Sunday, November 19, 2017

What's the Difference between a Buffer and a Shock Absorber?

Lately, in industry publications and social networks, you can read a lot about shock absorbers and how to apply them in supply chain planning to reduce variability or absorb it. Apparently, many thought leaders have been replacing and complementing buffers with these shock absorbers… at least in theory. Not wanting to get into anybody’s conversations but I do have my own view on the topic. In my opinion a buffer and a shock absorber are not the same thing and therefore I do not use them interchangeably.

Let me explain my thinking. A buffer is a medium (inventory, capacity, time) that blocks and swallows up a signal... like a firewall prevents heat getting through to the other side. A shock absorber takes the signal in and absorbs it to a certain degree. When the signal is big enough and the absorber‘s buffer is exhausted, there is no holding back anymore.

To use an analogy, think of  driving your car into a pool of mud at 50 miles an hour and you know how a buffer acts. Drive that same car through a pothole 10 inches deep and you get an idea what shock absorbers do… they let you feel what a big change in variability there may be, while a buffer simply stops you right in your tracks.

The problem with buffers is that a part equipped with a buffer doesn’t ‘see’ when a big change in demand is on the horizon. The buffer simply blocks all that’s coming at it. The shock absorber, however, swallows up the little changes but carries through the big ones. That can be very helpful in some situations but annoying and inefficient in others. This is why both buffers and shock absorbers are awesome tools to take the noise out of your supply chain. You just have to use them in the right situation and not mix them up because that can lead to chaos and bad planning.

In the same way one certainly doesn't want to drive a brand new BMW into the mud to slow down when there are a few bumps in the road. But coming down from a mountain pass in a 14 wheeler with failing brakes, you’d be happy if there’s a buffer coming up absorbing all that energy.

Friday, November 10, 2017

Who needs a rough-cut capacity check anyway?

Just recently I was visiting with a manufacturing company that builds parts for the automotive industry. It’s a repetitive business and they use Make To Order and Make To Stock strategies. The problem is that there is a humongous backlog, because there isn’t enough capacity and there is too much of the stuff nobody requires and too little of the stuff they desperately need.

I do not want to go into the specifics why I think the backlog is there (no flow and therefore wasted capacity, missing parts and therefore orders stuck in the line wasting capacity, no scheduling system and therefore inefficient schedules and therefore wasted capacity). No… I want to ask why there is no planning? All over the internet there are discussions going on where everybody pitches in with the newest and greatest planning methodologies… Advanced Planning & Optimization, DDMRP, IBP, S&OP, agile, lean. Then there are the software solutions that supposedly make it all happen… and eventually everybody is so keen to improve forecast accuracy with demand sensing or probabilistic forecasting. Everybody is out there to get the best and coolest planning solution.

Why? Why is so much money spent on implementing the greatest software tools, applying the best methodologies and engaging thought leaders and evangelists when - in the end - orders are scheduled in MS Excel (where there is no information about capacity, inventory or demand), carefully crafted strategy (like are we making to stock and how big is our shock absorber) gets blatantly ignored and all the big rules fly out the door when the sales rep claims a super important customer request?

The last six customers I have worked with did not do planning. That is a heavy statement. Especially if I would tell you the names of the six… you’d know them all by name. So what do they do? Let me start with what they do do and then we’ll discuss what they don’t do. Some of them work with a forecast (four of the six).  Of those that do, the forecast comes from the Sales department. Some of these forecasts are in a different system than the system in which production orders are scheduled (two companies of the four forecasting, maintain an interface) This raises the question about what objects are transferred through the interface. Both companies that maintained a forecast, and had two systems, transferred planned orders (supply elements) over into the system that takes care of scheduling. The other two companies that maintained a forecast - but within the same system as scheduling - did not make a difference of a long term, a mid term, a short term horizon or a frozen zone. It was all one long period in which there were planned orders, fixed planned orders or production orders. The two companies that were not maintaining a forecast were going by customer orders only.

What does that say about planning? We have three situations (please note that a forecast can also drive a schedule for MTO products, as in that case capacity for MTO production is reserved on the line):

  1. No forecast, customer demand driven
  2. Forecast drives schedule, two different systems
  3. Forecast drives schedule integrated in one system

Refutation for situation 1.: if you go by customer orders only, then there is no planning. And unless your customers accept the time it takes to procure materials, fabricate components, build sub assemblies and assemble the finished products, they will go someplace else to get the product they need and want.

Refutation for situation 2.: if you maintain a forecast in one system, perform a rough capacity check for the long term. This is necessary so that you can decide whether you have the capacity to fulfill the expected demand in-house or you’ll have to talk to a trusted supplier who’s helping out and provides some extra capacity. If you don’t do that and simply hand over supply proposals (planned orders) to the scheduling system, then I’d call that bullying but certainly not planning. If anything, hand over demand (instead of supply) so that your MRP Run can at least do some planning in the short term. It’s probably too late to do anything good anyway (what is a production scheduler to do if the demand causes a utilization of 400% plus?).

Refutation for situation 3.: The two companies that forecasted and scheduled in the same system had discrepancies between how they wanted to operate and how the system was setup to support those efforts. In other words: they were trying to do the right thing but didn’t get the right advice on how to set up that system that was supposed to ensure an integrated planning process. There were no planning horizons, the planning strategies were not setup correctly, the sales orders didn’t consume the forecast properly, the sales availability check did not work and therefore demand was transferred incorrectly… etc pp.

And here is the sore point. I don’t think it matters much what system one uses or what methodology one applies when the construct (the third floor) is not supported by the basement. All those methods are great when applied to the proper situation. And all these system have all the function and features one needs to plan effectively and efficiently. But deciding on the right methodology and right tools isn't even half the way towards the end goal. You must support the philosophy with sound structures (planning horizons, planning strategies, planning hierarchies). What good is it if you place strategic buffers but you send the wrong elements (supply orders instead of demand) through to the wrong transaction at the wrong time?

It is absolutely justified to invent newer and better systems and methods, but we also - or should we say primarily - need to make sure the foundation is build first and is solid. Too often I hear „this software is crap“ or „we need to replace that old method with a new one“. Remember… just because you can‘t fly an F16 doesn‘t mean it‘s a shitty plane!