Smart Manufacturing and the Problem of Legacy Planning Process Models
Smart or advanced manufacturing is getting a lot of attention for the last few years. Obviously nine manufacturing innovation institutes have already been established and six more are planned by the end of 2017. There are many related other important concepts such as Industrie 4.0 and industrial internet which cause a little bit of confusion because many of these terms refer to very similar concepts. There is also an intense competition between the U.S. and the E.U. and as argued by Dieter Wegener, Siemens’s coordinator for Industrie 4.0, this process is fueled by the consumers' demand of customized products - efficiently manufactured and quickly delivered.
To eliminate the possible confusion, let me start with what I mean by smart manufacturing. National Institute of Standards and Technology has a very neat definition which states that smart manufacturing is "fully-integrated, collaborative manufacturing systems that respond in real time to meet changing demands and conditions in the factory, in the supply network, and in customer needs".
What is obvious from this definition is that it describes a manufacturing enterprise which behaves essentially like a living organism. And no, I am not going to be pedantic and explain the similarities between potential smart manufacturing sub-systems and their living organism counterparts :) However, it is straightforward to see that if a so-called smart manufacturing system only responds to a disruption with a myopic behavior, then the overall efficiency won't be what it is supposed to be. Therefore, it is natural to expect that a smart manufacturing system will incorporate a planning sub-system which can project into future to see what possible sorts of outcomes are possible from a course of action and which can do these in real-time or near-real-time.
At this point it makes sense to take a look at the classical planning process upon which many - if not all - state-of-the-art advanced planning and scheduling applications are built.
It is a very straightforward process model which says define your objectives, collect necessary data, the cool part of your software starts working and computes many different alternatives and chooses the best for you and then you execute your plan. Although this looks very nice on paper as a process model, it is very problematic in reality. Let us try this process model:
Suppose a customer requests a change in some of his orders. What you should do in this process model, because the process model is linear, is to plan everything again.
Say you don’t need to care about your customers because you are a monopoly. You want to see if it is better for you to work three shifts a day and omit working during weekends instead of working two shifts throughout the week. Because this means a change in data, you need to plan everything all over again.
Say you are a monopoly and you are so much flush with profits that you don’t care about anything. Even in this situation, there is something that none of us can escape, and that is the fact that every plan inevitably deviates from its course. When that happens, you need to plan everything again.
So if you have this process model and if you need to re-plan repeatedly whenever a change in your assumptions happen and if your planning engine is not blindingly fast, then naturally you will not use such a software to manage your daily life let alone in a smart manufacturing environment which requires at least a near-real-time response. What is needed is a significantly more agile approach to planning.