3d printed fuselage plug for a new moulded F5J

[CTR 81]

Some of you may know that John Hovell and I are tinkering with design and build of an F5J. To that end, the split in resources is that John does the wings, I do the Fuselage. Not wanting to reinvent the wheel so to speak, we have extracted some very useful ideas and methods from Colin & Kevin and the Pro Glide project (many thanks to them for reducing our workload )

Getting to the point: I successfully produced a plug (the master shaped solid model) of the fuselage using a 3d printer which Colin thought may be of interest to others crazy enough to walk this path.

As the plug is used to produce the mould for laying up the final job, the finish needs to be as good or better than the finished fuselage. I think this is the main point Colin was found of interest as 3d prints don’t have the best surface finish; possibly the size management was also of interest. So here goes!

This is a list of initial considerations: (and a taster of things to come).

1/ Basic design for lending itself to 3D printing and best form/function (only my opinion so don’t shoot me down; just take any positive ideas and adapt them as you wish)

2/ 3D filament to use for the best results/minimum work.

3/ What printer to use.

4/ How to create a long plug from a smaller printer.

5/ Finishing methods for the surface.

6/ Laying up the mould.

7/ creating the final fuselage. (Not there yet!)

I’ll also cover the pitfalls encountered on the journey to help others avoid wasting time and cash.

As my time is limited, and I’m no expert at typing, these points will be spread over separate posts in the next few days. Apologies to those of an impatient disposition, but family life comes first!

Photo is the first incarnation as an appetiser.

Neil P

 

[CTR 81]

Here we go with the stage 1:

Looking at numerous fuselage failures, it appears the mass of the tail feathers, with it’s inertia causes a high percentage of failures just in front of the tail area. This leads me to thinking the typical round/oval section is not the ideal for resisting these forces. It also appears that failures at slower speed occurs when there is also a wing tip ground contact involved I.e. a spin/sideways twisting at moment of impact.

Resultant drag from different sections is typically small (assuming similar cross section areas) so a roughly triangular section was settled on. This gives a flat top with two small radiused corners for strength where compression failures could occur, and a semi elliptical lower shape which would be under tension at moment of impact. Adding a significant fin skeg also helps with the transition from the mass of the tail to a relatively thin boom. The logical step was to incorporate 40mm of fin base into the fuselage design, leading to selection of an all moving stab too. As far as I’m concerned, the aerodynamics improvement in an AMT is probably countered by the less than perfect joint between the stab and fin sides, but nothing’s perfect!

The flat top also lends itself to top wing seating and a better transition at the trailing edge.

Working forwards, the triangularish section under the wing is not conducive to gripping during launching so grip groves were designed under the wing seat.

Typically the hatch is also a weak area and is the main restriction in width for gear/ batteries etc. Logically, continuing the same fuselage shape forwards would help here too. The top edges of the hatch cutout need to be as straight as possible and contain a corner, all be it radiused (stronger in compression than the typical curved hatch sides). Also stringers of carbon tube can be moulded in along these edges from nose to wing leading edge. Result is that the hatch is almost as wide as the fuselage max width and very stiff.

Transitioning to round at the nose finishes my theory of a good shape for form and function, strength and stiffness. No doubt others will have differing views which I would like to hear.

I’m fortunate in having a powerful CAD system able to create the 3D model and able to save it as an STL file ready for 3D printing software.

[CTR 81]

2/ What filament?

Two reasonably priced options here: ABS and PLA. PET polycarbonate and Nylon could also be considered but more expensive and not easy to smooth. The rationale would be to use the lower cost, more common filaments if they are fit for the purpose.

The dilemma is that ABS is great for chemical smoothing using acetone vapour = less issues with prep’ for a good finish but is more difficult to keep it stuck to the printer bed; whereas PLA sticks much easier but has to be mechanically smoothed.

