These felloes helped salvage half a dozen taiko drums

Ten nice looking barrels for making drums. At Oakland Taiko we had 10 new barrels donated to make mid-size taiko drums ('chu-daiko'). That was summer 2023. The barrels were 20-gallon, made in Fresno, and cost $500 each, new. That' $5000 donation was a really big deal at the time. The group of us preparing to make drums collectively thought that these small barrels, 40% the capacity of common wine barrels in California, would be just about the right height and head diameter to make serviceable chu-daiko.

In contrast, common used wine barrels are taller and fatter.

Common wine barrels are about 35" tall, containing 59 gallons 

Before discussing the challenges we faced with the small, 20-gallon barrels, let's look at the shape of barrel staves, particularly the ends, where there is a groove called the 'croze'.

Oops, not quite the right height. Barrel staves have a deep groove on the inside near each end, called a croze. The croze receives the barrel head and is a deep groove that leaves only about 1/2 the thickness of the wood. A group of us concluded that using the complete height of the 20-gallon barrel stave, including the ends beyond the croze, would result in a structurally inferior drum. As a result, we cut down the height of four barrels at the croze, losing about 4 inches of height and resulting in four stubby drum shells. 

Sketch of how we cut off the stave ends at the croze (line A) and added reinforcement (E)

Compare the stained shell (with the cowhide on it), made from one of these 20-gallon barrels, to the shell to its right. The stained one is about 4 inches short.

We headed two of those 20-gallon drums with skin and put them into service. This led to unintended consequences such as needing taller stands, and those two drums being of limited use. The remaining two stubby shells have been languishing, not yet headed, in storage. Worse yet, 6 brand new 20-gallon barrels were languishing totally untouched in storage. That's $3000 of unused donation, was there anything to be done?

Short drum from 20-gal barrel on right

Isn't Reinforcement Learning a fancy AI thing? Necessity is the mother of invention. There had been some discussion of spending funds on buying new drums, and that's what prompted me to think about getting those 6 20-gallon drums into service. I hated to see the $3000 of donation going to waste. Sure enough, I came up with a solution for keeping the full height of those barrels. The idea was, make a really, really good reinforcement that could allow the full height to be used, despite that croze groove. It would have to be stronger and more rigid than the other three reinforcement methods I'd tried so far. My idea? Put the head back in to fill the croze, and then add a bent hoop of oak on top of the head. Finally, cut away the center of the head, leaving just a ring of reinforcement.

Some context about taiko drum reinforcement might be useful. In the US, and in many places (Africa and Asia too), many drums are made using stave barrel technology. Here in California it's easy enough to take apart a used wine barrel, then glue it together without leaving the metal hoops on, to get a nice drum shape. Wine barrels are made from oak that is almost 1" thick, which is strong enough to build a drum, but there is a problem over time, as chips of wood can fall off the inner part of the rim. According to taiko lore, this is due to the fact that taiko players often hit the rim with the drumsticks ('bachi') to make a sharp rapping sound ('ka!'). Eventually, after enough 'ka'-s, the wood may crack and flake off on the inside. This leads to undesirable rattling sounds on the bottom of the drum and an inferior sound on the playing head itself. Since a taiko head can last 10 years or more without needing to be replaced, it's worth our while to prevent rim chipping. In order to stave off this chipping and splitting, many taiko makers in the US add some sort of reinforcement ring on the inside of the rim at each end of the drum. This reinforcement is designed to prevent chips from falling away, but it's not necessarily strong enough to prevent cracks from forming in the stave itself. 

In the course of the last year I learned and tried 3 methods of reinforcement, and observed another two methods. Each had its pros and cons. There was a lot of learning about reinforcement going on, Reinforcement Learning, you could say. Eventually I got to thinking, would any of those 5 reinforcement methods be a solution to the 20 gallon drums and their problematic croze weakness?

It would take a while to describe the 5 reinforcement methods, so I'll just summarize and say they were either relatively thin, or made of many pieces. And since the problem we are trying to solve here is a pre-existing weakness in the stave, thin or many-pieced additions don't seem strong.

