Resurrection

A few years have passed, and this blog kind of fell into disuse. No, I never did build the craft I was planning... It's still a collection of ideas floating around in my head. However, with my recent graduation from school, I a) have a bit more free time to burn, and b) have the prospect of living in a place with a garage, so the time has come to revisit the design of a second generation hovercraft. The basic idea is still a 12 to 15 foot-long craft, with a tapered bow that is squared off. However, the plans for the drivetrain, powerplant, hull construction, and superstructure have all changed.

The hull dimensions will be determined by the powerplant selection. Ideally I would like to find a lightweight 4-stroke engine in the 45 to 60 hp range, but this is in the awkward place between lawnmower-type engines and car engines. One possibility is a geo metro 3-cylinder engine.

Instead of having a separate lift engine I am planning to use a belt driven lift fan. Some type of automatic speed governor on lift engine speed would be ideal, but I haven't given this much thought.

After doing a bit of reading I decided to shift away from the extruded polystyrene insulation foam as a core material, and use a structural foam or other engineered core material. I'm still doing a bit of research to see what would best meet my needs, but so far structural foams are looking very promising.

Update

Let me try to bring this blog up to speed on how the project is progressing...

The short version is nothing much has happened, and is unlikely to happen in the near future: currently my energies are directed towards cycling and classwork, with little left over for designing and constructing a hovercraft.

Having said that, parts for the next craft keep showing up in my parent's garage!  So far I have most of the thrust drivetrain, and a 36 hp two stroke motor.  Still to come is a matched thrust prop for the motor, and a new lift fan since my current one is, to put it lightly, showing its age.  All of these parts are coming from my mentor for my high school project, off of a previous craft of his.  

HOVERCRAFT 1.0 END-OF-LIFE HULL DESTRUCTIVE ANALYSIS

8 AUGUST 2008

INTRODUCTION

In order to clear out space in the hovercraft shed, it was decided to dispose of the hull from the first hovercraft. Prior to disposal the hull was sectioned in multiple places to examine how the interior had held up after several years of use.

TOOLS USED

• Electric chainsaw
• Sawhorses
• Camera

PROCEDURE AND FINDINGS

The hull was laid on the sawhorses and the first cut marked on the hull, just aft of the seating ply. The chainsaw was used to cut the hull.

Massive Great Stuff expansion was found, similar to other areas of the hull examined previously while the craft was in service. No meaningful moisture was found.

Crack formation in the foam core was noted where the plow plane blocks lined up with the top panel edge. Not good.

It was also frighteningly easy to separate the Great Stuff bond. This proved to be a common theme throughout the demolition.

Equally frightening was how easily great swaths of fiberglass skin were pulled away from the foam core.

Original Glassing Procedure:

The core surface was lightly gone over with a rotary sander to take off the “skin” that forms on top of the panels. The fiberglass cloth was laid down and positioned on the core, and the epoxy was applied over the cloth and “squeegeed” down and evened out. The obvious error here is inadequate surface prep.

Places that had been repaired later adhered to the core much better.

This shows a section cut down and through the in-hull wood fittings that held the prop cage. You can also see how an entire piece of the top layer was easily peeled off. Also visible are screw holes protruding through to the lower layer of foam, which provided a water entry point deep into the core of the hull. Some slight moisture was noted in the Great Stuff layer.

Another example of a structurally failing Great Stuff bond. This was the story all across the hull.

Showing the aft prop cage attach strip. Unlike the forward strip, this one was showing some fairly advanced rot. Evidence of some slight water penetration is visible along the aft edge of the hull where the fiberglass has been removed. The skin that has delaminated here (from cutting with the chainsaw) is 3 layers of 10 oz. fiberglass; no difference was noted in how it could be separated from the hull.

These images show the mounting block for the anti-sway brace for the Thrust Engine Mount and its mounting plate. This was added at a later date, over Christmas 2006, when the craft was outfitted with the more powerful thrust engine. The backing plate was bonded with an epoxy slurry made with microballoons, and I was unable to separate it from the hull – the bond was extremely strong.

Upgrade Fabrication Method:

The original skin was sanded away with a belt sander, and the piece of plywood epoxied directly to the core. I do not remember what degree of surface prep was done, but I suspect the surface was roughed with the belt sander to some degree. Then 2 – 3 layers of 6 oz. fiberglass were used all around the edges of the plywood base plate. Again, this piece was firmly bonded to the hull.

This image shows a cross-section of the plywood block for the cleats. These were glued (epoxied) down with a piece of Kevlar felt in between to strengthen the bond, and that worked very well: this bond was still very strong.