First tests were with ABS as it is tougher than PLA (less brittle) and easier to smooth. However, this very fact means it is not able to withstand the resin mould materials. My thought was to use a painted coating to protect it from the chemicals in the resin. Good theory but even after many days, the solvents in the paint are still being released and cause a breakdown of the release agents used in the mould making process. Not a happy bunny if the plug gets stuck in the mould!

Next was to go with PLA and accept the need for mechanical smoothing (lots of elbow grease).

A word of warning here: There is a new trend in PLA filament, often referred to PLA+ or some such. These are modified by mixing other polymers with the PLA to make it ‘better’ for 3D printing. This is to be avoided at all costs because they are usually attacked by acetone; whereas pure PLA is not affected. If in doubt, put a short piece of filament in a small jar of acetone and leave it submerged for 30 minutes or so. If it’s pure PLA it will survive this treatment, otherwise you will have a sticky mess or at least a sticky surface to the filament. Although I’m not a particular fan of HK, their own brand of PLA is good to use and reasonably priced.

OK, so we’ve settled on PLA; now we need to minimise the elbow grease needed to get a smooth finish. Here’s where we go to the next step of selecting the best printing method for our purpose. Probably tomorrow’s posting for that info. We will get there in the end.

[Tony 18]

Fascinating Neil, way above my league of construction, really like the fuselage details and why you've made it that way. Looking forward to more of this.

Tony

[CTR 81]

Stage 3 & 4 What printer & handling the length.

Well, first did some trials using my cartesian style. 3d printer with mixed results. Issue is related to the geometry of the parts.

I thought it would save time and effort by splitting the fuselage model lengthwise down the vertical centreline, then add a flat surface in that plane to form the split line surface for the mould. Hope you can visualise that! This means that the print for the first half is made up of a flat rectangular base with a half section of the fuselage protruding out of the top. Good for adhesion to  the printer build plate, and simple to lay up the first half of the mould.

Reasonable start and logic except for the following:

Consider the typical fuselage shape; it is roughly circular cross section with slow tapering changes.  Referencing this to the way a 3D printer works with layers at say 0.1mm thickness. A long taper in the Z plane perhaps increasing by 0.1mm in the Z for 10 to 15mm X or Y travel results in an exaggerated stepped surface in the final print. This takes quite a bit of smoothing to remove the staircase effect.

Logically the next step was to print a test section with the fuselage vertical. This way, its like layering a big stack of 0.1mm formers together to make up the fuselage. As each layer can be printed to a shape accuracy of better than 0.01mm, the resulting steps between layers is going to be much smaller. Test results backed up this theory.

Following thought trail was to invest in a delta type 3D printer because these are great for taller, roughly cylindrical shapes. This was the best solution I came up with.

First photo is the nose section being printed vertically, having added a base raft to ensure it stayed stuck to the printer bed. 2nd photo is of the much modified, stiffer Anycubic Kossel plus 3D printer used for the job.

The printing height limit of the delta printer is around 300mm, so the Fuselage was divided into lengths  within that limit. I chose locations for the section divisions that would be easy to join and smooth. To add accuracy to the joints I added small peg holes in the mating surfaces and a triangular section rib (see photos 3 & 4).

More to follow.

[Marc Sinclair 27]

Amazing work and project Neil! 

Thank you for sharing.

Will you guys be producing and selling your designs in the future? 

[CTR 81]

Thanks for the encouragement guys.

Production and selling isn’t the driver for this project, just the challenge of doing it! That’s what happens when you retire from a life of design and manufacture. Somehow the creative instinct doesn’t die with the last pay packet!

However, let’s see how well it turns out and reevaluate the next step then. In the mean time, I’m fairly sure I speak for John too that we hope some of our ‘research and development’ will be of benefit to others.

BTW I’ve named my design files R&D1 ( rev4 so far) so RAD1 is probably as good a name as any for the glider. This name has personal history for me too; Long time ago when I ‘dropped out’ of a high pressure computer related career, I started a kite making business (about 1990) when serious kites were rising in popularity. Not the cane and paper variety  but carbon frames and advanced ripstop fabrics etc. At that time there were very few ‘inflatable’ kites like Jalbert ram air parachutes. I designed a kite using this principle under the design name of R&D1.  A 3.5M span elliptical wing plan job and it pulled like a train! That was the design goal, so hope the glider also meets the mark too.

