Let’s face it, the term “Supply Chain” has never really represented the real world of logistics and the relationships between buyers, makers and sellers. Multiple suppliers serve multiple customers who might in-turn serve other customers who use their products along with other items and materials to make other products for multiple customers, and so forth. The so-called supply chain more closely resembles a web or network than a chain. Always has.
But today, more than ever before, the network nature of the extended enterprise is an important reason why model-based planning and management tools have become an essential factor in the evolution of supply and distribution logistics operations.
It is said that a chain is only as strong as its weakest link; that is in the nature of a linear chain arrangement. A linear chain of activities can be planned and managed using simple rules and calculations. In the real-life supply network, complex relationships present a number of options and alternatives. When one “link” becomes unavailable, for whatever reason, there are likely to be one or more alternatives such as an alternate supplier, a 
different carrier or route, another mode of transportation, etc. And even when all links are fully operational, the existence of alternatives provides opportunities to optimize – engage the most effective supplier, partner or strategy whether that is the fastest, lowest cost, most reliable, most conveniently located, or however you care to define relative value.
Making best use of such a network is all about trade-offs; having an array of options from which to choose and recognizing the interactions and impact of each choice on overall results. Finding the “best” solution is a challenge, even for a seasoned planner who knows the industry and the partners intimately.
Of course computerized planning systems have been central to this task for many years. Planning programs are able to sort out all of the factors and possibilities, and make recommendations based on a set of rules embedded in the program code. And it works… more or less. These programs develop a plan that satisfies the embedded rules as much as possible but may find situations that are unresolvable or they may recommend actions that are simply impossible to carry out. And rules-based planning programs do not really think or reason – they just follow the rules embedded in the code. Helpful, certainly, but human reasoning is still required to resolve conflicts and choose the ‘best’ from among the choices.
A new generation of planning systems, now available, takes a different approach from the rules-based systems of the past. New model-based planners use a digital representation of the supply network that embodies the important characteristics of each resource. The planning process is carried out through simulating various combinations to test how each works and determine the overall result. After thousands of tries (this is carried out a computer speed, of course, so there is no long wait for results), the best alternatives are presented for the human planner to approve or adjust according to their own experience and reasoning. Remember, too, that the current situations and characteristics of the resources are based on real data (much of it collected from IIoT sensors and recent history) so it is much more detailed and up-to-date than any fixed rules or assumptions could be.
You can call this artificial intelligence. Or you can call it machine learning since the planning is continually adjusted and improved based on the system’s ‘experience’. Either way, the new planning paradigm is just what we need in today’s world of complex supply networks and increasing global competition.
