Go Back

Principle (Capacity): In steady state, all plants will release work at an average rate that is strictly less than average capacity.

Motivation

This principle states that it is impossible to operate at or above full capacity on average.  Note that this is true even if capacity is measured in average effective terms (i.e., by adjusting for downtime, yield loss, and other detractors).   The reason for this is that all manufacturing systems include variability--machines fail, quality problems occur, customer demands fluctuate, etc.  Eventually, this variability will cause the bottleneck to starve (or get blocked in a system with finite buffers).  Since time lost at the bottleneck is lost forever, such an event guarantees that the output will be below capacity.  The rarer we want this type of capacity loss event to be, the more WIP we must have in the system.  By Little's Law, more inventory implies longer flow times.  Hence, as we approach true theoretical capacity, both WIP and flow time will grow rapidly.  This is illustrated in the following figure (click here for mathematical treatment).
 
 

Of course, in practice, one never encounters plants with infinite WIP or flow time.  The reason is that in practice when utilization approaches or exceeds 100% and WIP and flow time starts to grow, some corrective action is taken.  Overtime is scheduled or demand is reduced (either through rescheduling or via unhappy customers canceling orders).  The result is that utilization falls back below 100% and the system recovers.  But, while capacity might be exceeded for a short period, over the long-term, utilization will be less than 100%.

Example

MRP vs. Pull System:  In a classic MRP (material requirements planning) system, work releases are controlled by a Master Production Schedule (MPS).  By determining the input rate, the MPS effectively controls the utilization of the production system.  Because higher utilization translates into higher revenues, it is common to aim for an input rate close to capacity (i.e., utilization close to 100%).  Unfortunately, this is precisely the portion of the curve where WIP and flow time increase most sharply in utilization.  Furthermore, because the true capacity of the system is impossible to know exactly--it depends on many uncertain parameters, including downtime, operator efficiency, yields, etc.--it is very difficult to manage the MPS so as to set the utilization precisely.  It is quite possible that errors in capacity estimates could lead to a utilization level above 100%.  Because of this, traditional MRP systems were commonly plagued by the overtime vicious cycle (Overutilization leads to increasing WIP and flow times, which cause late customer orders.  Eventually the system requires intervention, often in the form of overtime.  This brings utilization down and the system recovers--only to have the sequence of events repeat.)

An alternative to the "pure push" approach of MRP is pull production.  In a pull system, there is no exogenous schedule that determines the release rate.  Instead, releases are triggered by changes in the production system itself.  Production occurs to replenish voids in inventory stocks.  The result is that WIP is controlled directly, instead of resulting from an implicit choice of utilization as occurs in MRP.  Since WIP and flow time are tied to one another by Little's Law, flow time is also controlled.  This is a fundamental reason that pull systems are less prone to the overtime viscious cycle than are MRP systems (click here for principles related to push and pull production)..
 

Go Back