 

[SVSUSteve]

Too bad. Times change, usually as a result of legal action.

As a student, I designed, for credit, a structural load test for a proposed STC.


BJC

The funny thing is my wife is an attorney and would have been the one handling the review of any waiver on my end....talk about a very defensible document in court should something happen. But whatever....it is what is it.

 

[SVSUSteve]

A diploma alone does not make an engineer.

One of the professors I know (and asked to help before his department told him that he could not) commented that he was very surprised to learn about my disability. It's not something I exactly mention to most people. It has always been a sore spot for me ever since my second grade teacher called me a 'retard' and said I would never become a scientist because of it. There's a joke about the fact that a major motivator of my pursuing a PhD is revenge. LOL

I took his comment of the fact that I would have made "one hell of an engineer" otherwise-- because of being able to see that it's not just the calculations that matter but also human factors, ability to build, maintain, etc that matters-- as a very high compliment.

 

[wsimpso1]

Speaking from the perspective of a mechanical engineer (but not the kind who has done much practical work with structures), one oft-overlooked part of any new design is testing, testing, testing. Good test engineers are worth just as much as good structural engineers.

For instance, my background is autonomous flying robotics (aka drones). I can-- and have-- written a lot of autopilot code but I don't really have the easy ability to test it. (It's hard to test flight code from a city center, or when the season is inclement, or when I don't have access to the right kind of aircraft.) Some of my happiest days were writing autopilot code that others were then able to go fly that very minute. I had a tight collaboration with my testers.

But a good tester isn't just a crash-test dummy. Test engineering is not done seat-of-the-pants any more than structural engineering is. It is, however, much more experience driven than theory driven. So it can-- and can only-- be learned outside of formal classroom environments.

The upshot is that if you can figure out how to offer excellent test skills to a designer, the designer might be able to offer you excellent design skills in return. If this sounds useful to you, my advice might be to develop and market the kind of test skills that we're looking for. Then its win-win for all involved.

This reply and several others are thread drift...

That being said, I feel it is necessary to point out that structures have been around a long time and folks have been figuring out how to analyze them almost as long. No one wants to find out their bridge design sucks after the bridge is up. For the most part, there should be no surprises in structures unless you are trying to design within a gnat's eyelash of failure. In the vast majority of instances, gnat's eyebrow margins are not needed. More than that, our aero structures have multiple load cases, most with multiple loadings superimposed upon each other. Load tests are rarely able to fully duplicate the full load situation.

So what do we do? We analyze in ways that do combine the loads and check the failure criteria. We analyze previous successful products the same way and see how they do for comparison. In a production situation, we do a couple types of serious tests, one that simulates the biggest single stressor from real life, the other that simulates running the product through a lifetime of load cycles. Three possibilities exist:

• It runs well past the analytical predictions - congratulations, it is sturdy to this load case and is likely OK on the lessor cases too. It is possibly heavier than it could otherwise be, and management will likely direct a review for weight and cost savings;
• It fails somewhere near where your thought it would - congratulations, you just demonstrated that you know what you are doing and that it is likely OK on the lessor cases too. Doubtless management will direct you to look for cost savings;
• It fails way short of prediction - You now have a serious problem. You can fix the thing that broke, but how many other understrength parts must go through "find and fix"? How many more cycles of build and test and adjust design again are needed? Will it be done in time to launch a successful product? Besides jeopardizing the entire project, it means the structures team failed.

Test of structures is meant to be a confirmation that the design is adequate after competent folks did the work, not a way to find all of the places where they guessed and/or failed.

Moving to other types of systems beyond structures, just imagine trying to do 4000 systems that all had to work to land people on the moon. You do not really think that most of those systems were crappy and then "find and fix" was applied until they might be OK to trust lives and treasure on, do you? No, most of them where schemed up to work pretty reliably just so the whole thing would have a decent chance. Even then, they screwed up and had to invent new stuff on reliability analysis of systems. But that was over 50 years ago, and Failure Modes and Effects Analysis has made its way into all kinds of design and systems too.

