[Drrrrring Drrrrring]
<Bob> Hi Lesley! How are you today?
<Leslie> Hi Bob. Really good. I have just got back from a well earned holiday so I am feeling refreshed and re-energised.
<Bob> That is good to hear. It has been a bit stormy here over the past few weeks. Apparently lots of hot air hitting cold reality and forming a fog of disillusionment and storms of protest.
<Leslie> Is that a metaphor?
<Bob> Yes! A good one do you think? And it leads us into our topic for this week. Perfect storms.
<Leslie> I am looking forward to it. Can you be a bit more specific?
<Bob> Sure. Remember the ISP exercise where I asked you to build a ‘chaos generator’?
<Leslie> I sure do. That was an eye-opener! I had no idea how easy it is to create chaotic performance in a system – just by making the Flaw of Averages error and adding a pinch of variation. Booom!
<Bob> Good. We are going to use that model to demonstrate another facet of system design. How to steer out of chaos.
<Leslie> OK – what do I need to do.
<Bob> Start up that model and set the cycle time to 10 minutes with a sigma of 1.5 minutes.
<Leslie> OK.
<Bob> Now set the demand interval to 10 minutes and the sigma of that to 2.0 minutes.
<Leslie> OK. That is what I had before.
<Bob> Set the lead time upper specification limit to 30 minutes. Run that 12 times and record the failure rate.
<Leslie> OK. That gives a chaotic picture! All over the place.
<Bob> OK now change just the average of the demand interval. Start with a value of 8 minutes, run 12 times, and then increase to 8.5 minutes and repeat that up to 12 minutes.
<Leslie> OK. That will repeat the run for 10 minutes. Is that OK.
<Bob> Yes.
<Leslie> OK … it will take me a few minutes to run all these. Do you want to get a cup of tea while I do that?
<Bob> Good idea.
[5 minutes later]
<Leslie> OK I have done all that – 108 data points. Do I plot that as a run chart?
<Bob> You could. I suggest plotting as a scattergram.
<Leslie> With the average demand interval on the X axis and the Failure % on the Y axis?
<Bob> Yes. Exactly so. And just the dots, no lines.
<Leslie> OK. Wow! That is amazing! Now I see why you get so worked up about the Flaw of Averages!
<Bob> What you are looking at is called a performance curve. Notice how steep and fuzzy it is. That is called a chaotic transition. The perfect storm. And when fall into the Flaw of Averages trap we design our systems to be smack in the middle of it.
<Leslie> Yes I see what you are getting at. And that implies that to calm the chaos we do not need very much resilient flow capacity … and we could probably release that just from a few minor design tweaks.
<Bob> Yup.
<Leslie> That is so cool. I cannot wait to share this with the team. Thanks again Bob.