Hey guys, I wanted to share with you my latest post on http://www.theleanedge.org. w/ Michael Balle'. The question is below from Andrew Turner, please see below. You can go to the website to see other Lean author responses or see mine below the question:
Andrew Turner: Where do we start in a Press shop?
“Our company is split in 2 sections, the one a JIT assembly plant, the other a mass production Press Shop. Implementation of Lean in the JIT plant has been relatively simple (not that Lean is ever really simple), however, we are struggling with the implementation in our Press Shop. I know the importance of items like SMED and Heijunka in driving this journey, yet we are battling to get the ball rolling forward. Where do you think we should start the process in the Press Shop?”
Hi Andrew, I will answer to my personal experience in regard to this question. I think its a good one, it can bring out many dynamics that fall under that umbrella of thinking “flow vs batch” so I will try to cover several of them within my answer. When I was first exposed to the Toyota Production System (TPS) “thinking” in 1988 at Toyota Motor Manuf. KY (TMMK) I made an assumption that if you weren’t practicing one piece flow then you weren’t effectively practicing TPS. Now to explain that statement I was in a 2-week assimilation class before I ever was exposed to my work area, so we learned how to be “toyota team members” and for us; that was learning discipline and accountability to the methodologies that created our culture. I was hired into the Plastics department where we had Injection Molding processes and other similar “molding” areas. I soon realized that one piece flow wasn’t part of the mix there. We almost felt like we were “breaking the rules”, I quickly learned otherwise. I think a common mistake with Lean today is that if you aren’t practicing or creating flow then you aren’t necessarily Lean. I guess we were never Lean in some areas- ah the horror!!:).
So if you take a few steps back and look at a bigger picture you can see there are many things involved with implementing “flow type” thinking in batch driven areas. So in my experience it started with our Production Control Planning department (PC). As a leader in production I was always given a forecast each month from PC for the next month which helped me drive our production numbers to a very close average based on the last 3 months of pull from the customer. Now some may ask, how can you forecast customer pull so precisely? I say this because its not so linear in other industries, as it is perhaps in automotive, but when you think about it- can I really predict exactly what car the customer will purchase on the lot down to one for one? Unfortunately we can’t, nor can healthcare leaders know what is always coming through the door so you do the best you can to minimize waste and understand and track pull.
So if we use the previous 3 months of data to drive our future need we would try to predict within a 10-15% window, meaning sometimes there was flucuation in customer demand based on options/colors etc, and we tried to deal with that with our daily “leveling” (heijunka) if necessary. Just because our takt time (pull dictated by the customer) may have been 60 sec a part “off the end of the line” didn’t mean every car had to be 60 sec. We had to average 60 seconds at the end. So for example, a high end Camry may take 67 seconds to complete a specific process and a base model may only take 53 seconds, so Im averaging 60 seconds which is part of our production planning to try and forecast this to the best of their ability at level daily which met the montly expectations as close as we could (remember 10-15%).
So if Im given a production sheet for example stating we are making X amount of Camry’s and X amount of Avalons and X amount of Venza’s then I look at those high level breakdown points as my first level of “kanban” creation. A kanban is like an instruction for production or inventory regulation whether you are running one piece flow or batch type. The next level breakdown may look at specific exterior colors or interior colors, then even type of car for example sunroof or not, leather versus clothe.
So to countermeasure our batch areas to be the most efficient the Japanese trainers introduced us to a concept called signal kanban. This allowed us to run just in time with Assembly even though we had 3-4 mold changes per day (sometimes up to 45 min downtime per shift). Of course we looked at SMED too and got our changes down from 30 min to 10 min, but to know how many you were running and when to do the mold changes was mimicking just in time in many ways in an area where you couldn’t do one for one.
So based on the monthly forecast I was responsible for determining how many kanban (we had 20 headliners per a kanban)- in our case the kanban was an empty cart brought back from assembly. I knew how many assembly were going to pull based on the PC forecast, so when they emptied a cart the kanbans were posted on a board signaling us when it was time for a mold change and this same method went into our next shift, so they ran different molds than we did minimizing the changes per shift. The shift leaders worked together and rotated molds often monthly or quarterly based on the pull and to minimize waste. If you can envision a board with hooks by type, color and model we hung our kanban cards showing how many were pulled and then needed to be built by the next shift, it worked beautifully and yet we were still running “batch” but the most efficient way, maximizing every second of time to create as much value as possible. This was similar in our Stamping department as well. The signal kanban was necessary to compete in an Assembly driven one for one off line process, otherwise you would need lots of space to stack parts up if we were just mass producing.
Note** Within our assembly processes for the headliner pieces it was one for one applying hardware for the lighting and sunroof. So we embedded work cells within our mold changing areas that were driven by a one piece flow but at a higher level pulled from a signal kanban mold change system, we implemented the best of both worlds.
The key point is ensuring that your people understand the purpose/importance to stick to the kanban set by production control as much as possible, often times there can be situations where you would have an urge to “stock-pile” parts or “add” extra kanban to boost your inventory for those dreaded downtime days. It’s an absolute discipline when you really run a “batch” flow when you stick to your guns and run to what the kanban orders (in essence the customer) not to an inventory level that doesnt relate to a pull type system.
You also must create standardized work that meets the expectation per process that meets the assembly pull that creates your signal kanban, there was minimal ability for buffer creation if you followed the rules. In our case had about 3 hours of buffer between us an assembly, that kept us just in time and following the rules. It also enabled us to be in a continuous improvement “problem solving” mode- we encouraged a problem awareness type culture(problems were good). We used the Define-Achieve-Maintain-Improve model (DAMI) to ensure if we met a standard we raised the bar and improved it, again this thinking ensured we were respecting our team members ability to think and continuing to raise the bar on our processes allowing us to see waste. Waste can lead to excess inventory if not addressed and that isn’t a friend and shouldn’t be welcomed so we must create that discipline with our people in understanding expectation and teach them to see abnormality to standard at all times. My role as a leader was to live that and develop others in that thinking.
It’s hard to tell you exactly “where” to start without seeing more, but I can share with you my experiences in how I learned based on what my trainers taught me in a “mold changing batch area”. I think the key where ever you decide to begin based on all the replies you will receive is ensure you are utilizing the power of your people, ensuring standards are set so people can see abnormality and ask why, then lastly have a production planning process that forecasts very closely to customer demand and set up your systems with the flexibility to change monthly, bi-monthly, or even quarterly depending upon your product(s) demand. I think our flexibility was the “cake” and knowing the expectation within a tight window (10-15% for example) was the icing. I hope this helped answer your question.
Until next time,