In this early drawing you can see a number of configurations we considered.

As you can see from the drawing, the earliest shell concept was symmetric top and bottom. Overall shell dimensions were set at 36 inches deep, 120 inches long and 20 inches wide. This original design was economical because you could build a mold for one half of the shell and then make two pieces. The two identical pieces could be modified as appropriate for the upper or lower section and joined together to make the complete shell.

The above drawing shows 2 possible fuselage designs and 5 different locations for the wing spar.

In the early designs, the wing spars are large diameter carbon tubes. The wing spars would slide into a still larger carbon receiver tube that was integrated into the fuselage structure. The small drawing below and to the left of the main drawing shows the supports necessary for the wing spar receiver in the earliest configurations.

This first configuration was discarded because I did not feel it could be built. The curved tubes could not be made as single pieces if we went with a carbon as our building material. We could make them of course, but we would never be able to get them off the molds without destroying the molds. Making the individual pieces from carbon and gluing them together was considered, but discarded as being too heavy. Making the parts out of aluminum was also considered and also discarded as being too heavy.

While our RAVEN design effort was in progress other HPA projects were experimenting with fully recumbent designs. Wayne Bliesner was incorporating a similar shape and structure on his latest Man-Eagle HPA. While in Germany, Per Frank ("Velair") was also experimenting with this fuselage style. We watched both efforts carefully. These two designs showed that fully recumbent designs could fly.

You can see the strong design influences of the Man-Eagle and the Velair HPAs on the RAVEN.

Both Bliesner and I were making a significant departure from what is now the conventional practice in HPA structures. Current "Best practice" is to construct a "bicycle" style Frame. Per Frank continued to use the more conventional frame style.

At this time the plug still did not have the wing stubs or the inlets (they weren't designed yet). Wing sizing was not complete and final wing location was not decided on. Where to put the wing is an entire design exercise in itself. The plug mold only followed the Crown, Keel & Mid lines. In this photo you can see our first departure from the Man-Eagle design. With the RAVEN, the crown line of the fuselage drops to the centerline of the fuselage instead of continuing horizontally out to the end of the shell. This was done to reduce fuselage wetted-area and to allow for including an efficient cooling system in the shells.

From this plug we made two fiberglass flashes, an upper and lower. A flash is a quick and dirty test part use to check dimensions and test fitting parts inside. In the above photo you can see the upper shell flash resting on the mold.

As the first iteration of the plug mold neared completion some potential structural designs were explored. Ground rules for the structure were:

It couldn't poke out through the skin
It needed to be able to support the secondary structure (the skins)
It needed to support a retractable landing gear <<
There needed to be room for cooling airflow.
The pilot needed to be able to get in and out of it. (doors have always been a problem on HPAs)
It needed to be sufficiently rigid to resist the pilot pedaling forces.
The pilot needed to be able to comfortably produce 0.25 HP for six hours.
There needed to be a force path from the landing gear to the wing spar.
The wing and tail need to be held in the proper relative positioning.

It was clear that if we were to fit an efficient structure inside the RAVEN shells, we would need to depart from conventional thinking.

We took our first serious design and constructed a mockup. The mockup was made from scraps lying around the basement workshop of the Museum of Flight. The mockup included the central support beam, the pedals, and the pilot seat. The wing size and location were still undecided.

Mockup of RAVEN structure in lower fuselage shell flash mold

This design had a problem. It would be difficult to include a retractable landing gear. You can also see that it wasn't going to be easy to attach a wing. The mockup would serve well to help us study pilot positioning.

It was at this time that reports were beginning to come back from the recumbent Man-Eagle taxi testing. The reports weren't good. In fact they were down right discouraging. So I went on out to the next test session to observe for myself.

Because the pedals and propeller system were cantilevered way out in front, the structure would swing back and forth with every pedal stroke. You would think that an HPA structure would swing up and down as the pilot pedals, but that's not the case. Without the ground contact on the wheels holding it straight, the fuselage structure wants to swing side to side.

As the pilot started to pedal, the nose of the airplane would swing side to side relative to the wing. Under high power settings the nose swing was greater then twenty degrees. It was obvious that the drag losses from the design were would be unacceptable. Binding in the drive chain would cause power losses. It was also clear that such wild swings would damage the shells of the RAVEN design. The problems were so severe that Bliesner abandoned the configuration and it never flew.

A 1/5 scale model of a possible structure modification.

