First F-8E Crusader prototype: the story of a long project.
This blog is about the journey that brought me to introduce the Crusader to the market.
It all started with the passion of one man: Woody Lee, COO at Ultimate Jets has always been a Crusader lover. He is one of the few owners of a still flying JD Enterprise Crusader. He built it about 10 years ago and still flies it on a regular basis. This is one of the very few F-8 I have ever seen flying successfully as a model.
A few years ago , as we talked about the Crusader over a drink, he told me that he knew a group of modelers from Texas who had made a 1/7 scale plug but never flew it. After discussing about this at length, we decided to get in touch with these people and see what was left of this project.
It appeared that the molds were on storage in an attic somewhere and that we could get them to our Houston facilities.
So we did. Unfortunately, the molds were quite thin and warped in all the directions. The wing mold was un-useable and the fuselage was bent. Anyway, the size of the plane was nice. Not too big, not too small. So I decided to salvage the fuselage and shipped it to my Dubai factory. After a couple of months, we got a thick fuselage plug out of the mold. The fuselage was bent, but I managed to straighten it decently by cutting it into 5” sections. At that point, It appeared that getting it back to good scale standards would be a lot of work. Also we just had received our first Kuka milling robot. So I decided to 3D scan that shape and convert it into CAD.
Here is a picture of the bent fuselage plug
That turned out quite well and I rapidly got a very decent first shot at the Crusader. However some shapes were not complying to the blueprints we had, so I spent quite a bit of time digitalizing the Crusader skeleton and fine tuning the scanned surfaces to match the documentation.
This CAD was then converted to Gcode and milled in two halves from a solid blocks of MDF with our robot.
The surface came out excellent and after some primig/ sanding, we had a very nice looking Crusader fuselage! However, this was not enough to me as the pictures and visits to real Crusader showed the beating of war operations of these embarked airframes. The skin of the plane was extensively dented and wrinkled. So we decided to include these features in our plug. The wavy pattern of every stringer was reproduced by hand in a subtle manner. This cannot be seen in straight light but is very evident in tangential light conditions. I believe this has never been done before on a model to this extend.
In parallel, I had started looking at what I could do for the wings, stabilizers and fin. The original plane had a relatively thin and pointy NACA airfoil, optimized for supersonic flight. I was concerned about using this at very low Reynolds numbers of 200,000 on a model. So I started working on different aero options in XFLR5. It was immediately evident that the original airfoil would be unsuitable for this size and wing loading and would result in a disaster. The simulation was showing very high stall speeds and wing tip stall tendency. I looked at my airfoil library and decided to go for an American Selig design, optimized for lift and low Reynolds numbers. However this airfoil was not optimal for slat operations and I really wanted to make the scale wing with low dropping slats that is so typical of the Crusader. So I worked on the reverse design module of XFLR5 and changed the camber of the upper front of the airfoil to make it more tolerant to high slat deflections at 15% of the cord.
Similarly, I was concerned about the possible elevator masking effect with flaps deflected and with the wing at 7 degrees up ( tail raising by 7 degrees ). Furthermore, XFLR5 was showing quite a bit of turbulent stream in this configuration around the stabilizer. This coupled to a stabilizer tip cord of 2” made me study that aspect of the plane aerodynamics as well. Sure enough, XFLR5 was showing early stabilizer tip stall and very low control efficiency at high angle of attack. So I decided to depart from the original stabilizer airfoil and go for a much thicker design, using airfoils optimized for super low Reynolds number of 50,000.
I chose a hand launch glider stabilizer airfoil for the tip and a modified NACA 0009 at the root. This stabilizer was designed to generate about 3 times more lift than the original one at moderate angles of deflection of 10 degrees. This in turn would allow me to use much smaller stab angles to achieve the same pitch authority and thus keep this flight control far from its critical stall deflection.
This would certainly get the plane to stay away from the dreaded stabilizer masking effect, but would require a precise and stiff stabilizer rotation system to give nice handling characteristic on the pitch axis around zero.
Further study of the fin area and short rear arm showed to possibility of fin flutter in high vortex energy environment. So I decided to go for a thicker airfoil that is more self damping and easier to stiffen. Also such a fin would give a better dampening on the yaw axis. Early Crusader planes without the ventral fins were known to feather-spin on takeoff in certain conditions due to yaw unstability.
