Been Workin

We finished the testing in August and you may start to wonder – what the hell were we doing since then???   It’s the end of January 2016 now.   So sorry to have been absent, but we've been working hard!  We have been back to the lab again, doing a lot of detail designing and modeling.  We’ve taken the lessons from the load testing and empennage builds, and been busy incorporating them and fleshing-out the design of the whole machine. Lots of small changes, and a few big ones! Our whole design team went on a field trip to Titan Aircraft in Ohio in September.  John Williams spent over four hours of his weekend with us talking shop, showing us his whole operation and how they make every part.  John was very encouraging and interested in the Hellcat.  He is a real presence in the replica fighter world today as you all might well know, and we learned a lot from him. A primary reason to for the visit is that after some looking at it, we intend to leverage the landing gear components from the Titan 51 as a base for our own gear.  The landing gear of the Hellcat impact the layout of the center section in a big way, and without a solid design for them we can’t proceed with the rest of the wing.

View fullsize

We did get another distraction this fall in the form of more equipment.  My son Scott decided to buy a small vertical mill and convert it to CNC.  This “two week project” turned into 6 weeks or more of tinkering and fabricating.  Usually projects just do that.  This mill is no Haas or even Bridgeport, but what we need is the ability to machine small, complex parts without blowing the budget. In fact it’s quite a nifty little unit called a Precision Matthews PM-25MV, sold by Quality Machine Tool and Scott purchased a separate CNC conversion kit from a guy going by ArizonaVideo99.  It was a quality kit but more DIY than I was expecting thanks to the tolerances on import equipment.  Scott decided after getting the thing working and crashing more than once, to go ahead and wire-up good proximity sensors.  After two rebuilds it’s almost done now – just one or two cables to go.  Then we need to spend time getting used to what it can do.  But sorry to say, we didn't make more parts for the Hellcat, so we’re looking forward to get back to that this winter.  Back to work!

Loading Up

The tail frame is now finished and tested!  We’ve been busy pushing to finish the sample frame so we could find out if the design held up as expected. We would have been done sooner except, well, a little thing like OSHKOSH AIRVENTURE 2015 got in our way. It was a great event and we learned a lot and met some cool people.  We’ll have more on that shortly.  It did put us back a bit though.  Also as always you run into some delays for personal reasons and everybody in the team is super busy at home.  Anyway, we did it!

Finishing the Frame

First, we finished the tail frame with some bulkheads and hard mounts for the horizontal. The bulkheads were originally to be single piece with large cut-outs in the middle. Well, if you have a precision CNC bending brake, or a big press with forming blocks, you can do that and it will all fit up perfectly. We don’t have one of those, we have an eye-ball and hand adjusted bending brake, which meant that even with much care there was a little misalignment. So, realizing why other contemporaries seem to have mutli-piece bulkheads, we changed ours to be built with multiple pieces in a picture frame – like the Sonex and other designs do. This method worked out perfectly, as each piece just mates to the skins and then is clamped together for drilling and riveting the corners together, square and happy. Then as mentioned there are the mounting brackets for our horizontal/test load platform.  The original design was probably fine, but it had a few potential flaws that could be avoided with a simpler design just using some L-angle. This new approach comes out very rugged and while we had lightening holes in the wrong spot, that doesn’t really matter. The vertical on the other hand mounts up to a heavy-duty end bracket that ties the sides together. The ideal U-bend form of that plate was going to be a nightmare to bend with our limited shop tools so, we made it two pieces and hammer bent each side in a vice with a forming block. It’s not a bad looking install even so, and structurally it really made the aft end super solid.

Building the Test Fixture

When we finished the tail frame, we needed a test fixture to perform the load tests.  We designed and built a frame of 2.5” square steel tubing to support the tail off the ground and keep it stable.  To mate the test article to this test fixture, we made a mounting adapter frame.  This was a picture frame of steel tubing and plates to bolt and rivet into at the front of the test article.  This mounts exactly as the fwd fuselage frame will do.  Then, we built a load platform out of the same square tubing and capped with a deck of ¾ plywood sheathing.  This is mounted exactly as the horizontal will be and was easier to stack bags of gravel on than the horizontal would have been. Once everything was welded up, it was primed and painted a Rustoleum Rattle-can Blue. When dry, the mounting adapter ring was then riveted and bolted to the test article – permanently – just as the forward fuselage would be. The pieces were set before Oshkosh got in the way. But once we got back it was time to bolt the final assembly together and put it on the pad out back.

The Load Test

So we thought it over and decided the best way to perform the test was to load the frame in 100 pound increments and see how it did. The test was a very simple cantilever load setup, fixed at one end and then loaded on the other; similar to what anyone does, using a jack under the tail end as we loaded and then removing it to see how much deflection we would get.  We measured the height of the tail above the concrete pad and marked our ruler to set the base-line. For each load we waited over 30 seconds for each cycle and then jacked the tail back up to match the baseline before adding more.  After your 6G worth of load of 800 pounds, we removed all the weight from the fixture and removed the support.  At that point – there was no permanent deflection – zero.  Perfect!  We then loaded it back to 800 and then proceeded incrementally to 1200 pounds – 9 G’s.  Then we removed the loads and support again to check for permanent deflection.  This time she took a permanent set, but only 3/8”.  After this it was pointed out we also need an asymmetric load test – simulating rudder deflections and air-pocket/turbulence, and that at up to 9 G’s because why not?  At that load, it settled at approx. one degree of twist and then returned to level after the load was taken off, so nothing to worry about there it seems either. After that we decided to keep adding weight, see if anything started to change dramatically, so we went incrementally again to 1500 pounds – all the weight we had on hand.  This is about 11 g’s equivalent, and she survived, with ½” of permanent deflection.

