MRP

You don't trust what you can't understand

In the ´70s and ´80s, decision support technologies were developed with the aim of planning the material requirement. The basic principle of these Materials Requirements Planning (MRP) systems, and later Manufacturing Resource Planning (MRP II) systems, is that you enter conditions such as orders and forecasts, and receive a "wish list" with capacity requirements and suggested start days for production orders.



MRP is a complex mathematical model which is hard to survey. A small change of data, such as the size of an order, can have consequences that seem illogical and are therefore hard to understand. As a result, the planner often makes his own decisions based on practical experience, because:

  • you know that the calculations are based on average values and guesses at lead times, and are therefore probably wrong anyway,

  • the software system is not much help in pointing out what and how much needs to be changed, and
  • time is of the essence, and a planning cycle can take as much as several days to compute. This means the schedule is obsolete when it is presented, and therefore useless.

Seeks the optimum at each level, and forgets the whole

MRP is frankly ill-suited for use in a changing environment, where new calculations have to be made often. When we make a new calculation, several optimizations will show a different value than last time. These are moved up to the next level and lead to suboptimization. The interesting thing is that the whole, the big picture, is absent from MRP.

Requires long delivery times to the customer in order to be stable

In order for MRP to work, the delivery time from the supplier plus the production time need to be shorter than the delivery time to the customer. When this is not true, you have to resort to emergency solutions such as forecasts and frozen schedules.

Forecasts are guesses, and should be used for strategic decisions ? not operative ones. And in a time when adapting to the customer is a success factor, freezing your schedules is not a very clever thing to do.

Slow systems cannot control a dynamic environment

The complexity of the calculations means that it is realistic to compute MRP once a week. Using a week-based method when each hour is a competitive advantage means that you will have to "put out fires" every day to make deliveries on time.