The wings were CAD designed to include functional slats, aileron and inner flaps, just like the real plane. The wing of the Crusader is one of the important elements to get a scale look of the model. The other important aspect is the wing incidence system. This was also CAD and CFD modeled and extensively computer tested with finite analysis software. The slats and flaps had to incorporate an aerodynamic seal.
The sliding junction goes into the wing to masks the upper slot when deflected. This improves the wing aerodynamics tremendously. Live hinging was also chosen for the aileron and slats to ensure clean aerodynamics and maximum vertical hinge stiffness.
Another important aspect of the study was to make the plane easy to ship, transport and handle at the field. For this reason, the wing was split in 3 parts, joining just outboard of the inner small flaps.
Similarly, the fin was made removable as well as the front nose section. A large engine hatch was made just rear of the wing center section, big enough to pass the steel exhaust pipe in one piece. Although the center wing section also opens up a large access area, this one is deigned to stay on the plane in normal conditions. So the plane is easy to carry in a SUV with only the wing outer panels removed, making it a breeze to setup at the field. The large engine hatch opens up a big service area that allows to give access to all the required operations to refuel, switch on, fill air and start the plane from one single point.
Here is a picture of the fuselage fully open with the engine hatch and the wing removed. The main gear and wing hinge bulkheads can be seen there.
The front canopy was designed to be removable but normally not touched except to remove the batteries and charge them. The short nose makes it a breeze to access these batteries and make it very easy to charge them out of the plane for maximum safety. The full depth cockpit tub is inserted into the canopy frame and the whole unit is removed as a block. Enough room is available in front of the cockpit tub to fit a GPS unit and pitot sensor board or ASSI board.
I managed to get my hands on the drawings of the Martin Baker Mk5 ejection seat and that was the occasion to CAD design a super scale cockpit. Using the pictures of the seat, side panels, front instrument panel, I went full scale and included every screw, switch, dial and knob available.
These components are either 3D printed in-house on my SLA laser printers or available as a simpler kit of cast resin panels. The cockpit tub allows to place all these components as well as a full height ejection seat and allows for a very easy access to the nose section once the canopy/ tub assembly is removed.
All of this CFD study, iterations tests and troubleshooting took about 1 year before I would be confident enough to cut a wing and stabilizer plug. By that time, we had received our Kuka robot and these were precision CNC machined in the same MDF material as the fuselage.
Once I had a set of plugs finely sanded and primed, I started working on the paneling and rivetting. Upon careful study of the real plane pictures, it appeared that these Navy airframes had sustained some severe abuse during Vietnam war and most of them had marked wrinkles and wavy skin between the stringers. I thought that it would be really cool to reproduce these waves that I first saw on the very impressive Skygate Collection Hawk many years ago.
So I started tracing all the panel and stringer lines and sanded the plug along these in a wavy manner to create the skin ondulations. However I carefully made these so that they would only be noticeable in tangential light or when touching the surface of the fuselage. These came up extremely good, and after many hours of fine tuning and sanding, I started proceeding with the panelling and rivetting.
Panelling was made form the pictures of the real one and the Vought panelling manual, with a mix of different thickness adhesive aluminum like Flight Metal and primed surfaces. Rivets were punched into the panels / primer using a mix of tubes and rods of different diameters relevant to the airplane scale.
As explained earlier, the stabilizer design required a very precise driving mechanism. So I elected to make a very stiff setup. The pivot point was computed to be at 17% of stabilizer MAC for a good natural control stability, moderate oscillation damping and reasonable servo load. The pivot shaft itself is made from aero certified alloy 2044 from Alcoa. We purchase all our Alcoa aluminum directly at the Lafayette factory in the USA with certificates of origin. Al 2044 is specified by Alcoa as an aero alloy specifically resistant to shocks and fatigue creeping. Very useful on parts that are subject to transport shocks and vibration.
The 10 mm shaft is glued within the carbon fiber structure and double pinned between the dual carbon spars. This very important pin securing/ gluing is certified in our quality system ( Enata Intranet ) along with photographic proof.
I designed the shaft so that it would be as long as possible to butt against the exhaust pipe. The protruding segment is not that long, however, and I had to use a pair of super stiff 12 mm wide needle bearings encased in a Al 7075 solid block. The control arm is located between these two bearings for maximum stiffness and is pressure clamped as well as keyed on the shaft for maximum safety. The pushord is made of a pair of stainless steel 3 mm precision ball links and our aero certified 3 mm stainless threaded rod. The assembly is exceptionally stiff, precise and strong. This prove to be very welcome in this plane and gives ultimate precision and super high authority to the pitch axis.