This image shows the second large cross-section cut on the hull, across the seating ply. Extreme Great Stuff separation is readily apparent. The fiberglass under the seating ply was still firmly bonded to the hull and showed much less inclination to delaminate than the rest of the hull. If I was not already planning on it, this would convince me to skin the top of the hull of the next hovercraft with plywood.

These three images show how frighteningly easy it was to pull apart the entire section of hull! It took almost no force to simply peel my entire core apart. This is what really, truly convinced me never to use Great Stuff again: it has no chemical bond with the foam, it absorbs water, is weak in tension and shear, and compresses easily.

Great Stuff Surface Prep:

As you can see in the images, the surface is aggressively scuffed and sanded with very coarse sandpaper (60 grit range). No pinholes were punched in the foam, but in retrospect that would have been a good idea.

Another mistake made here was the sheer amount of Great Stuff that was used in the joint. Even with 4 large garbage cans full of water on the hull plus other random heavy items, it still expanded over an inch in some places.

This image shows water penetration under the port side legs that held the lift engine to the hull. The foam had pockmarks in it from water sitting there for long periods. The fiberglass was heavily delaminated around the lift engine mount legs.

The starboard side lift engine leg mounting area. Gasoline was spilled on this area on 4 July 2006, and ate into the foam underneath the mounting leg. It was repaired using foam chunks and an epoxy slurry. This repair held up reasonably well. Again water penetration is present.

Sectioning the lift duct to look for moisture or damage. Besides the shoddy construction, no glaring delamination, structural issues, or water was found (the piece of foam with water damage on it in the foreground is from the hull, not the duct).

CONCLUSION

This hull was in extremely hazardous condition. The core had mostly separated from itself, and the skin was, at best, tenuously bonded to the core. It is, frankly, a miracle it did not fall apart sooner with the larger thrust engine installed. Specific key structural points on the hull were very close to catastrophic failure. It is a good thing this craft was decommissioned when it was.

The Great Stuff failed from the beginning to form a chemical bond between the two layers of foam in the core, and the skin was never strongly or correctly bonded to the core. Yet despite these glaring shortcomings in construction methods, the hull managed to hold together for the better part of three years. This is a testament not the skill (or lack thereof) of the builder, but to the incredible strength and forgiving nature of composites.

Lessons Learned:

• Don’t use Great Stuff!
• Surface prep is KEY to getting a good bond between foam layers and between the skin and the core.
• Use more than one layer of fiberglass to seal the core – one layer has lots of holes, even with an epoxy topcoat.
• Skin the top of the craft with 1/8” plywood or doorskin.
• Epoxy slurries seem to bond MUCH better than straight epoxy.
• Be careful of water penetration points, such as screws.
  

Updates

As you may have inferred from the lack of updates, the project schedule has been pushed back just a bit ;-)

But don't fear, the build will still happen, but with the following changes:

• The thrust powerplant is now a 36 HP two-stroke motor from a hovercraft acquaintance of mine. He has also graciously agreed to provide essentially the entire thrust engine drivetrain, less the engine stand, AND a lift fan in good condition, for a reasonable price.
• The emphasis has shifted from a "quick-and-dirty" build to a quality craft that will last many years and be a fast, fun toy. To reflect this shift in purpose, the design stage is going to be extended quite a bit - the goal is to get the details so exact that I only have to order materials once (not sure this is anything near realistic, but I've got to have goals!)

Thrust Assembly Design

Below are some pictures of a more accurate and detailed thrust assembly on the hull model.

side view:




top view:




isometric view:




I decided to make the beam spanning the prop cage a 2 inch square box beam of aluminum, but I need to discuss its suitability with my solid mechanics professor to make sure it doesn't fail. In theory, all the thrust loads are transferred via the shaft to the forward bearing and then out to the diagonal braces, but a) I don't know if this would actually happen in real life, and b) there are still some very weird gyroscopic loads to contend with.

The other challenge was to devise a way to brace the forward inverted "U" without taking up a lot of cockpit space as my previous design did. On an outside suggestion, I tried bracing it to the top of the prop cage, but the angle was too steep. I decided to combine the lateral and axial braces into one, which sends them off at a strange angle, but gets them out of the way while still extending forward quite a bit. The modeled braces are a bit rough because of how I built them up, but in actuality they will be blended into the corners and the deck.

Except for the aluminum box beam, all of the thrust structure is made out of wood. The advantages of wood are that a) it's cheap, b) it's easy to shape, and c) the joints can be epoxied and fiberglassed, making them ridiculously strong. While aluminum is stronger per pound than wood, to fasten members together you either have to use bolts, or weld them. Both of these joining methods produce stress concentrations, which reduces your overall strength.