Billski

 

[stanislavz]

I could really benefit from having someone who is more experienced with structural analysis help me out. I can handle most of the other aspects-- although, as has happened before

Not to being sarcastic - you will make life much easier, if you provide load data to engineers - to be like spar is loaded at lb/foot, skin loaded at shear and surface etc.. Otherwise it is asking to develop airframe by provided sketch..

 

[ragflyer]

Testing as a primary means (not just final validation of structural analysis) is a legitimate approach to proving airworthiness particularly in amateur designed build aircraft. It is particularly indispensable when novel (experimental ) structural forms and or materials are used (for example hollowed out foam core wing in a recent thread). It has also been relied on significantly in new designs such as the mosquito bomber when a non traditional approach was taken.

In the UK virtually all experimental/homebuilt aircraft (outside the new unregulated UL class) have to meet an airworthiness code. Prof. Guy Gratton ran this at the PFA/LAA for many years. He is a rare professional in a field otherwise dominated by amateurs. His quote below from the book, "initial airworthiness" is based on approving/analyzing a large number airplanes

I have advanced degrees in engineering and am well versed in analytical methods. I have spend many hours back analyzing wood/metal and composite airplanes. In a separate thread I will enumerate my thoughts but suffice to say using pure analysis is very challenging for homebuilt composite airplanes unless one stays very close to form of existing designs.

The reality is test driven validation is a valid and very practical approach to amateur designing safe airplanes for a large number of folks on this forum.

 

[wsimpso1]

I agree 100 %...you have to work on it.

I was just wondering as I have a structural innovations that would it possible to make them into ½ scale and run loads accordingly in relation to the scale.

Can the test results from exactly similar structure be valid also in 1:1 scale ?

The benefit of this would be less work in the actual testing...also I find it very difficult to find several hundreds of kg:s of weights to actually make 1:1 testing for a wing .

To put is shortly...could I assume the real plane is safe if the ½ scale model of it has been tested to + 10 G and -5 Gs ?

You are talking about scaling, well scaling means making all of the stressors work the same. Trouble is the many stressors work in different proportions. Let's get into a few of the possible issues. If you make the entire thing half scale, the abilities to resist certain loads and deformations change in differing proportions...

Tensile members will have same stress at one-quarter the load;
Elements in bending/torsion will have same stress at one-eighth the load, but the same angular deflections at one-sixteenth the load;
Columns in compression will have the same ability to resist buckling at 4 times the load;
Weights will be one-eighth, but spring rates will vary by other fractions, greatly changing natural frequencies of the systems.

That is just a few of them. You can not just scale things... You would have to figure out the loads for the small size system, then design the structure to carry the loads per your principles of your "innovations".

If you have something that really does not show up on a Patent search, and really has advantages, do not share it on here. If you intend to use it or make any money on it, your first public disclosure really ought to be a filed Patent. First secrecy, second do a Patent search, third a Patent filing. Good luck!

Billski

 

[ragflyer]

Elements in bending/torsion will have same stress at one-eighth the load, but the same angular deflections at one-sixteenth the load;
Billski

I think you meant to say one eighth the moment rather than load- the load will be a quarter just as in tension. The lever arm is halved and so the moment will be an eighth.

For beams in bending/torsion such as a wing spar and even wings you can absolutely scale and test. If you go half scale all dimensions including thickness would have to halved and the load will be a quarter of full scale. Ofcourse it may be difficult to scale thickness in practice (minimum gauge available). There may be a few secondary effects but in the main can be valuable. A lot of the design calculations used in industry are empirical in nature and they have been validated by scale testing.

 

[addicted2climbing]

I think I asked this once before but the project was shelved before it went anywhere due to other things coming up on my end that took precedence. I am getting to a point where I have more free time to get back to working on a design but the same issue creeps up.

Those of you who have spoken with me privately, many of you at least, know that I happen to have a math learning disability. It does not impact my ability to do research since I can understand the concepts and rely on statistical analysis software to do the actual calculations. I cannot exactly do that with the structural analysis of an aircraft design and I am not willing to bet my life that I got it right (let alone the lives of anyone else) even if it is something that I have another person look over.

