A Guide to Implementing the Theory of
In this discussion we will follow the general argument of Umble and Srikanth (1, 2). We will assume a “perfect” operation, one with just process time and queue time or safety time. The objective is to show how rolling up local safety into global safety makes the whole operation safer and potentially more economical as well.
Below we have simple process of 6 sequential operations separated by local safety time buffers. Operations and buffers are drawn in proportion to the time taken; we will assume each operation takes five hours. Therefore actual processing time is 6 steps of 5 hours duration. The safety time between each operation is also 5 hours. Therefore total lead time is 6 by 5 plus 5 by 5 = 55 hours. We can, in this simple example, quote to our client a delivery time of 55 hours after starting the job. The blue marker represents the delivery due date. The 5th operation is a constraint.
Now let’s consider what happens if the first job is launched 5 hours late. Not a lot, the first local safety buffer is absorbed as it should and the job can start at the second operation on time.
What happens then if we have a 15 hour disruption at the first operation? Operation 2 starts 10 hours late. Operation 3 starts 5 hours late, operation 4 starts on time and thereafter the sequence continues as scheduled. The local safety buffers are adequate to protect the constraint and the delivery time from a 15 hour disruption at the first operation.
What then is the effect of a 15 hour disruption somewhere else in the system? Let’s say operation 4. Several important things happen when the disruption originates further into the system. Let’s draw it.
Operation 4 starts 15 hours late, operation 5 starts 10 hours late, and now operation 6 starts 5 hours late – BUT now the job is completed 5 hours past its due date.
One moment – operation 5 started 10 hours late. Operation 5 is our constraint! That means our constraint stood idle for some time. We lost output – we didn’t lose operating expense unfortunately. So throughput went down, and we started shipping late. The local safety buffers failed to protect us from this disruption originating deeper in the system.
Let’s rearrange the existing local safety then to better protect the constraint. All the local safety before the constraint is “rolled-up” into a global safety buffer in front of the constraint.
Now the system is protected from a 15 hour disruption at operation 1.
The system is also protected from a 15 hour disruption at operation 4. So the global safety protects the whole system from disruptions, regardless of whether they are near the start or are deeper in the system.
But there is more. Now, in our example, the rolled-up global safety is actually more than is required to protect the system from the disruptions. Therefore let’s reduce the size of the global buffer by 5 hours. This will still adequately protect us from 15 hour disruptions anywhere prior to the constraints. Let’s draw it.
Now we can protect our system adequately with less global safety than our previous local safety. Our throughput is protected, and total lead time is reduced. We can do more with less because we know what is important and what is not.
A major people issue is to have local workers and management understand the power of global safety and then to get them to relinquish their local safety and adopt global safety – and then once global safety is in place to make sure that it is not harmed by other erroneous policies. Generally these issues become one of “fear of losing control.”
Creating global safety is an exploitation step, reducing the amount of total safety as a consequence is a subordination step. We aggregate and subordinate our local safety everywhere to a small number of more effective global safety buffers located exactly where it is needed – in front of any constraints, control points or assembly points, and at shipping.
To return to the previous page press Alt key + left arrow.
(1) Umble, M., and Srikanth, M. L, (1996) Synchronous manufacturing: principles for world-class excellence. Spectrum Publishing, pp 139-144.
(2) Srikanth, M. L., and Umble, M., (1997) Synchronous management: profit-based manufacturing for the 21st century. Volume One. Spectrum Publishing, pp 195-201.
This Webpage Copyright © 2003-2009 by Dr K. J. Youngman