With this model I was attempting to solve several problems. The nose swinging, the wing mounting, pilot seating, and the retractable landing gear. I replaced two of the tube sections in the design with composite panels. There was a panel that served to rigidly fixed the wing spar tube to the structure and served as a seat back. There was a lower panel that served as the seat bottom and as an anchor point for the pedal system. Along the underside of the two panels, I ran an open "C" channel. The "C" channel was added to increase the torsional rigidity of the panels, provided a mounting point for the tail boom, a mounting point for the landing gear, mounting for pedal system, and a post for the control column. (At this time the RAVEN design still used control cables)

We came very close to using this as our structural design. We decided not to use it for several reasons, the torsional rigidity still wasn't high enough and, it was heavier then we liked, and it required a forward swept wing. Knowing what we know now, I believe this design could work.

The next idea I had used a horizontal frame.

Another fuselage idea

This model changes from a single central tube structure. I decided to go back to the rectangular frame idea. But, instead of having the frame vertical, I placed it horizontal. By placing the frame horizontal, I was able to do several things.

1) I could make the frame narrower. A vertical frame needed to be at least 44 inches tall. By putting the frame horizontal, that spacing was reduced to 20 inches. I didn't need tubes any more, the side members could be constructed from composite panels. Just as important, the side panels could butt up against the wing root. The inertial mass of the wings would constrain the fuselage from swaying side to side.

2) The fuselage frame extending forward provided mounting points for the fuselage shells

3) The pedals presented a minor problem. It would require the crank arms mount to the fuselage on the outer frame. This would require four crank arms instead of the two found on a bicycle. Believe it or not, the drivetrain design we came up with is lighter and more efficient then anything else out there. A lot of bicycling purists will disagree with that, but I'm pretty proud of this design.

4) There was more flexibility in designing the pilot seat.

This model also shows that I was looking at getting away from using tubes for the wing spars. In this design, the wing spars slide into hollow boxes that were an integral part of the fuselage structure.

Wing bending forces were carried around the pilot using a ring structure. A ring structure was chosen as the baseline design because it eliminated the problem of the wing spar needing to go through the fuselage at the same location as the pilot's chest. With the RAVEN, we decided to go with a blended wing-body shape. (See the Where to put the wing link). Compressive loads carrying straight through and the tensile loads being carried around the lower ring segment. Because I was trying to use a cantilevered wing, I needed to go to a structurally more efficient rectangular spar.

The lower ring segment didn't just carry the wing spar torsional loads; it would need to provide a mounting point for the retractable landing gear. By this time, the first wing configuration had been designed and the airplane Center of Gravity (C.G.) had been calculated.

Further refining the structure

In this photo the above concept had been developed a little further using a foam and cardboard mockup. The dimensions and location of the pedal/crank system are well on their way to being defined. The original wing design (small area/chord) has been abandoned and a new wing configuration decided on. You can see an outline of the original design on the plug mold.

The outer dimensions of the fuselage frame are well defined by this point. In this design we were looking at having two wing spars, a change from the previous design which used a single carbon tube. The "sockets" for holding the wing spars have been replaced with I-beam stubs. Carry through of the wing bending forces from one wing to the other is still a major problem. This design explores the idea of using straight tubes to transfer the loads from one side of the fuselage to the other. The fuselage frame has been extended aft to support the detachable tail boom. The upper straight tube segments for wing compression load transfer were a bad idea and this configuration was never seriously considered.

Mockup with ring

In this picture I'm still trying to get the idea of a fuselage ring to work. The rear spar now goes through the frame and serves as a back support for the pilot. There is a cut out in the lower ring to accommodate a pivot for the landing gear. I've added the triangular section to the front of the frame to increase the frame rigidity. The frame now has two structural triangles, one in front and one behind the rear spar carry through supporting the tail boom. This arrangement produces a rigid frame. The ring though, would require much more structural weight then was allotted.

Photo of model

This is the final attempt at having the fuselage structure carry the wing loads. Wing compressive loads are carried through the straight upper ring segment and the tensile loads through the looped lower segment. The ring had extension stubs, which extended out and into the wings. Try as I might, I couldn't make it work efficiently. I calculated the weight of the ring assembly at about 10 pounds, much more then allowable. It was at this point that I gave up on putting a ring structure around the pilot. This concept of a ring structure should, and does work, but alas, not for an HPA. There have been several successful designs that split the wing spar and place the engine inside. The DC-10, the MD-11, and the SR-71 are three successful airplanes that immediately come to mind. The small size of the fuselage pod. This concept fails because the pod is much larger then the wing depth. The root airfoil is 9 inches thick and the fuselage pod is 36 inches thick. The 36 inch thickness of the pod represents the minimum height required to fit the pilot inside the fuselage. To efficiently blend the wing and fuselage together for this configuration would require that the wing root airfoil be scaled up to near this dimension. Since the DAE41 airfoil is a 13% thick, it would require a root chord of about 277 inches (23 ft or 7 m) for a smooth transition into the body. And….. If the wing root was 20 feet, the span would need to be about 200 feet to get a reasonable efficiency. That would give you a wing area of over 3,000 sq-ft. If you were able to build it super lightweight, say 200 lb. You'd need two power supplies. Anyway, the point here is that isn't the direction we were wanting to take with the RAVEN design. It's kind of an interesting idea though…. Be the first on your block with a 3/4 scale Human-Powered B2 bomber.