Would any of you be interested in collaborating on the design of a new light aircraft that is designed to be as safe as possible while still practical. The current plan is for a metal tube fuselage with wood wings and tail. This mix of materials is simply because I would rather build with wood. The basic layout is complete more the most part and many of the systems have been laid out to a decent degree (figuring that it helps to know weights and placement before working on structural detail design).

I could really benefit from having someone who is more experienced with structural analysis help me out. I can handle most of the other aspects-- although, as has happened before, I will be picking many of your brains for things related to engines, electrical, and aerodynamics-- but the structural detail work is something that I keep stumbling on.

Hello Steve,

I too am in the same boat and am working as an Engineer for 30 years but learned on the job and mostly self taught. I know what I need to know for my work and industry, but have many holes in my knowledge when it comes to statics and such. I asked nearly the same question years ago. One reply turned me onto the book "Stress without tears" its more fundamentally based knowledge for calculating aircraft structures and written in more laymen's terms, but I will say I was still stumped a bit and that was usually fixed by asking here as many on here have the book as well. Also bought many aircraft plans that are similar to study various methods of construction. I figure its enough for me to get very close on my own one day conceptually then could once again reach out for help here as things become more finalized.

Oh and I so hear you on the learning disability with math. I am so lazy now that rather than use trig. I just draw the geometry in solidworks and dimension it to get the angles and lengths...

I am about to build a cnc router in the spring 2021 with hopes of designing a simple single seater that will use a Briggs Converison.

bets of luck to you

 

[Mad MAC]

Not to being sarcastic - you will make life much easier, if you provide load data to engineers - to be like spar is loaded at lb/foot, skin loaded at shear and surface etc.. Otherwise it is asking to develop airframe by provided sketch..

Hmmm, thats tends to the big company specialist approach, at normal GA level best results come from the generalist approach, if its insufficently strong, one can apply a better stress anaylsis, a better load analysis or redesign the part without getting up one ones desk.

Given Steves high level of safety focus, the first step would be to write the engineering standard that he wants to use (AKA copy and paste from FAR23, FAR27, etc) so that the implications on the stress analysis can be readily established consistently by any collaborators. Many of the safety requirements be the critical load cases, negating the requirement to more than roughly calculate the flight loads.

Structural test is just like FEA and hand calulations its just another method to pick from, all must be considered with the resources to hand.

 

[kubark42]

This reply and several others are thread drift...

I apologize if my comments struck you that way. My understanding is that @SVSUSteve isn't looking for advice on how to design a part, so much as he is looking for someone else to do it. That's a very fair approach, since he offers his talents in exchange for others' talents. (I'm sympathetic to his plight, I too have some skills useful in homebuilt aviation but not all the set of skills required.)

I think all the people in this thread offering design guidance is great, because maybe that will work for him, and then he can go it alone. Teach a man to fish kind of thing. But I feel his original question hasn't been unanswered yet so I took a stab at describing how I might find the designer, instead of how I might design it.

 

[Speedboat100]

I think you meant to say one eighth the moment rather than load- the load will be a quarter just as in tension. The lever arm is halved and so the moment will be an eighth.

For beams in bending/torsion such as a wing spar and even wings you can absolutely scale and test. If you go half scale all dimensions including thickness would have to halved and the load will be a quarter of full scale. Ofcourse it may be difficult to scale thickness in practice (minimum gauge available). There may be a few secondary effects but in the main can be valuable. A lot of the design calculations used in industry are empirical in nature and they have been validated by scale testing.

Right this is what I am after...I may have some advantage in optimizing as I don't have to brake 8 x heavier structure/machine. Also easy to fix in the next model as the work is lesser...and way cheaper.

 

[Geraldc]

If you intend to use it or make any money on it, your first public disclosure really ought to be a filed Patent. First secrecy, second do a Patent search, third a Patent filing. Good luck!

With the problems of a person like myself protecting a patent I would rather post it here so everyone can use it.Open source principle.
First challenge is to find something worth patenting.
Edited spelling and grammer other mistakes.