First reinforcement method I tried: plywood disc.

The disc is then ground down to taper away from the drum head.

Second method I tried: kerf-cut plywood and fiberglass.

Third method I tried: re-use the stave ends and glue them with epoxy.

Method #1 (plywood) might appear strong enough for reinforcing the croze, but the reason I rejected it is that the staves are much thinner above the head, where the chime tapers to the top. When we used the plywood disc method, we actually cut a rabbet into the stave. Not only was that a lot of work, there's just not enough material to allow it in the chime area.

Rabbet for plywood reinforcement method, this is a barrel that we cut off at the croze

Reinventing the wheel.  None of the five reinforcement methods struck me as strong enough to prevent a stave from caving in right at the weak spot, the croze. However, working with barrels has the effect of heightening one's awareness regarding bent wood. After all, staves are bent using heat (open flame) and some surface steam (water splashed on the hot wood while being exposed to the flame). So it occurred to me that one might steam-bend oak into the right sized hoop and insert it as reinforcement. This pretty much amounts to making the circular parts of a wagon or carriage wheel, and instead of adding spokes and a hub, simply gluing it to the inside surface of the barrel. Part of this idea was to glue the barrel head back in place, thus filling the croze itself with perfectly fitting oak, and then laying the new oak hoop onto the head. Together, a multi-layered extremely rigid reinforcement structure of the best possible material (oak, like the staves). I reasoned, if the hoop is almost as thick as the stave, then it will make up for the weakness introduced by the croze. 

Serendipitously, while planning this project, I visited Death Valley National Park and saw quite a few wagon wheels.

Hoops vs felloes. You can easily take a thin narrow strip of wood and loop it into a hoop (a full circle). You overlap the ends and somehow bind them together. Voila, a flimsy hoop. It's not really circular, more of a spiral, since the ends overlap. You can make it stronger by using slightly thicker material and maybe layering a few of these together. But the overlaps and layering are a lot of work and not as strong as one thick piece. I learned from Professor YouTube (one of the primary advisors in the School of 21st Century Life) that above a thickness of 1/4", hoop-making results in not very circular hoops with nasty ends that don't meet. Professor YouTube introduced me to Engel's Coach Shop, a family-owned business in Montana that publishes videos about restoring wagon wheels, wagons and more, often by steam bending massive pieces of air-dried white oak. You can really learn how to do this work from Engel's - and I did. The strong wooden rims of wagon wheels are called felloes.

The only thing is, Engel's uses a power winch, a massive steel belt, and air-dried lumber. None of these were readily available to me, so I needed to learn enough to make adjustments in my method. I decided to learn the why's and wherefore's of certain aspects of wheelrighting. 

Nobody expects a chariot to be part of a story about taiko, yet here we are. About 5000 years ago, spoked wheels were invented and were fairly quickly adopted to make chariot wheels - machines of war. 

Chariots were a big deal, so sometimes they would be carried

And who can forget that Tutankhamun's tomb contained a chariot?

The key thing is that even in ancient times, spoked wheels were made of bent wood. This design preference is because having the wood grain follow the wheel curve is stronger than piecing together multiple pieces of wood. and for millenia, two half-circle pieces, called felloes, are assembled into the full circle of a wheel. There's a lot of engineering detail here, but the summary is: you can make half circles easily and assemble two of them with two joints. If you try to make a full circle and only one joint, it becomes much harder to do successfully. If two pieces of bent wood are strong enough to carry a warrior with armor at high speed in war, then I guess a similar design will be adequate to reinforce our 20-gallon drums with their croze "weak links".

So, we went ahead and glued together the 6 20-gallon barrels, gluing the heads in place as well. The heads were glued with paste epoxy. We had a crew of over a dozen helpers to get this gluing done.

