The planning engine has always been at the core of manufacturing management systems from Material Requirements Planning (MRP) to Manufacturing Resource Planning (MRPII) to today’s Enterprise Resource Planning (ERP). For much of that evolution, the effectiveness of the planning process was limited by the initial design of MRP planning, which was designed to work within the limitations of early computer technology.
Certainly, computer technology has evolved a great deal in the last fifty or sixty years, and planning systems have advanced as well. The first major breakthrough in planning hit the market in the 1990s in the form of what is called Advanced Planning Systems (APS). The original MRP planning systems were (and still are) based on a two-step process: plan material acquisition from the master schedule using bills-of-material and inventory data without regard to resource capacity (infinite capacity assumption); then identify and handle any resource conflicts manually with the help of Capacity Requirements Planning. This is an iterative process that relies heavily on “standard” lead times, capacity, and other assumptions.
Advanced Planning Systems
APS, introduced in the 1990s, fundamentally changes the way planning is done, most notably by discarding the infinite capacity assumption. Materials and resources are planned simultaneously, resulting in a more complete and realistic plan. Even more importantly, APS uses real shop data – a current status of running jobs, actual (calculated) lead times for planned work rather than standards – and advanced mathematics to develop a plan that fits the materials and capacity available.
APS is a relatively expensive optional module in most ERP systems. The extra cost is justified because of the added value and it is optional because many companies don’t actually need advanced planning, their processes are relatively uncomplicated and resource availability is not an issue, and many that could benefit from APS are simply incapable. APS is far from “plug and play”. It takes smart, dedicated planners to properly manage the data and the process, and interpret the information the system provides. APS also relies on good, timely reporting of shop activity, accurate bills, and solid up-to-date inventory balances and availability. And even advanced planning is a reflection of the limitations of computer system power and speed as it existed just before the turn of the century.
Welcome to Twenty-First Century Planning
What’s different about the new generation of planning systems? The new planning harnesses the power of artificial intelligence, machine learning and simulation to automate the planning process and bring it closer to reality and experience. It doesn’t eliminate the role of the human planner, however. It empowers the planner to make better decisions and get better results – more on-time completions, higher utilization and efficiency, higher quality, and a more controlled and reliable plant floor.
The new planning approach builds a detailed model of plant behavior from activity data (largely collected from machine controls and sensors) that reflect actual status and experience, rather than averages or standards. As new data is collected and new requirements entered, the planner simulates what is likely to happen based on that detailed model and statistical analysis of the way things have gone before. The new plan is grounded in reality and accurately captures realistic work flow, efficiency and utilization experience.
With a more realistic plan, the company can quote real completion dates to customer, and meet those dates without resorting to extraordinary measures or at the expense of other customer orders (delays caused by expedited orders taking precedence) – in other words, no surprises.
Automation and Continual Improvement
While the new approach to planning is certainly more complex and sophisticated that APS, it is actually easier to manage and use because most of the functionality is automated. Managers are given alerts and suggestions – managing by exception as the software takes care of the routine, minor corrections needed to adapt to normal day-to-day variability.
The best news is that the system actually “learns” the longer it is in use and the more experiences and variations it encounters. The model is continually adjusted as new data flows in.
Furthermore, the model also becomes a stake in the ground against which a continuing stream of data is compared. Any deviation from the model’s prediction is noted, and the model can be updated to create a new plan and also to alert managers to evolving issues or concerns. This is machine learning and artificial intelligence in action.
This new approach to planning is truly a revolutionary change; a result of the evolution of computer hardware and software that continues at an ever accelerating pace.