 

[stanislavz]

Hmmm, thats tends to the big company specialist approach, at normal GA level best results come from the generalist approach, if its insufficently strong, one can apply a better stress anaylsis, a better load analysis or redesign the part without getting up one ones desk.

It is and it isn't. Yes, in next 20 years, you will load your napkin sketch to big computer, and it will print you your dream airplane with minimum weight penalty.

Now - you divide and conquer. For wing - it is safe to assume, wing with one spar is build from on flat plate (spar) and four curved panels with know radius. Each part gets loaded. Skins - surface load (from lifting) and shear load (from traveling C.P and ailerons), Spar is got tension and compression on caps and shear on the web. Plus you have ribs which have to translate lifting force to spar, and difference pressure between top/bottom skin.

Yes - it is wood thinking of some kind, but you are able to build one or two section of wing, and test all components. And sandbags are not your friends anymore. Or just run numbers for preliminary testing.

 

[drgondog]

For those of you that are curious regarding 'old school' structural analysis (good from 1930s to 1970s) of airframes, look to the MISC folder in the Smithsonian P-51 drawings package. I used the same methods in late 60/early 70s as NASTRAN and Stardyne were emerging.

500 page P-51D-5-NA Structural Analysis - including bomb and pylon loads.

Also included (imbedded in the report is the P-51D-5 Performance Analysis - included is detailed parasite and pressure drag and Thrust Hp and exhaust thrust calcs.

As to 1/4 scale for Static Test? Staggers the imagination regarding building one to scale, much more when contemplation of sheet/frame buckling results..

Fine for preliminary lift and drag data if you have $$ for wind tunnel testing.

 

[Speedboat100]

For those of you that are curious regarding 'old school' structural analysis (good from 1930s to 1970s) of airframes, look to the MISC folder in the Smithsonian P-51 drawings package. I used the same methods in late 60/early 70s as NASTRAN and Stardyne were emerging.

500 page P-51D-5-NA Structural Analysis - including bomb and pylon loads.

Also included (imbedded in the report is the P-51D-5 Performance Analysis - included is detailed parasite and pressure drag and Thrust Hp and exhaust thrust calcs.

As to 1/4 scale for Static Test? Staggers the imagination regarding building one to scale, much more when contemplation of sheet/frame buckling results..

Fine for preliminary lift and drag data if you have $$ for wind tunnel testing.

How about ½ scale ? Same thing ?

 

[drgondog]

How about ½ scale ? Same thing ?

yes - but the reason is practicality as well as dimensional scaling to achieve the same stress. It will not be 1/2:1 for dimensional combinations to achieve same Stress threshold for tension, compression, shear, bending or buckling for 1/2 design load.

For example - if full scale contemplates 0.040 skin and 1/8" flush rivets, you can't countersink in 0.020 skin for a flush 1/8" rivet, and 1/8 rivet would probably be oversized for even 2x rivet spacing in .020 skin. To achieve the same stress levels in 0.020 skin due to new (1/2 scale) shear loads, the rivets have to be smaller pan head rivet sized to accommodate 1/2 scale spacing. Another factor is that rivet spacing and edge spacing are different between pan head and countersunk rivets.

Secondly, the structural analysis demands performing Stress calcs due to applied loads to allowable stresses, at or below yield stress for the material (or per the tables), for the Limit Load condition. To achieve this across your entire airframe, a.) you need to be a structures expert to begin with on your 1:1 scale airframe/components - where presumably you have designed to standard sheet and extrusion thickness for say, 2024, and standard diameter rivets and commonly available extrusion/tube dimensions. Ditto for machined stock and fittings and bolts.

When you 'scale down' , but hold same desired stress margins of safety at 1/2 scale loading, you will be 'juggling' smaller scale (area/dimensions) components (such as an I or H or T section extrusion for a spar build that does not exist in stock form at the smaller dimension) in combination with smaller (not necessarily 1/2:1) dimensions, to achieve the full scale stress allowables at half load.

Bolt sizing is based on tensile loads applied against bolt area, but shear loads are against X-sectional section the cross section of the adjoining shear area of the other part - and they don't scale simply by choosing 1/2 of the area or the original diameter for both conditions.

