WIP and Pull system to increase Cycle time, Kanban System (PSK) - The Boat experiment

Last post 02:53 pm June 12, 2019
by Wilmark Johnatty
02:53 pm June 12, 2019

There is an experiment introduced by Klaus Leopold in his book Practical Kanban (Recommended reading for PSK Cert) in Chapter One where he folds paper boats to show the improvement in Cycle time and other improvements by limiting WIP. A video of the demonstration can be seen at the link at the bottom with reference points. He demonstrates building ships with an unlimited Push system and a Pull system with a WIP of 1 for each stage shows a much lower Lead time (he calls it lead) when he tabulates the results (see below). I dont think there is any difference in Lead Time between these two comparisons (The Unlimited Push vs the Limited Pull). Assuming that he is comparing apples to apples, and customer orders are arriving at the same rate in both situations, the work is still piling up in the Limited Pull system - just outside, and he basically admits this at 36:15. His Cycle time calculation comparisons (Red Ships) ignore the queue outside his measurement boundary that is building up faster in the Pull system, making the Total work in both scenarios basically the same, as the proverbial question for Cycle time: "When will it be done?", should include the FIFO queue before the first step, and should be basically the same in both systems. There is benefit with the Pull system from process examination/improvement, observing the efficiencies of each process with work limited, lower risk for less work started etc. By bypassing this FIFO stack of unstarted customer orders, he is in fact creating a CoS/priority situation for the Cycle Time measurement in the Pull system by bypassing these orders. After looking at this experiment and being encouraged that we can easily show improvement by the limited WIP system with simple examples (like a simple 3 step process of paying an invoice for example), it was disappointing to  realize this. However, in a system with more complex work like during a Scrum Sprint where that work isnt piling up at the front end, limiting WIP can result in increased Cycle time. Have any of you done the Boat experiment? What do you think?




6:50       Start Push System

8:50       Stop Building, Measure WIP

9:35       Introduces Red Ship

15:00    Red Ship finished

18:25    Start Pull System with WIP =1, Practice

20:00    Start Pull System

20:03    Stop and Measure WIP

22:38    Red Ship Enters Push

24:23    Red Ship Exits

25:40    Tabulate Results

29:00    Throughput is basically the same (earning the same rate)

29:45    Earning the same amount of money with less work in Pull

34:51    Lead Time Comparison