From my experience, there are a handful of pull system design steps. This post seeks to "simply" outline those steps and some of the math that should be considered.
However, don't let the brevity of this post mislead you. It isn't necessarily simple.
We will address more and more of the referenced math through future Lean Math posts.
- Understand and segment customer (internal or external) demand. The lean practitioner may find the following lean related math useful: average period demand, coefficient of variation, demand segmentation, ABC inventory analysis, and days inventory on hand.
- Understand supplier (internal or external) reliability, quality and availability. Relevant math includes: operation ratio, on-time delivery, scrap factor, etc.
- Select best pull system type – supermarket pull, sequential pull, or a hybrid. Not a lot of explicit math here, often this is driven by lean principles and value stream characteristics and value stream and organizational maturity.
- Understand supplier lead times. The devil is in the detail and the detail includes, but is not limited to: available time for changeovers per period, changeover opportunities per day, changeover distribution, and replenishment lead time.
- Understand pull system design constraints - working capital, floor and shelf space, shelf life, organizational discipline, etc. The following math may be meaningful: inventory carrying costs, inventory turns, square root law of inventory, and Pick's theorem (to measure square footage).
- Select best kanban type (if the decision is kanban). The basic options are production instruction (in-process kanban or batch kanban. If batch, triangle, pattern production, or lot making.) or withdrawal (interprocess or supplier kanban). Does your head hurt yet?
- Calculate and size the pull system. The typical kanban calculations are based upon average period demand, replenishment lead time, factor of safety (considering both buffer and safety stock requirements), and container capacity. And, there's math behind this math. For example, there are lead time nuances to consider based upon the type of kanban and there are things to think about when calculating trigger points and FIFO lane sizes.
- Simulate/test the system. PDCA often includes stock-out analysis math.