The more sophisticated your foundation airframe is (i'e' retractable landing gear vs fixed), the more complicated "1/2" scale sizing becomes to select components that scale properly with respect to holding calculated Stress allowables the same for 1/2 applied loads.

Your Static Test plan is another 'interesting' component at any scale, much less reduced scale - when the desired outcome is finding out where the airframe 'starts to bend' - to all the way to 'done broke it'.

For anyone building 'just because you want to' - and not for series production in a competitive market where weight and performance based on cost/price - build it like a bridge that will easily take +3G/-1G for symmetrical AoA loading and 1.5X margin for all components directly related to flight safety.

Smarter guys than me have already commented on the above issues to scaling down dimensionally to prepare for Static Testing.

My head hurts - I did all this professionally, but I'm not sure I could not reproduce those skills today If I was building an airplane for which I would assume responsibility for another person's safety.

 

[Speedboat100]

yes - but the reason is practicality as well as dimensional scaling to achieve the same stress. It will not be 1/2:1 for dimensional combinations to achieve same Stress threshold for tension, compression, shear, bending or buckling for 1/2 design load.

For example - if full scale contemplates 0.040 skin and 1/8" flush rivets, you can't countersink in 0.020 skin for a flush 1/8" rivet, and 1/8 rivet would probably be oversized for even 2x rivet spacing in .020 skin. To achieve the same stress levels in 0.020 skin due to new (1/2 scale) shear loads, the rivets have to be smaller pan head rivet sized to accommodate 1/2 scale spacing. Another factor is that rivet spacing and edge spacing are different between pan head and countersunk rivets.

Secondly, the structural analysis demands performing Stress calcs due to applied loads to allowable stresses, at or below yield stress for the material (or per the tables), for the Limit Load condition. To achieve this across your entire airframe, a.) you need to be a structures expert to begin with on your 1:1 scale airframe/components - where presumably you have designed to standard sheet and extrusion thickness for say, 2024, and standard diameter rivets and commonly available extrusion/tube dimensions. Ditto for machined stock and fittings and bolts.

When you 'scale down' , but hold same desired stress margins of safety at 1/2 scale loading, you will be 'juggling' smaller scale (area/dimensions) components (such as an I or H or T section extrusion for a spar build that does not exist in stock form at the smaller dimension) in combination with smaller (not necessarily 1/2:1) dimensions, to achieve the full scale stress allowables at half load.

Bolt sizing is based on tensile loads applied against bolt area, but shear loads are against X-sectional section the cross section of the adjoining shear area of the other part - and they don't scale simply by choosing 1/2 of the area or the original diameter for both conditions.

The more sophisticated your foundation airframe is (i'e' retractable landing gear vs fixed), the more complicated "1/2" scale sizing becomes to select components that scale properly with respect to holding calculated Stress allowables the same for 1/2 applied loads.

Your Static Test plan is another 'interesting' component at any scale, much less reduced scale - when the desired outcome is finding out where the airframe 'starts to bend' - to all the way to 'done broke it'.

For anyone building 'just because you want to' - and not for series production in a competitive market where weight and performance based on cost/price - build it like a bridge that will easily take +3G/-1G for symmetrical AoA loading and 1.5X margin for all components directly related to flight safety.

Smarter guys than me have already commented on the above issues to scaling down dimensionally to prepare for Static Testing.

My head hurts - I did all this professionally, but I'm not sure I could not reproduce those skills today If I was building an airplane for which I would assume responsibility for another person's safety.

Right....local FAA was on these lines as well when I asked a week ago. But making the structure in a model could still provide ample information.

Also many supersonic VT models are very tiny....maybe even 1/16 sized.

 

[drgondog]

Right....local FAA was on these lines as well when I asked a week ago. But making the structure in a model could still provide ample information.

Also many supersonic VT models are very tiny....maybe even 1/16 sized.

aerodynamic modeling is a whole 'nuther dscussion - and easily to scaling based on Mean Aerodynamic Chord.