Improving my Violin Bow Rehairing with Lean Manufacturing Process Timings

As I walk up to a work station at my day job holding a clipboard and stopwatch, I notice the operators tense up. I can guess what's going through their mind: here comes another engineer to tell me that I'm not working fast enough, and he's going to time me to prove it. Time to put on a show and cover up all the issues until he goes away...

It couldn't be farther from the truth. The essence of lean manufacturing is to remove waste and burden from a worker. The philosophy holds that people are our most important resource (not the fancy new CNC machine), and that our continuous improvement initiatives are best served by a focus on the shop floor---where the value-added work that customers are willing to pay for truly happens. From my perspective as a manufacturing engineer, my job is to set people up for success and remove the burden from their tasks. Some of the best tools I have for that are an objective look at how long things take. I do that with a simple clipboard and stopwatch.

Violin bow rehair setup, with the fixture to hold the bow and the toolbox with all necessary tools

I was working on several violin bow rehairs (replacing the worn out horse hair in a bow) one weekend in my violin shop, and started thinking about what my customers are actually paying me to do. Am I delivering them good value and a fair price? How would I know? Even more than that, I am a deeply committed lifelong learner, and I wanted to know how I could be better at rehairing bows. I could certainly watch Youtube videos of other luthiers, attend a workshop, or scan online violinmaking forums, but in the end, those are either passive or expensive ways to learn. The best way I could improve right that moment would be to pull out my stopwatch and start timing how long it takes me to rehair a bow.

Determining the Process Steps

The first, and often hardest, step in taking process timings to is write down the procedure. So often, we think we know how we can improve without even knowing what we're doing in the first place, but the reality is that any improvements without a baseline are subjective at best, and might turn out to be detrimental. As Taichi Ohno, the founder of the Toyota Production System, famously said:

Without standards, there is no kaizen (improvement)

The simple act of writing down the steps of how I rehair a bow led to an immediate improvement. When I rehair a bow, there's a little piece of metal called the ferrule that has to slide over the hair after one end of the hair is attached but before the other end goes in. I'd often forget this little detail, and it would turn into an ordeal to undo one end of the hair and possibly even recut the plug holding the hair in. Needless to say, a simple operation that I found out only takes 17 seconds, if skipped, results in several minutes of rework! But once I wrote down my steps and started following them to get my timings, I never once ran into that issue.

The importance of writing down a procedure is simply to create a standard way of doing it the same every time. It's something to plant a flag in the ground, but the first way you write out your procedure isn't a hill to die on. In fact, it should change frequently as the process improves. For example, there are two schools of thought in the bowmaking and rehairing world: one says it's best to start by rehairing the tip, the other advocates for a frog-first approach. As with most things, there are good reasons for both. Pick a method, rationalize it sufficiently to yourself, and move on. If you are dedicated and take process timings of each method, and the time it takes to rehair is your most important consideration, you can find good proof of which way will be best. If there are other things more important in a bow rehair, such as avoiding crossed hairs or using the better part of the hair, then choose metrics for those and test out both ways to see which is better, always keeping in mind that if you change one thing, new problems are either created or uncovered.

Taking the Timings

I did this part on my own, simply hitting "lap" each time I moved to a different step. In most cases, though, you have someone to observe and take timings. There's a nuance here, that if you have someone taking timings of you, you want to establish the exact moment when each step starts. For example, the step to "Polish metal and lubricate" starts when I grab the tube of metal polish compound; "Tie frog end knot" starts the moment I grab the thread. You don't need to sweat the details, but it's important to be consistent. You'll find that it's far easier to time something like a touch or grab or taking a step, as opposed to trying to get the exact moment someone releases an object or stops walking.

You want to take as many timings as possible; statistics are usually better interpreted with more samples. For something like rehairing bows, I did 6. In a perfect world, I would have done all 6 the exact same way, but I already started making minor adjustments after the 1st bow as a few improvements became evident.

Below are two ways to look at the exact same data, the time it took to rehair 6 different bows. Both are interesting and informative, and it's simply important to point out that they emphasize different things. The first looks at the minimum time that it takes to do each step (the blue bar), the median time (green dot), and fluctuation (orange bar). The way I stacked up the bars tells me that if everything goes perfectly right, with no rework or anything, it takes me just over 30 minutes to rehair a bow.