As the plugs and molds were in the process of being finalized, I started working on the landing gear CAD. I started from the prototype landing gear that the Texan group had made. However I wanted my gear to be 100% scale. So I worked from the pictures Woody made for me and the hundreds of measurements he took from the real plane. All the tiny details of the struts, wheel rims, hubs and braces were reproduced on the CAD.
Some of the details were too thin and intricated to be CNC machined, especially on the nose gear. So I opted to work with the technique I had implemented on our SR-71: a hybrid build from 3D printed and CNC machined parts. The idea here was to use laser 3D prints where the details are too thin to be CNC machined and cut aluminum where strength matters. In practice, the sleek machined struts are covered with SLA 50 micron 3D printed resin covers. The covers contribute to reduce the gear weight and are interchangeable in case of damage. This gives very impressive lightweight and visually intricated results. The non 3D printed parts are in-house CNC cut on our Siemens machine centers from shock resistant Alcoa Al 2044 and Al 7075.
Some details like the main gear strut joiner ( or upright ) are quite intricated and require many machine hours to achieve the perfect shape with the proper angles.
The one part I was not sure about was how to actuate the main struts. The real plane is using a brace rear brace actuator that is hydraulically operated. This at 1.7 scale was way too thin to be efficient when powered by a pneumatic system. So different options were considered, like going hydraulic, or increasing substancially the size of this ram to make it functional. However no solution were satisfying. So I had once again to deprt from the original design and completely think “out of the box”.
I ended up creating a side retract box that was operating a trunion/ torque rod connected to the lower strut bracket. This retract would include a piston facing the rear, passing through the main gear bulkhead and locking flats at each ends of the torque trunion. This showed good promises on the cinematics workbench of the CAD software, so I 3D printed prototypes to test them in the plane.
The rear brace actuator did not end up becoming a dummy unit, however. I found out that it could be used as a rear stop to allow fine tuning of the wheel toe ( the toe changes as the gear moves forward and proper strut stability is essential to directional stability on this plane ).
The main wheels are 100% scale designed and feature a series of 12 custom manufactured M2 x 24 mm titanium socket head screws. They include fully sealed and dust proof ABEC 5 ball bearings and exactly scale rubber 3D printed graphene 3D structure tires as well as laser printed scale brake assemblies.
While working on the CAD of the Crusader gear, we had already started building the first Crusader prototype. Airframe number 1 was made from monolithic glass and carbon fiber with our super strong Finnish birch aircraft grade plywood. This ended up in a very heavy setup and as I was processing our weight and balance worksheet, I quickly realized that the plane would end up well over 20 kgs wet.
This was well over my design wing load of 350 grs/ dm2. So I decided to use this for static display only.
This is the plane that was on our stand at Top Gun 2017.
In the meanwhile, we tried some sandwich structure for the wings and fin, as well as stabilizer, and the weight started to drop, but not quite enough.
Prototype number 2 was scrapped as we ended up having problem with a test epoxy resin that was supposed to give great results.
The airframe number 3 was made with internals in Airex/ carbon fiber and ended at just 19 kgs wet, which was the max design wing load. So I decided to proceed with assembly of this airframe and started the ground phase testing with fixed wing incidence and fixed landing gear.
The airframe was extensively tested at idle and max thrust on the ground for optimization of the bypass system cooling and airflow. 8 temperatures sensors were installed in the plane, connected to our CAN-TEMP sensor board and ASSI module for telemetry. Also a FLIR camera was used to check out the temperature of different flows from the outside.
Initially, the plane had a nasty tendency of running hot and imploding the pipes. This was due to a combination of the long and narrow inlet duct, high thrust engine and short pipe. I went through 6 iterations of the pipe/ inlet duct shape/ bypass venturi taper before I could comfortably say that the internal air duct was properly working. This process was conducted over a course of 3 months, during which I gained a considerable experience on pipe design and computation, sheet metal spot welding and pipe material required. Once this process was completed, we went for the first taxi tests.
This airframe showed us during high speed taxi tests that the main gear and nose gear that gear support bulkheads had to be made from stronger material that airex/ carbon fiber and I reverted to plywood for the main gear bulkhead and 3D core material for the nose gear structure.
After all these researches and optimizations, I was finally ready for the maiden.