[ThermalBoy 81]

Fantastic work Neil. Very innovative use of 3D printing. The sense of satisfaction when you fly your first home built moulded model is one you and John will never forget. I'm looking forward to seeing how you smoothed the surface of the PLA. I made one small 3D printed mould (Tailplane support pod) using my Anycubic i3 Mega Printer in PLA which I still use today but getting it smooth was very difficult.  I've been fortunate enough to see a preview of the moulds you have made from these plugs and they look terrific.

I feel sure that 3D printing is set to become an ever bigger part in producing composite planes/components going forward even for the professional manufacturers.

Keep the great work up.

Colin

[CTR 81]

OK, so here we go with step 5 which is all about getting that finish.

As already said, building the 3D print vertically is probably the most significant step in reducing the graft of smoothing.

PLA filament is also quite hard which is better for sanding as the paper does not clog so quickly as is the case with ABS, Nylon etc.

Start by carefully trimming the typical flash edge where the print and bed meet. This is caused by a small spreading of the first layer of print where The filament is forced out between the nozzle and the bed; the first layer being slightly less height but extruding the same amount of filament. This is to help with the bonding of the print to the bed because we are printing with a relatively small bonding area and a tall print. The last thing we want is any wobble as the print builds in height. Another advantage of the delta type printers is that the print bed is static meaning there is far less chance of print oscillating as would be the case with a moving build platform. Also the mechanical load on the moving printer mechanism is a constant load whatever the weight of the printed part.

Next I used a very sharp wood chisel with the bevelled side up, flat side on the surface of the print. Sliding this over the surface removes a surprising amount of the ribbing left by the printing process.

I chose to do a lot of the smoothing on each separate part prior to gluing them together. This was learnt from experience with the plug shown in the first post. This was glued first, but the thin section of the boom split several times during over exuberant sanding! Each time it needed repairing, the new glue joint meant more smoothing work. On and on 

When doing the separate parts, be careful not to sand the ends! Leave 15-20mm to be smoothed once the parts are joined. However, if you do make a mistake and dig into the surface, baking powder and thin CA makes a great filler as it is very similar to the hardness of the PLA. Very fine edges to the fill are possible meaning they won’t show on the final mould.

As for sanding, I used a finer side of a  Permagrit 300mm tool, the extruded aluminium one with the 2 different grits on it. Use it lengthwise on the part until there are virtually no raised rings around the part. This is OK to use where the surface is flat or convex (bulging outwards) but not to be used where it is concave; these areas must be done with 80 grit paper. I used silicon carbide which has a good life on PLA. This stage took about 30 minutes in all.

Next, change to wet and dry paper, 120 grit, used wet. Even then, be careful not to work in an area too long as it may get warm resulting in  small rolls of PLA forming between the paper and plug. Use the 120 until there is no more evidence of the 80 grit score lines. A tip here is to sand in a slightly different direction with each finer grit. This makes it easier to see if all the previous grit marks have been removed.

After the 120 grit I glued the parts together. I used 80 grit again over the joints, then the 120 grit until the whole surface was to the same roughness.

After that, work up in grit levels 180,  220, 300, 400, 500, 600, 800, and 1000. Don’t skip any if you want the quickest progress! It only takes about 5 to 10 minutes per grit, but much longer if you try and skip even one of the grit levels. Also make sure you use fresh water and wipes each time you change grit to ensure there are no coarse grit particles to mar the finer surface. I only needed a piece of each grit paper about 100mm wide torn off the width of an A4 sheet. This did the full plug. 

When finished with the 1000 grit, go to using a rubbing compound with cotton rag. I used Paiboat NW1 Premium from Easy Composites. This worked really well and can be used on the finished mould too. It took about 30 minutes for the polishing. That’s it for this stage.

