scroll

Continuous Planning

Continuous Planning

As data becomes more and more available in real-time, supply chain planning systems must respond to the events as needed on a continuous basis. Much like an autonomous car responding (or not responding) to the data it receives from its sensors and makes changes accordingly. “As needed” is the key. It may sound simple but it is very complex to decide when and why we should respond to changes that may or may not impact the supply chain plan and execution thereof. Let’s say a part number is arriving 2 days late. Do we need to re-plan? Does it impact delivery of a very important customer? Does the customer care if it is 2 or 3 days late?  Or, making the product 2 days later might create a huge disruption in so many other orders because of shortage of capacity already allocated to other orders!

With every event, we need to have a mechanism to evaluate the potential impact on the plan. This means possible changes in our delivery performance, in cost, or even the strategic impact of it such as late to market or dissatisfaction of a very strategic customer.

One simple and unwise approach is to plan based on every event. Even if this were computationally possible it does not work because it causes instability of the outcome. Imagine moving all the gate allocations in an airport if one airplane arrives earlier than expected just to accommodate disembarking of the passengers! Another approach would be to use AI strategies in order to make intelligent decisions based on cost, lateness, asset utilization, customer importance, order volume, strategic objectives of the company, the plan stability, and the computational complexity (number of events received at once or number of orders impacted). Adexa has taken the latter approach by creating algorithms that can evaluate the impact of the event and decide to what extent re-planning is needed. The output of the algorithm could be re-plan all, partial repair of the plan, incremental plan, or do nothing. The system makes use of a self-adjusting “point allocation” mechanism to figure out what the best strategy is. The best is decided by the one that produces the least impact on the above factors combined.

As an example, a late arrival of a material may cause 50 orders to be late by an average of 15 days. Given the current allocations, moving the orders by 15 days or so can cause under-utilization of equipment. In addition, 5 of the 50 are high priority customers. By making such evaluations, the implications are that a partial plan change is the best approach to take in order to come up with a better short-term plan. Another scenario: let’s assume that a  high priority order with large volume comes in unexpectedly. The immediate reaction of the system would be to do an incremental plan. An incremental plan may cause displacement of lower priority orders and potential lateness thereof. Since the resulting change produces a low impact score then this path is taken and a commitment is made to the high priority customer! As one can imagine, there are many possibilities such as dock worker strikes, cargo ships on fire, earthquakes at the supplier location, or simply an equipment breakdown for a day or two. Each of the above events may or may not have a major impact on the plan and the delivery performance and cost. The key for the system is to recognize at what point a “re-plan” is needed!   Watch the short video on how continuous planning works from Adexa’s perspective.

Share:

Share:

Leave a comment

Support
Request Demo