You may notice the prop cage is a lot bigger than the prop I have sketched in, and there's a good reason for this. Initially I will be running a prop I already have, which is pitched to be run directly on the engine with no reduction (36" dia, 18 degree pitch). At some point in the future when I have more time, I will fabricate a larger diameter prop with a higher pitch and adjust the pulley ratio to reduce the prop shaft rpm. The prop cage is sized to this future prop (48" dia).

You may also notice that there is no bottom on the prop cage: this is on purpose. The side legs will be glassed directly into the deck. Putting a bottom on the prop cage on the last craft caused debris to build up in the cage that would eventually get sucked through the prop, damaging it. This design allows a free drainage path to prevent this.

I am still working on how to mount the engine to the platform (3/4" plywood). The most elegant solution would be to drill and counterbore holes in the plywood, so that the bolt heads sat flush from the underside, and then epoxy them in place when the plywood was laid down on the hull. However, there are 2 main problems: 1) they have to be in EXACTLY the right spot the first time. From personal experience, I can tell you this is a lot harder than it sounds - it normally takes 3 iterations of drilling and "adjusting" the bolt holes in the mounting substrate before they are right. And 2) if the bolts separate from the epoxy later from vibrations or what-have-you, it will be very, very, very difficult to fix. I'll have to sleep on this problem some more.

That's all for now, folks.

Hull re-design

On the advice on some very experienced hovercraft designers, I have modified the hull design to maintain the full width for 7 feet of the length. This will increase payload capacity and increase the craft's lateral stability.

This does complicate the fabrication of the hull somewhat, as now the sides and their plow planes must be made in 2 cuts, and the walls must also be fabricated in 2 pieces. I am becoming concerned about the amount of 1" foam the design will need for the walls, bow panel, and top of the bow. If necessary, the top of the bow could be fabricated from 1/4" foam, since it is not structural.

I have also decided I'm going to need 2 more gallons of epoxy ($200), and most likely another roll of 10 oz. fiberglass ($7.xx / yd.). Many of the components could be fabricated with a lighter weight glass, (probably around 6Oz.), but purchasing a roll of 6 oz. in addition to a roll of 10 oz. would eat up a large portion of my already cramped budget. This weekend (easter weekend) I will visit home, so I will have the opportunity to ascertain exactly how much fiberglass and skirt material I have left over.

Anyway, renders of the re-design, courtesy of SolidWorks as usual (click for full resolution).

Side view:




Top view:




Underside view:




Isometric view:



Happily, modifying the hull allows me to take a much more conventional approach to constructing the lift duct. This will decrease construction time and angst quite a bit.

There is still a lot of design work to be done, even though it looks pretty finished. For instance, I'm going to need to install a tow hook in the bow in the event that the craft ever needs to be towed. This will require routing a hole in the forward section of the hull, epoxying in a large block of wood, and cutting another hole in the bow panel to allow the hook to pass through, then sealing the entire thing up. Little details like this are almost not worth the trouble to put on the plans, because all the fitting is done to other components, so any dimensions you give in SolidWorks are useless. The main thing is not forgetting to put the tow hook in.

As a second example, the back of the craft will not be left open: there will be small panels on either side of the prop cage that prevent water from entering the cockpit (for lack of a better word) when off cushion. In the bottom corners of those panels will need to be some drain holes that allow any water that enters the cockpit to drain out. After the panels are glassed in place, I will need to take a drill to them, and then seal the holes with epoxy.

As a third example, on my last craft I had issues with sand and other debris being carried by water into corners near the prop's plane of rotation. The debris would then get sucked into the prop, tearing up the tip tape. To prevent the same from happening on this craft, I'm going to make two low "dams" to direct any water and debris around the prop cage and engine mount area. But you won't find these on the renders, because they're a minor detail.

As a fourth example, (by now it should be obvious that I'm cataloging these here so I don't forget them), I have not designed the lift engine mounts that must be embedded in the lift duct. To be perfectly blunt, I have not given much thought to how these should be fabricated.

So as I said, a lot of little details remain to be planned out. Almost all of them can be dealt with "on the fly" as I construct the craft, but the challenge is to design to a high enough level of detail so that build time is minimized.

Pictures

Here are some renders from SolidWorks of my rough ideas. The model is low precision and only serves to get the basic ideas down and make sure it all fits together; it is not precise enough to actually start spec'ing out materials such as lumber.

Top view:




Side view:




Underside view:




Isometric view:




I need to do a post about the lift duct, which is probably the most complicated piece to fabric on this entire project.