Hats off to anyone who reads through all my ramblings. I hope you get something useful from your perseverance!

Next instalment will be mould making.

[CTR 81]

Just had a thought that someone may want to know the 3D print settings I have settled on for these results.

I used 0.1mm layer height with a 0.3mm nozzle. 

6 solid base layers and top layers with 5 perimeters. 10% infill with a 2 layer multiplayer (I.e. it prints infill every 2 layers which reduces the print time).

Each fuselage section took from 10 hours to 17 hours depending on the volume.

Temperature for the HK PLA was 215 deg and I have the bed at 50deg but that’s not essential if you don’t have a heated bed.

Speed was 100mm/sec.

The plug was destroyed in removing it from the mould because of the week infill. If you want to reuse the plug it would be better to use 30% infill.

I have done a trial since doing this one using 0.05mm layer height. This was less work for smoothing as you can go straight in at 120 grit wet & dry but the print time is just over double. As I wanted to get results in a short time, the 0.1mm layer meant I got parts off quicker and could start the smoothing process while the next was being printed. Overall a quicker process.

As for gluing the parts together; PLA isn’t good for gluing. I drilled a load of 2mm holes in each mating face of the individual prints. Using slow araldite and the 2mm carbon pegs gave a reasonably strong joint as the glue was squeezed into the holes in each surface giving a mechanical joint as well as the glued surface. Don’t bother with the quick setting glues as they remain rubbery.

[CTR 81]

Stage 6: making the mould.

There’s a good video on line by Easy Composites on mould making. The process is universally virtually the same.

I thought their idea of using polypropylene sheet as a split surface would save a lot of work. How wrong I was! The theory is good in as much as resins will not stick to PP.. However, it isn’t very stable and expands significantly with heat; and heat is generated as the chemical reaction takes place as the resin and catalyst react. See photo 1. It clearly shows the rippling along the edge despite the buttons along the sides, which started off flat, and this was only the gel coat. You can see the wood buttons in photo 2 and 3 which wasn’t close enough to the blue plug to stop the ripples. The soft wax pyramids in photo 3 are covering some of the screws and used to ensure accurate positioning of the final 2 parts of the mould. PP tape visible on the remaining screws.

I had to peel off the gel which proved that the release agent worked, and start again.

Anyway, the theory is to have a split surface that will divide the mould into parts where the resulting moulding can be removed from the mould. Obviously the plug has to be removed too once the mould has been made. 

Previously I have used melamine covered chipboard as the split line sheet without problems, but the quality of these now is way below the old standard. Still useable but you will end up with a mottled surface which in this case shouldn’t be detrimental. I stuck with the PP for this mould but had to add a significant amount of wood batons screwed to the back of the PP sheet to stop the rippling. This gave small dimples where each countersunk screw head was, but that was tolerable. Each screw was covered by PP tape to stop the resin sticking in the screw heads.

Photo 4 shows the completion of the first half of the mould. This must be well waxed before the second half of the mould is built up on top. Don’t be tempted to remove the plug from the first half to check if it is good. It will be impossible to ensure it is replaced exactly to ensure accuracy of the second half of the mould.

[CTR 81]

Despite carefully sanding and polishing the plug, there may be blemishes that went undetected. However, visual inspection of the polished black mould surface will show up these areas. Don’t despair! The mould can easily be flatted and repolished in these areas (see photos).

Photo 1 shows the areas I was not happy with. Photo 2 is a closeup of an area sanded with 400 grit wet & dry. Notice the main smoothing was done lengthwise because the blemish was across the fuselage; but then I used a circular motion. Photo 3 is the sanded result using 600 grit W&D, again used wet. Notice that the sanding strokes are only lengthwise and done until all the circular marks from the 400 grit sanding are removed. That’s the reason for the swirls otherwise some deeper 400 marks may remain, discised buy the 600 finish. If there, they would be very obvious when it is finally polished.

