Here’s a bold assertion: The best schedules result from using the right descriptive model of the manufacturing environment.
Some would argue that the best schedules result from some type of optimizing logic. The fact is that optimal solutions are obtainable for only very limited special cases. Few real environments fit into those special cases–which is unfortunate for everyone else.
There is no doubt that a good algorithm is necessary, but even the best logic applied to an poor model of the real-life production environment results in a sub-par result, with limited utility for the company it’s supposed to support.
Here’s a good example where modeling is important. Imagine a very common and simplified manufacturing process with several steps. Each step requires something from the previous step–it’s like an assembly line where the output of one step is the input to the next one. The inputs and outputs could be materials or partially-completed products. Here are two ways to model this type of production:
Generated and Consumed Resources
You model each step as an activity that generates resources (such as partially-completed products), and eliminates the resources generated by other activities. These consumed and generated resources create work-in-process (WIP) inventories that are dynamically increased and decreased as the activities are executed. The scheduling logic won’t schedule an activity unless its input resources are available. The processing sequence is indirectly imposed by the availability of resources.
Predecessor and Successor Activities
You model the sequence of manufacturing steps attaching predecessor and/or successor requirements to each activity. That is, you define a precedence network that does not model the intermediate resources generated or consumed. In this approach, each step is defined and sized to produce exactly what the next step needs. This approach is common in project scheduling and make-to-order manufacturing environments.
If you can use either modeling approach, how do you choose between them? Here are some questions that can help sort this out: (Spoiler alert: The answer is usually a hybrid of the two.)
- Are any intermediate products sold or outsourced for finishing?
- What are the reasons for maintaining an inventory of intermediate products or materials?
- Are you modeling intermediate products as a legacy of an MRP system that is designed around materials availability?
- Do you have existing materials or intermediate products in inventory that should be used up as the production schedule is produced?
- Does one step in the process sometimes produce more or less than the next step requires?
By now you know that a good scheduling app should support both modeling approaches. Utilizing both approaches to your best advantage is worth some thought–perhaps even some trials of the alternatives. And your production needs may evolve or change as time passes, too. A flexible approach in your software will let you do just that.