The orange bar tells me a worst-case scenario for each step, but significantly it doesn't count that for the cumulative time to rehair. Those fluctuations are anomalies. If the green dot is pretty close to the blue bar, it means there was a one-off instance that a certain step took a lot longer---maybe something like a redo or a particularly tough ferrule to remove. A green dot near the top of the orange box means that most of the time I take closer to the max amount of time, and the shortest time was a one-off. For example, in the "Inspect, charge with rosin" step, I usually spend about the same amount of time, but on one instance I didn't do a very through charge with rosin. The red flag to me is when the orange box is fairly large and the green dot shows up near the middle. That means that there is a lot of inconsistency in how I perform a step. High, consistent fluctuation is a hallmark of burden, and a prime opportunity for improvement.

Another way to look at the timings is the cumulative time by the median. This is much more useful for me to determine how to set my price, since nothing ever goes exactly right. Sometimes I'll do a rehair faster and sometimes it will take longer, but the typical time it takes is currently about 50 minutes. I still show the fastest and slowest times on here with the error bars, and I can make the same conclusions as I do by looking at the fluctuation and if my median is closer to the minimum time, maximum time, or somewhere in the middle. In general, the first graph helps me quickly identify the areas for the most improvement, whereas this helps me to set my price and apportion a realistic amount of time in my day to rehairing bows.

Last note on these charts. Notice that I always used the median, not the average. Averages can be skewed high or low by one anomaly, and anomalies shouldn't dictate everything we do. The median is a much better representation of what "typical" looks like. It's important to note anomalies, and they may justify work to solve them, but more often than not, they're a distraction from the 99% of problems I face on a regular basis.

Lessons Learned

If you had asked me what I needed to work on to improve my bow rehairing before I took an objective look, I likely would have said something like combing the hair or cutting the wedges. While cutting the wedges certainly takes a lot of time and has lots of fluctuation, indicating it's a prime target for improving my bow rehairs, I would be chasing down a fruitless rabbit hole if I worried about shaving a few seconds off of a relatively consistent 30 seconds of combing hair out.

Below are a few of my ideas for tests to run and see if I can improve my bow rehairing:

• Cut the plug blanks down to a more useful size. The current blanks I use are large and take a lot of excessive cutting to get to the final approximate dimensions.

• Experiment with different types of wood for the plugs. Harder woods are harder to carve and less forgiving; I may be using too hard of a wood with low tolerances

• Polishing: is there a more aggressive method to polish the metal?

• Try out new tricks and methods to remove obstinate ferrules and slides

• Try wedging the bow hair earlier on, such as right after inserting the slide and ferrule

• Charging the bow with rosin using a powder-dispensing brush or a rosin cake with alcohol

There are a few nuances that I want to address. I can, of course, question whether or not I need some of the steps, or if they're in the right order. I could easily cut down on the time required if I didn't wet the hair; it would save me time in gathering supplies, cleanup, and completely eliminate the wetting hair step. However, if I do that, I have to be aware of how other steps might take longer as the hair is harder to manage, and if the quality turns out as well. Also, as I test out these new improvements, I don't need to worry about timing a whole new bow. I can simply time the individual steps that I'm working on. For example, if I'm trying cutting out new plugs, I can gather half a dozen plugs and a few different bows, and time myself cutting the plugs in the new way. Now that I've identified the major problem steps for me, I can narrow in and focus on them.

Conclusion

At no point in this did I ever feel the pressure to simply "work faster." That is never a sustainable solution, and almost always creates more issues with quality or rework. The focus of taking the timing was an objective self-reflection that helps me to see how I can better serve my customers and alleviate the burdensome parts of the task. The principle below is known as Pearson's law. I find it extremely profound.

When performance is measured, performance improves. When performance is measured and reported, the rate of improvement accelerates.

Most often, when we have these process timings in front of us, the first ideas are about how we can improve the value-added parts of the work---the parts of the work the customer is paying for. You can see this tendency in my list above. The customer is paying me to cut new wedges, and I'm looking for ways to do it faster. But is the customer paying me to gather all my supplies in step 1? It would make no difference to them if I had to walk around the whole shop gathering the disparate tools from hidden corners, or if I had all the tools for a rehair in their own little box that I can quickly grab (incidentally, this is exactly what I do). But the former method would take me several minutes and have high fluctuation if I couldn't remember where I put a specific knife or forgot to grab the string, whereas my current method takes me a minute to mainly go to the sink for some water, then quickly grab my rehair toolbox. I would guess that well over 20 of the 30 minutes in my best-case scenario time is doing tasks that are not value-added.

With all of this, my goal is to continuously improve my bow rehair techniques, and then pass on the savings to you, my customer. The next time you come in for a rehair, my process will likely be radically different, thanks to small, incremental improvements.