A really concise and excellent article on how to apply takt time to a jobshop. Actually, jobshops are a more common phenomena in many businesses. The thinking can be applied to hospitals, retail and financial services, R&D or in the government and universities. Essentially, the trick is to find patterns of work in a sea of chaos. The closest hint you will find in mainstream lean literature is the concept of pitch. This is what the article is really describing, the application of pitch to different value stream products: easy to handle, harder to handle, toughest to handle. The article calls finding this pitch the common denominator. We can find common denominators in many things: container sizes, product types, cycle times, ease of use, etc. The trick is in quantifying it in terms of pitch.
I've worked with teams to apply this concept in two different ways: one was with the building of customer configured control panels. All the panels were "different." But by finding the common assembly patterns, we were able to create a flow line with the right amount of workstations and feeder lines for the manufacture of customized control cabinets on a four hour takt time.
Another application was in injection molding. The average plant takt time was about 6/1000th of a second, spread across forty machines that can run many different products at any given time. Kinda makes it tough to use one takt time, doesn't it? Most people would say lean and JIT doesn't apply. But the group found the common denominator: how often does a person have to pack product, i.e., take product away just-in-time for the machine to be able to eject more product without disrupting the cycle? This is something the people quantified and level loaded.
Bottom line: when takt time is too variable or too small to balance against - you need to find the pitch or as this article calls it - the common denominator. This is the first step in creating level flow.