I remember very vividly the first time I worked on a manufacturing line that was considered "lean". I was a quality engineer, but my mentor told me that to become an operations leader, I needed experience supervising a "lean line".
When the second shift supervisor was on vacation, I stayed and supervised his shift, just for the experience. My first day was an absolute failure. I had forgotten to make sure that there was enough material and we ended up missing takt time. In fact, by the time my shift was through, we were about 4 hours behind in production. Oops. Fortunately, by the end of the week, I figured out my rhythm.
That experience resonated with me in a couple of ways. The first is that the concept of Takt time does not leave room for error. Think about it - Takt is customer demand divided by the available production time. The result is a single number. For manufacturing facilities with standard products and little variation, this is wonderful. However, for manufacturing job shops and high variation production, more mistakes happen. How do you measure improvement to hitting takt without defining the variation?
The second thing that resonated with me was the use of a physical Gemba board, where magnets represented the parts and notes were added in dry erase marker. How can you identify systemic issues and trending if the data is constantly erased? How can we predict the production output for the rest of the week when mistakes happen and the data is no longer available?
Looking back, I think digitally capturing the data on the Gemba board would have been a good investment. Well, maybe not then since technology solutions were expensive; however, these days you can find low cost options, or create your own solution. By capturing the data, the variation to the takt time can be defined. Additionally, analytics can extract insights from that data to remove waste.
When parts are manufactured behind takt time, you miss customer deliveries. When parts are made ahead of takt time, you end up with over production. Over production increases inventory and ties up cash. Analytics can help improve operations. One analysis that is useful is to compare the production runs when takt time was hit vs. those production runs that are missed. Is there a pattern in the shift, the day of the week, or an employee?
With enough data, a predictive model can be created to determine if the production demand could be met or not. If you know early in the week that production will be missed, actions can be taken to offset. One of my mentees created an analytic that looked at the historical production data and forecasted the output for the week, based on how much of the standard work was completed. The analytic went on to look at the employee training records and recommend a person to add to the line, based on who was trained. Pretty slick!