The spinner controls how much work you do. The average
amount is 4, click the spinner to add or subtract a random
number from that average. The Variability dropdown lets
you choose between ±1, ±2, and ±3.
Every time you click the spinner, work happens at the current
workstation, and you move on
to the next workstation.
It's best to start out manually clicking the spinner until
you understand what's happening and what the numbers mean.
Once that gets tiresome,
click the "Go" button to automate the
process and work successively at each workstation
for 10 complete cycles. Reset sets everything back to zero.
The five workstations each contribute some amount
of work to the final product.
They represent activities that rely on a
previous activity having been completed.
In a factory: these are construction stages.
In a phase-gated process, they're
SDLC stages (design, dev, test,
deployment, etc.).
In Scrum, they're 5 back-to-back sprints, etc.
The spinner controls how much work you do (the number of
coins that flow through the station). The tin can represents inventory—incomplete work.
Workstations depend on each other, so
you can only do as much work as is in your inventory.
If you spin 6, but your inventory is 3, you
can only do 3 units of work.
The plot keeps track of how far ahead or behind you are from the
average. It's a cumulative measure.
With no variability, work marches through the
system at the average rate, and we'll plot points
on the dashed line. You add points based on actual work done.
If, in this cycle, you do two units of work less than average,
the dot will be two points down from the previous dot. If you
do more, it goes up.
The biggest misconception that people have about the way systems
like this work is that they think that averages will work for them:
If you're down a bit this time, you'll be up a bit in the future
and work will average out. Unfortunately, when you have dependent
operations, it doesn't work that way. Put another way, if you
estimate by summing the average time required to do some set of
dependent tasks, you will always be late, usually by a
significant margin. The higher the variability,
the later you'll be. Note how the first workstation, which
has no dependencies, performs much better than the others.