Holiday Agility In The Workshop - Part 2

Last time we started looking at my holiday workshop experience and seeing how it related to agility and infrastructure agility in particular. We looked at why the two are similar (long lead times for materials, limited rollback options for mistakes and so on). We then started to step through the process of building something out of timber and discovered a few useful rules for infrastructure agility along the way. We looked at the planning and material buying stages and discovered the first two rules for infrastructure agility -

• Have just enough information to get started. The detail will follow.

• When buying materials, give yourself a little slack. A little upfront cost leads to a lot of downstream flexibility.

Today we are going to cover some more of the process and see if we can discover some more useful rules.

The next part of the process is actually building the thing. Taking the raw timber sheets, assembling them into panels. Cutting to size. Cutting the joints. Gluing up and clamping. This is the real non-reversible part of the process. Once you cut a piece of timber you can't un-cut it. It can be somewhat nerve-wracking looking at a $200 sheet of timber and hacking it in half with a large drop saw. If that cut is in the wrong place, that's an expensive mistake. You don't get much leeway either. A mm too long or short and the finished piece won't go together without a bunch of crooked sides and visible gaps. I used to make a lot of those sort of mistakes. It's hard to be fractions of a mm perfect with large power tools.

This time though, I didn't make any cutting mistakes. Everything went together perfectly. I didn't do it by becoming more accurate at measuring or cutting. I did it by changing the way I built things. In the past, I would measure everything out on the timber sheets, fire up the drop saw and start cutting. I would cut all the pieces because setting up a machine is a pain and takes a while, so you want to maximise the number of operations with that machine before you need to change to another one. So I would do all the cutting to length at once. 

Now, I have the workshop set up better so that machines are already set up and accessible with minimal effort. I have become much more efficient at machine set up. So now it's no problem to take a piece of timber from the stack, go over to the drop saw and cut one piece. I don't need to cut everything at once. What this meant is that I could build individual key pieces first then use those to size the next pieces. So any mistakes making those first pieces, I could allow for when cutting the next pieces. I wasn't relying on every cut being perfect. That's a handy rule right there - "Automate setup tasks so they are easy".

I was building in a way that allowed me to compensate for previous errors (or design changes) as I went. I built in the order that gave me the most future flexibility. That's a good rule - "Build in the order that maximises future flexibility".

I built the legs first. I could change the length of the legs without impacting any other component. So I made one leg and checked its length inside, in the space the finished piece was going. I even built a mock up using come scraps and off-cuts from a past project to give a visual impression of how tall it was going to be. I checked that with my PO and agreed on the height. There is even an old woodworking term for a quick model knocked together to see how the finished piece will look. It's called a maquette. So that's a useful rule right there - "Use maquettes to visualise how the final build will perform in place".

Once I had one leg, I made the other three and I cut them exactly to the size of the first one. It didn't really matter how long that first leg was, I just made the other legs the same. I wasn't relying on an abstract plan, I was relying on the thing I had just built. Next I cut the mortises in the legs to fit the side panels. When doing mortise and tenon joinery, the tenons are easier to adjust than the mortises, so cut the mortises first then do all the adjustment on the tenons. That's another useful rule - "Do the parts that are hard to adjust first - do all the adjustment on the easy side".

Once the mortises were cut, I assembled the side panels slightly oversize then trimmed little by little until I had them exactly the right size. I measured and cut the tenons not off the plan but off each of the mortises I had cut in the legs. That way they fit perfectly. That's another useful rule - "Don't size subsequent pieces off the plan, size them from the performance of what is already built". Remember that leeway you gave yourself in materials? This is where that comes in really handy.

The sides were glued to the legs giving me two complete side pieces. Next were the shelves. Again, I used a maquette to see how the planned length would look in place. As it happens it would have been too long and impeded access to a chair, so my PO and I settled on a new length 200mm shorter than the original plan. Good thing I hadn't started cutting joints yet, wasn't it? That would have wasted a bunch of time. I then cut the pieces to the right length, made the joints (sized from the previous pieces rather than the plan) and it all fitted together perfectly. Then it was time for glue up and assembly.

Last of all, the final pieces - doors and other trim get sized and cut based off the finished frame. The whole thing gets assembled and final finishing (fine sanding and a couple of coats of oil in this case) takes place.

So there you go. A fully finished piece of fine furniture. And some (hopefully) useful insights into infrastructure agility. We came up with 7 basic rules -

• Have just enough information to get started. The detail will follow.

• When buying materials, give yourself a little slack. A little upfront cost leads to a lot of downstream flexibility.

• Automate setup tasks so they are easy.

• Build in the order that maximises future flexibility.

• Use maquettes to visualise how the final build will perform in place.

• Do the parts that are hard to adjust first - do all the adjustment on the easy side.

• Don't size subsequent pieces off the plan, size them from the performance of what is already built.

So the next time someone tells you that infrastructure can't be done "agile", what they really mean is that they can't work usefully in 2 week sprints. Which is probably true. But if they follow the 7 rules, they can certainly work in an agile way. The best thing is that building infrastructure in an agile way is easier, faster and a lot less stress than the traditional method. It also gives a much better end product.