Last photo is the polished finish achieved by using the rubbing compound used in step 5.

Note that in photo 1 there are semi elliptical areas in places along the sides of the mould. These were made by applying sheet wax pieces to the first half of the mould before laying up the second half. The resulting thin slots form wedge insertion places to aid opening the mould after completing each fuselage moulding.

End of ramblings for a while until I have made the first fuselage from the mould.

[Brett82 494]

Im very interested in this thread as I am hoping to make some moulds of my own soon.  Thanks for doing such a detailed report.

I have viewed the video on easy composites a number of times and was also going to use that method but I will heed your warnings and make sure if I use that board it has plenty of support. One suggestion if I may, when you use the yellow putty for your alignment pieces, place them over the screw holes so you dont have to worry about trying to cover the countersunk heads with anything to make it smooth. 

I live in Northamptonshire, when I receive my plugs do you think I might be able to come down and see what you have done in person and get some tips before I start?  Im looking to do moulds for a full F3F glider. I should be getting some plugs between now and January but if that falls through then I was thinking of trying to design and 3d print something in segments, similar to what you have done. That's a much longer and bigger project though.

Brett

[CTR 81]

Hi Brett,

Sure, no problem getting together over this. I’m about 45 mins south of Corby.

Where are your plugs coming from? Is it your design or a combined effort?

Regards

Neil

[Brett82 494]

Hi Neil

That's awesome, thanks I really appreciate it.  At least then I wont be going in completely blind. 

The plugs (if I can get them) will be coming from Germany and are for the Caldera R. I have someone who has a set of moulds an is willing to make me a fibreglass "shell" and send it exactly as it comes out the moulds, so I can make my moulds using that. It will be cheap to make a fibreglass shell and should have a decent finish. The designer has offered the design for people to use on non commercial grounds but getting his original plugs is proving impossible.

 I've also bees speaking to others about another possible model but they are currently asking the manufacturer about it. Otherwise I might go your route and design something then 3d print it, I might need to ask a heck of a lot more questions from you and your mate if i do that though.

I dont wish to hijack the thread so I will pop you a PM later, maybe see if there is an opportunity to come see what you have done so far over the next few months. 

Brett

[CTR 81]

Stage 6 and 7/

Layered up the fist moulding yesterday. This is a tricky part because I wasn’t sure how long it would take to wet out all the layers, but the bladder in, thread up the half mould (with the bladder in), put the halves together, remove the threads, bolt the mould up tight and apply pressure.

All this time would determine the ‘pot life’ needed for the resin, I.e. the time the resin would still be useable.

I bought the resin along with slow and fast catalyst so I could mix the amounts to adjust this time in the range of 15 minutes to 24 hours! I was aiming for about 1 to 2 hours, but it depends on temperature too, so I couldn’t determine an exact timing.

A word about pressures: Previously (more than 20 years ago) when I did the prepreg HLG fuselage, I used vacuum bagging for the mould consolidation and setting of the resin. I did this by using a bladder as is used in the pressure system of these days but with the neck of the bladder open to atmosphere outside the vacuum bagging envelope so air was drawn in as the vacuum was applied to the outside of the bladder. This worked well with the prepreg despite having just under 1 atmosphere of pressure difference (I have a very good vacuum pump ).  Rationale led me to use the pressure system (inflating the bladder) this time as it saved all the vacuum bagging materials and equipment. Logic said that 15 to 20 psi would be adequate as that’s more than used successfully previously.