To sum all that up, I feel we succeeded in meeting all our requirements: The frame survived our design load of 6 g’s with no issues.  She survived 11 g’s with no damage, just a permanent sag – It will get you home! But more importantly than the specific design we tested was that the available engineering predictions matched surprisingly accurately to the real world. Does this mean that FEA and basic calculations can replace building these test articles in the future? No! Well, not exactly; it does help add a level confidence to our abilities. I’m not the engineer but in many cases you can reasonably say a certain area is easily strong enough without going through a whole testing regimen to verify it; but it’s about having a method that can be followed to get some results that pan out in the real world, which applies to CAM and CNC as much as it does calculating structures and verifying design integrity.

The final takeaway though? We may be a bit too heavy duty.  The whole tail frame weighs in at 27 pounds which is right on the mark as designed, but there appears to be room to optimize. Then again, we like rugged airplanes.

Building a Box

So things have been moving along since finishing the Vertical, we're on to making a box. To elaborate, its a long sturdy box that will be the core of the fuselage frame. This build is specifically for a load test: the piece will be loaded to 9 g’s in increments – just as we did with our horizontal test piece. We are again looking to see how the sample will differ from the calculated expectation, and also whether this design as-is handles the load it will need to in flight, and to what degree over or under that is. I guess they are different aspects of the same problem; in my eye the pragmatic "does it work" viewpoint is the ultimate goal regardless, but having models agreeing to that end seems important. I'm not claiming to be the numbers man.

So before we even get into the actual fuselage stuff, I mentioned last update we were building a table from plywood and such. Well, yes, we have a table now. It's not a perfect plane, we diddn't plane every board to within .005 of an inch or anything, but it's about as dead flat as you need for building your aircraft on, easily less than 1/16 deviation over the thing. So here's what that looks like:

The aft fuselage that we're sampling here, goes from a bulkhead behind the rear seat of our main cockpit box, all the way down to the tail post. The design is very straightforward; 4 longerons made of of L-channel tie together 4 side-sheets to form a long box. Stiffeners and a few bulkheads are evenly spaced along the length to tie everything together, prevent buckling or twisting, and in the final full fuselage built the stiffeners actually act as attachment brackets for the outer skin bulkheads and stiffeners. Those will inevitably add extra support but the hope of the design is to make this box self-sufficient for the intended loads, and give us the flexibility to hang any type of outer shapes we wish. That's the whole objective here of ScaleBirds anyway right? Make it easy to go wild later.

For the build, I had to first generate good CAD data. I had a design in 3D already, but I had to further detail the aft part of the fuselage design in Inventor from where it was originally, mainly adding holes and details wherever it was needed. It took a while and I feel like the master-modelling method I'm using on the fuselage is reaching the limits of practicality; I started seeing crashes and long-long delays to update the feature tree after a change. At some point, I'll probably be having to go in and re-model the entire frame from smaller files once we have dimensions we like. The ideal would be to break it down to where each panel and part of metal gets its own part file, but it makes for a convoluted assembly where everything has to be interlocked anyway, so I figured doing it all in-situ in a single part file would give me greater flexibility while fleshing out the design and avoiding over-use of problematic reference dimensions. The good news is that unlike all the aerodynamic surfaces I've been having to deal with lately, I can directly model everything and export it to 2D as its all flat and I don't have to worry about curved-surface distortions and such. This means when I output the files into 2D I get true geometry that can be cut.

So it came to cutting out parts. I rigged up the patterns into CUT2D, and proceeded to cut the main panels from a pair .025 sheets. (The aluminum sheet we ordered from Aircraft Spruce, and chose to get it rolled for shipping which cut the delivery cost by a significant amount. I wasn't able to find much of any reference to ordering aluminum in this manner online, so we just took a gamble that the resulting sheet would unwrap flat. Turns out it does, it springs right back to flat condition as if nothing happened.)

So, Then we set to making some smaller parts that are made of thicker materials, flanges, doublers, etc. Some stuff is left to cut still, where we had to order more materials; but either way, the parts we did do came out looking great. 

And so we started doing some assembly. The process is pretty simple, though there's plenty of work involved. If you're building an aluminum plane I guess most situations call for a ton of rivets and this is one of them. It does make me think about going to something like a bonded construction: it's a simple enough form that I can see almost like pre-applied epoxy with tapes that you pull before assembly, or some kind of procedure like that. There's enough precedent that it might not be crazy. 

For now though, holes, rivets, repeat. 

As we finish this up over the next few days we'll post more pictures that show the overall form.

WELCOME!

Welcome to the new home of ScaleBirds.com 

This is the first post on our new website for the ScaleBirds Project. It's been a long time coming, and hasn't always been the first thing on our minds, but because Squarespace really makes it pretty painless to setup a website, we're able to get this up and running. I think having a place for the crew to share what's going on with the project will be good for both those involved in the project itself and those who are following along to see what is actually going into the Hellcat.

It's hard when in the midst of a project like this, which relies on multiple people who all are busy with life, to keep track of things on the macro and micro scales. But all elements fit together. Bringing this site online will help many things, even when the answer isn't obvious.

If you have been, thanks for reading, and visiting our project online. Feel free to leave comments about what you would like to see more of, and we'll be working on the pages and content for some time yet.

-SSW Out