Reworked structural mockup

Here we have removed the ring structure from the mockup and have replaced it with a spar carry through. By this time several other configuration decisions had been made. The wing size and shape were known. We've constructed two cardboard wing stubs to help with determining the fuselage structural details and spar location. The problem of landing gear attachment was deferred.

There were two things we needed to decide about the wing spars. How many spars we were going to have, and how deep to make the spars. The aerodynamic balance point of the airfoils is at the 44% chord location. Our goal with the design was to have the entire airplane balance at the 44% location. By doing so, we would minimize the control loads on the horizontal tail. Even more importantly, we could suspend the pilot seat directly from the wing spar. That would let us significantly reduce the fuselage structural weight. At that location, the airfoil depth is 13%. Since we have a 60 inch root chord, the maximum allowable spar depth is 7.8 inches.

Alas though, we couldn't put the spar where we wanted. The pilot was in the way. The spar would need to fit into a space forward of the pilot's chest, to the rear of the pilot's leg range of motion, and above the pilot's hips.

This mockup is what we used to experiment with the spar carry through location. You can see the quick and dirty pedal mockup and our first seat.

Our studies showed that we would need to position the spar 3 inches forward of the optimal location. Not only that, but the spar could only be 6 inches deep and still allow for the pilot to fit into the space. Since we couldn't move the wing forward, this meant that we would need to go with two spars at the wing root. A main spar, 6 inches deep, 3 inches forward of the C.G. and a smaller spar running behind the pilot's back.

The main spar would carry 80% of the flight loads and the rear spar, the remaining 20%.

We next entered our cardboard phase. The critical dimensions of the fuselage structure were defined and we needed to fine-tune the design to minimize the weight.

In this time period, we constructed several cardboard mockups of the fuselage frame.

In this photo, we've created just about the absolute minimum sized structure possible. This design is pretty close to what we wanted. The two problems are that the landing gear support is too far forward and the drivetrain support structure is too flexible. But all in all, we were pleased with this shape.

About this time we had committed to creating a display for Seattle's Museum of Flight(MoF). An important part of that display would be a cutaway view of the fuselage. And in that cutaway view, museum visitors would be able to see the inside of the fuselage.

A couple of note worthy things here:

The nose structure extends well forward of the main structure (not so good)
There is a strong connection between the landing gear support and the spar carry through location (good)
The landing gear pivot design (not good)
The rear spar carry through & tail boom support (about the best we can do)
The design we produced for the MoF display seemed pretty close to what we needed, so we cleaned it up a little bit.

Final drawings before we started mold construction

It was from these drawings that we started the construction of the fuselage mold. You can see a couple of changes from the MoF display mockup. We didn't include the landing gear support. The aft end of the frame swept down and smoothly merged into the tail boom. The side of the fuselage frame followed the outline of the wing root airfoil. As the aft airfoil thickness reduced, the fuselage frame remained at 4 inches. The tail boom where it joined the fuselage frame was an ellipse (4 inches tall x 3 inches wide).

This drawing also shows the preliminary layout of the ventilation system, the two spar locations, and the windows.

Structural mold

Here the first structural mold (inverted) is complete and is being prepared for a lay-up. You can see that we left off the landing gear support. Notice that we have placed a fence 1 inch in from the side face of the mold, as well as down the centerline of the stub that will receive the tail boom. We will use the fences as guides for cutting and trimming the part when it is removed from the tool. We made a fiberglass flash off the mold and brought it down to our MR&D facility, Skywalker Industrial, at Boeing Field.

Fuselage flash displayed at a team meeting

The fiberglass flash is being supported some wooden dowels. The dowels are holding the side walls 20 inches apart and are positioned at the two spar carry through locations.

Fuselage & former program manager Heather Costantin

This is an interesting photo for a couple of reasons. It shows that once upon a time, we actually had an office. Scattered about the office were our initial at building airplane components. On the top shelf above the computer you can see our first set of tail skins. To the right of that, just above the clock, is one of our 1/5-scale structure models. That row of white notebooks was once all the program's documentation. Lying on top of the notebooks is our first batch of propeller skins.

On the table is the fuselage structure flash. In the front section of the flash is a folded piece of plastic representing the propeller shaft support. The black object on the tabletop is the propeller driveshaft.

The chairs are positioned to model the pilot seating arrangement. Using the chairs like this allowed us to works with a couple of different spar carry-through and landing gear support configurations. You can see that here we are experimenting with having the spar carry-through arcing upward a couple of inches. This would allow us to move the spar further aft an inch or two.