That sinking feeling you get when you know something is probably wrong but don’t want to admit it yet hit me as I peered into the stern compartment at the autopilot shelf. We were at anchor in Friday Harbor, and I was doing a routine inspection of our new autopilot system now that we had put it through some rigorous testing.
Sometimes with boats there’s a temptation to stick your head in the sand and ignore possible problems, or not proactively inspect things because you’re afraid of finding a problem. Ignorance is bliss as they say. But this is something I always try to avoid. Ignoring a problem doesn’t make it go away, and with boats it usually will come back to bite you.
On this inspection of our fiberglassed plywood autopilot shelf I saw a small chip or crack in the top layer of ply, which hadn’t been there before. There was no indication the shelf had moved, but it certainly warranted further inspection. For a few hours I agonized over what it would mean if the shelf were failing. I had put a ton of work into researching and installing the autopilot system the best way I knew how, and yet somehow all that due diligence still wasn’t good enough. If this were a real issue it’d mean I’d failed at one of the most important parts of the project.
I knew if I had to rip the shelf out and rebuild it, it’d be undoing the work already done and doing double that to grind it out and build a new, stronger one. But it’s the right thing to do. The autopilot drive puts up to 650 pounds of thrust on this attachment point, and we want to make sure it’s something we can have complete confidence in.
As it turned out, the shelf was starting the buckle at the midpoint. This may have been due to not having backing plates or fender washers on the thrubolts, or due to having the wood grain aligned cross-wise to the direction of thrust. Furthermore I should’ve encased the whole shelf in fiberglass. I made a number of mistakes due to assumptions and lack of information online on how others have done this project.
What Went Wrong
My assumptions were based in the following line of thinking: 3/4” plywood must be really strong (it’s 3/4” thick!) and 5-6 layers of fiberglass cloth tabbing it to the hull is also quite strong. Figuring out exactly how strong it is though is very difficult – no one I found actually knew the breaking strength of the autopilot shelves they had built. They just tended to overbuild it (but everyone’s definition of this was different), and some people with more fiberglass experience had far better technique.
Through additional research and a few very helpful cruisers on CruisersForum I identified the issues the v1 shelf had:
- I hadn’t encased the top and bottom surfaces of the shelf in fiberglass. Plywood alone is strong in shear directions (like the chainplates that are sometimes sandwiched onto wood bulkheads in boats) but not in buckling directions. The autopilot base puts all sorts of twisting forces onto the shelf.
- The single “leg” of the shelf was aligned perpendicular to the direction of thrust, which isn’t as good as legs aligned in the same direction as thrust. This resulted in the 90 degree joint trying to wiggle and gradually work the fiberglass free at the hull tabbing.
- I used low quality Home Depot “outdoor rated” birch plywood. I didn’t realize there are wide variances in plywood quality, and Home Depot’s best plywood is often pretty junky. It had knots, voids, and only 5 layers of ply in the 3/4” thickness.
- I didn’t sand down to raw fiberglass on the hull enough. I did about 10 minutes of sanding with a Fein multitool, but should have done closer to an hour. It takes a long time to sand gelcoat off fiberglass.
- I should’ve laid a fillet of thickened epoxy at the joints before laying fiberglass. I filleted with GFlex, but a much larger fillet would’ve been better. The fillet increases the corner radius which strengthens the fiberglass tabbing.
- I should’ve glued the shelf legs to the top board instead of just screwing them together.
The V2 Shelf Design
Once we got to Campbell River we stayed three days (3x longer than our normal marina stay) to build the new shelf. Campbell River has excellent availability of hardware stores and tools.
I bought 5/8” marine plywood from Windsor Plywood in Campbell River. This plywood looked much higher quality than the Home Depot stuff. It had no knots, few voids, and 7 layers of ply instead of 5. I downsized slightly to 5/8” because I wanted to ensure I had enough room to build up more fiberglass on it – I’m working in a space with very tight clearances. And from what I now understand, the fiberglass gives the shelf most of its strength.
The new design has two legs, oriented in the direction of autopilot thrust. It took a full day to buy the plywood (walking a mile with a 4’x8’ sheet!) and cut it to fit the contours of the hull. I used simply a coping saw I had on hand. On the v1 shelf I used a Fein MultiTool, but that’s in our storage locker in Seattle now. And I found powertools are often a bit too blunt / crude for the tricky, contoured and beveled cuts I was doing. It required dozens of trim cuts and test fits.
To prep the hull I heavily sanded gelcoat off the fiberglass for upwards of an hour. I used a Dremel sanding drum and a drill-mount disc sander (which is cheap and doesn’t work that well, but I didn’t have our better sanding tools which are in storage). Sanding produced a ton of dust which I vacuumed and wiped up and then prepped the area with paint thinner.
The next day I screwed and glued the shelf together, using GFlex epoxy in the joints (West System actually recommends GFlex for joining wood that may expand or flex, so I think this is better than using a stiffer colloidal silica epoxy).
I also drilled the four holes for the autopilot mount now, because it’s much easier to do before the shelf is mounted in the hull. After fiberglassing over the holes later I would just redrill the holes – the holes in the plywood act as guide holes, and drilling the fiberglass isn’t very difficult.
I then filleted the bottom inner radiuses of the shelf with thickened epoxy (colloidal silica), wetted the whole shelf down with epoxy, and fiberglassed over the bottom using 3-5 layers of fiberglass cloth.
Next I worked on gluing the shelf onto the hull. I liberally applied GFlex to the bottom edges of the plywood – the parts that will rest against the hull. Again I think GFlex is a good choice here because of its slight flexibility. One CruisersForum member had recommended using a soft coring like balsa between the plywood and the hull, in case the hull flexes. That would be pretty tricky though and I decided GFlex is sufficient because our hull is quite stiff at the stern corner.
Next I filleted all the corner radiuses (shelf to hull joints) with thickened epoxy, and then began laying down fiberglass. I had dozens of fiberglass cloth sheets cut to various sizes, but I still had to do numerous rounds of recutting and mixing up additional batches of epoxy.
I worked in very small batches (~4oz of epoxy) because it was kicking fast (in about 15 minutes) due to the 80F sunny weather. I think I mixed about a dozen batches and laid in over 30 pieces of cloth (various sizes were overlapped, starting from smaller to larger). Overall the shelf top has at least seven layers.
Fiberglassing is a messy, sticky job! Due to cutting the cloth in the cockpit, we ended up having tiny fiberglass fibers stuck to the cockpit cushions and our clothing for weeks afterwards. These tiny fibers are itchy, and don’t vacuum up easily.
This was a really difficult project but I have much more confidence in the mount now. Our new autopilot continues to perform really well, and is one of the best upgrades we’ve made to the boat.