I bought 1" x 4" white oak boards (white oak is better for steam bending). Selecting straight-grained pieces is essential. Any knots or uneven grain will result in a crack or angle instead of a curved bend. Rule of thumb is that the grain must not "run out" within 12", meaning, it must not cross from one side of the board to the other in less than a foot. I decided to work with pieces 12 inches longer than the length I actually needed for the felloe. As you'll see, this extra length is needed for leverage in the bending process (and a bit longer would have made the work even easier). Then I used a table saw to make the profiles I needed. On the left, to pieces that will form the felloes. On the right, two pieces that will fit into the gap between the felloes and the chime. 

6 barrels = 12 ends
@ 2 felloes per end, that's 24 felloes and 24 wedge pieces

Many steam benders soak their wood for a few days prior to bending. This is intended to help when you don't have air-dried lumber. (Kiln-dried lumber is harder to bend because the lignin has already been heated in the kiln.) I made a frame from scrap wood and lined it with scrap plastic.

Wood is a complex structure. In this case, I wanted water to be absorbed from the sides of the board, not up through the vessels from the end grain. So, I applied a sealant to the ends before soaking. This is a tip I read in an online woodworking forum, so I can't vouch that the information holds water.

Next, I needed a steam box. I accidentally made mine twice as long as I needed. It's just a 5" wide, 5" tall box. The edges are rabbeted and sealed with caulk. It slopes up, away from the door. The door is also rabbeted and held shut with a rubber band. This is a common design I saw on YouTube. The only thing I bought for the steaming was a "steam generator" found on Amazon (about $70).

 

7' long, could have been 4' long for my needs

 

The box sometimes reached 220F

Usually I steamed one fellow at at time, but the wedges could be done in a batch. Note the plastic tray, for liquid fabric softener. Fabric softener helps the steam penetrate the wood.

Steaming is more efficient if you do many pieces in a sequence, because once you get the box hot, it's best to keep it that way. Tending the steam generator (filling with water), checking the temperature, timing the actual effective steaming duration (30 minutes in my case), and then bending and transferring to the drying rack, all this keeps you busy. I got most of the bends done in several 12 hour days (48 pieces total).

After the wood comes out of the steamer it is hot. You use gloves to reach in and remove the wood, but the gloves can come off within seconds. I was worried about how quickly I needed to work but in practice, I found that the wood remains pliable for several minutes at least. 

The bending form consists of several parts: (a) a circle of plywood, (b) a strap of steel, (c) end blocks attached firmly to the steel strap, with just enough space between them for your workpiece, (d) a clamp to hold the middle of the bend firmly against the form, and (e) a way to apply pressure to the workpiece ends (actually do the bend) while holding the form in place. There is also (f) a secondary block on the outside of each end block (and screwed to the end block), which helped keep the workpiece and strap and end blocks aligned during the process.

I tried simply bending the felloe by hand, but it was quite hard to do that and get a clamp on it. I did one that way and then set about building a proper bending setup.

Bending by hand (no winch)

I used cheap pulleys and an inexpensive winch. Also, simple two half hitch knots and a cheap rope.

A hand winch is connected by rope and pulleys to the ends of the bending strap

With this setup, bending 3/4" oak was a snap. Video.

One more detail: "springback" refers to the amount that the wood flexes back toward its original shape after bending. If you let it dry in the form for 12 hours, this minimizes springback. After trial and error, I found that a bending form 1" diameter less than the desired circle size worked well. I could transfer the bent wood from the form 30 minutes after bending and put it directly into the barrel, where it would dry into the exact shape desired.

I also built this fancy drying rack but later found it was easier to simply transfer the felloes directly into the barrels for drying.

Drying rack was more trouble than help

Letting the pieces dry inside the barrel end

Once the pieces have dried in shape, it's time to glue them in. The 3/4" pieces did not spring back much (maybe 1" at the most). The wedge pieces are more springy. 

Smear glue onto the chiv and chime and a little on the head

Glue on the reinforcement piece

Screw the reinforcement pieces in to make a tight fit, then clamp in the wedge pieces to fill the gap between chime and reinforcement piece.

I then cut out the center part of the barrel heads using a jigsaw and cleaned up the barrel rims as usual on the taiko-lathe-mill. After sanding we have a very sturdy, proper height chu-daiko shell. This is by far the strongest reinforcement method of all we've tried.