Unfortunately, I had an intermittent receiver failure for the maiden of this plane and one aileron as well as the rudder servos went to full stop a short while after takeoff. Anyway, I managed to land the plane in the grass with the engine shut down and minimal damage. This short flight showed that the plane was very tolerant to large flight control deflections at low speed and well mannered at high angle of attack, as I landed it at very low speed in the grass without any stall.
Because the repair would have increased the weight of this airframe and as it had already well served its purpose, I decided to retire #3 and proceed with the transplant of all the components to #4. This one had already been built and design benefited from the static and high speed taxi tests conducted on #3.
Prototype #4 was built with full carbon fiber/ airex internals for the wings, fin and stabilizers. The fuselage internals were thoroughly optimized with a mix of Finnish aircraft birch plywood for the main gear and wing bulkheads, 3D honeycomb shock resistant core/ carbon fiber nose gear bulkheads and airex/ carbon fiber everywhere else. This plane was built with a functional variable incidence system as well as the final version of the bypass and pipe system. It came out after 2 weeks of intense assembly at 18.5 kgs wet with all the systems implemented.
The second maiden was done in turbulent wind in the hot Dubai summer. Conditions were tough with an OAT of 43 centigrade and density altitude of 6000 ft. this flight was short as the plane proved to be a bit tail heavy and quite demanding to land. So I only did one circuit with it. Anyway, the landing was not bad at all and once again, airframe #4 showed some good manner at slow speed/ high AOA.
Subsequent flights with #4 were done at a COG moved further by almost 1” and with about 50 oz of lead in the nose. This plane was flown for about 10 sessions to fine tune the trims and throws and then sent to the paint booth to get the VF-111 “Sundowner” red scheme.
It was then flown according to my test flight program until development of full aerobatics capability and max design G load testing.
The test flight program took 3 months to open up the entire flight envelope of the plane and be ready to take the airframe to the market.
Crusader #4 is still flying on a regular basis and gives me a lot of information about the max load handling of the aircraft.
Crusader #4 has a takeoff weight of 18.6 kgs.
Typical production line airplanes come out at 13.6 kgs dry/ 16.6 kgs wet for the regular version with a B-140F engine and 12 kgs dry/ 14 kgs wet for the JWM version with 2 plasma bag tanks and 100N class engine.
A few more pictures of the flying prototype:
The wing bay with the carbon fiber molded actuator plate, designed to fit above the inlet duct, as well as the Jeti CB 400 distribution box.
The rear wing tilt hinges. The plain Finnish birch aircraft certified plywood bulkhead is a solid piece cut as a "O" shape around the inlet duct to ensure tilt stability of the wing. A large carbon fiber tape reinforces the skin at this area. The CNC cut hinges are designed to withstand 400 lbs of load each.
I spent a lot of time to carefully design the inlet duct. It is a 100% clean aero shape all the way to the carbon fiber bypass! This duct also follows the area rule with a continuous 4% increase. The inlet lip are sharp, like the real plane. As a result, the inlet "sounds" like the real one when the engine is running. One can see a condensation ring at full thrust on a wet day like the real one!
The nose strut after weathering look absolutely gorgeous. Its is scale to 1/10th mm. The tire is custom 3D printed from rubber and 100% scale as well. THis material has been tested in the hottest conditions and runs really well from -20c to +45c OAT!
The main gear, wheel, tire and brake assembly are also 100% scale!
The carbon fiber bypass and in-house manufactured stainless steel pipe work really well together. The pipe is made from 100% USA materials, including the highest grade stainless steel available on the market.
Here are some pictures of our thermal analysis. These were made with a FLIR camera. Full plane during engine run ( OAT showing at 41 c ):
The rear section skin temperature:
The hotter section gets to 65c. The stabilizer bearing area gets to 50 c on the skin and 55 c inside the fuselage during idle static conditions. We had 12 alumel-chromel thermo couples placed inside the fuselage to monitor the internal temperatures on telemetry via CArsten's CAN TEMP board.
The image above shows our 100% in-house made stainless steel pipe with the dual wall section behind of the bypass. The solid aluminum elevator bearing blocks are sandwiched between two carbon fiber/ airex bulkheads. These feature cooling holes that we can see in the background.
This validates the design of our carbon fiber full bypass system and our dual wall pipe. Sufficient cooling and efficiency is achieved with thrust figures of up to 160 N.
The Tg of the epoxy system that we are using is at 90c.