Seeing that others had success with 260 size modelling balloons and 6mm copper tube, this was going to be my starting point. However, about 1 minute after inflation the balloon failed. Don’t know why but thought it may be the circumference of the fuselage at the larges point being too much for the size of balloon. Fortunately I had the foresight to get the next size up when ordering the 260 size. I Quickly unbolted the mould, extracted the failed balloon, set up the larger size and reassembling the mould; applied pressure and it held. After 2 hours I heard the compressor start up! Obviously there was a leak. Yup, the balloon had failed. The dilemma now was not knowing if the slowed resin was set. This was unlikely because the small remainder in the mixing pot was still rubbery. Knowing that the thinner the layer of resin, the slower it would be to set. Anyway, I left it running for as long as possible with the compressor cycling every 3 minutes to refill the tank, but couldn’t leave that going overnight for the sake of the neighbours. As the pot resin was still soft I presumed the moulding would be a failure so I unbolted the mould to remove the mess! Surprisingly when half the mould was removed, most of the surface of the moulding was good but the failed bladder  resulted in some of the tighter edges having an excess of resin rather than the reinforcement layers being pushed into them. The resin pulled away in these areas. Somewhat disappointing but left the moulding in the other half of the mould until this morning.

Well, here we are, just taken the first fuselage front end moulding out of the mould. The school report would say “ Not a total success but shows promise.”

The closeup shows the failed corners; but encouraging to see the other side where the final finish certainly ‘shows promise ‘.

Things can be learnt from even this failed moulding. For example, torsional stiffness is really good as is longitudinal. The sides are a little too soft in front of the finger grip shaping, where the shape is reasonably flat. This will be remedied with modification to the next layup. 

I’ll let you know how it goes.

Not much advice being posted! All ideas welcome.

[Mikeb52 19]

Fair play to you sir! You are giving it a go, and I’m sure you will get there. I guess it’s trial and error that gets you there in the end. Watching this with great interest, although can’t offer any tips. 

[Brett82 494]

Hi Neil

You are doing a great job on this, probably why you are not getting many people offering advice. If that's the only blemish on your first pull then it definitely shows promise. 

Its a little worrying that the balloons keep popping. Have you thought about making a bladder out of plastic bin bags using a soldering iron? You mentioned you thought the failure could be at the wider part of the fuselage, a self made bladder from plastic bin bags can be shaped to fit exactly as required. Check the link below. 

 

Also, I know you watched videos from Easy Composites, are you aware they have a reduced price carbon fibre called "Black Stuff"  that they sell for £12 per meter or £100 for a 10 meter roll? It may be work considering while you are still in the uncertain stage, save a few bob until you are happy with a set process. Its standard 200g 2/2 twill weave but the weave is run a lot faster, so its cheap as they can produce more in a short space of time but there can be the odd blemish in the weave. Just a thought. 

https://www.easycomposites.co.uk/#!/fabric-and-reinforcement/carbon-fibre-reinforcement/woven-carbon-cloth-under-400gsm/black-stuff-22-twill-200gsm-3k-1m.html

Keep the info coming and it will be good to see how the wings are being done. 

Brett

[CTR 81]

Thanks for the encouragement guys.

Seen the black stuff Brett. Got an even better deal two weeks ago at HK. 1 sq M pieces of 200g carbon plain weave at under £3 per piece! Great for these trials. Twill drapes better but for the fuselage the plain weave cut at 30 deg. Is good enough.

I’ll give it another go maybe later today.  Will look into the bin bag route too. The 360 balloon should be big enough. The surprise for me was that it lasted 2 hours before failing. That’s the inexplicable bit. It’s worth a try using a balloon in the mould without the layup and see if that fails. If so, it is to do with the cavity shape/ size. If not, it may be to do with the semi square cross section along with sticky drag of the resin on the balloon. This could cause too much localised stretching on one side of the balloon.

Experiment, test, repeat ‘till we get there!

[Brett82 494]

20 minutes ago, CTR said:

HK. 1 sq M pieces of 200g carbon plain weave at under £3 per piece!

When I get to the point you are at I will look for that for the testing too. May even be an idea to back the fibreglass in the moulds themselves with a layer or two of carbon at that price!!!

Just regarding the balloons, I doubt it is the problem but what sort of temperature can they withstand. It might get hotter than you realise in the  mould and the balloons may not be able to handle as much heat as say some which you have previously used in the past. 

Brett