One of kaizen's unofficial taglines is, "Just do it." And it makes sense. We try to spin the PDCA wheel as fast and as frequently as possible in order to experiment and quickly learn and make adjustments. But, sometimes we should just do it AFTER careful and extensive simulation. It seems wimpy, but it's about managing risk. Lean leaders should care about that.
So, when does it make sense to simulate an improvement? We actually do it all the time when we trystorm. Trystorming is a melding of brainstorming and simulation. It can be really simple stuff or it can be much more involved. People tend to be fairly OK with the simple stuff, but start getting weak in the knees when meaty simulation is required. They don't want to take too much time simulating. It can be slow and tedious.
Simple simulation. People can tolerate simple simulation like pantomiming the new standard work sequence with a draft standard work sheet and standard work combination sheet in hand before they try it out for the first time. Then they can make adjustments on the way. Hey, who wouldn't be OK with that level of effort and spontaneity?
More extensive. The more extensive simulations take time and require a certain rigor. Why do we need to endure this pain? Because the implementation of improved or brand new systems can cause big problems if we don't iron out some of the more substantial flaws. Often we don't know what we don't know. Here are two types of extensive simulations.
- Many people apply 3P (production preparation process) when developing substantially new or improved processes and/or products. As we all know, locking in a poorly designed product or process is a recipe for long-term pain and suffering. In brief, 3P is a team-based methodology in which the members down-select from multiple alternatives to seven different ways for a new improved process (or product), simulate the new process with crude, inexpensive, and quickly applied materials (PVC, cardboard, wood, duct tape, etc.), then whittle down the options to three best process designs (as measured against predetermined selection criteria), followed by more trystorming and then ultimate selection.
- Supermarket pull is a wonderful thing when properly applied, but you've got to get it right in order to ensure that the downstream customers are not starved and that there is no excess inventory. Pull system or kanban system simulations are extremely valuable. Using production kanban as an example, after taking a first cut at demand analysis, percent load analysis, determining what the kanban strategy will be (i.e., in process, batch - pattern, batch board, triangle), sizing the kanban, formulating the draft standard work (how/who/when regarding kanban posts, emergency kanbans, scheduling protocol, etc.), etc., we need to simulate the system using real historical demand data and some invented surprises. The simulation requires cards for all of the inventory, mock kanban posts, "scheduling," capacity analysis...the whole nine yards! It is critical to find out when and where the system breaks in a big way and then figure out what needs to be adjusted...before it goes live.
So, what are your experiences with either high intensity simulations or implementations where it would have been a good idea to simulate (or simulate better)?