Understanding the Queuing Guy’s Definition of Utilization

November 2019

It is a long-standing tradition in manufacturing to try to get machines and people to work 100% of the available production time. If you don’t, you are not getting the maximum benefit from those resources, right?  To fully appreciate why 100% utilization is not a reasonable goal, we first need to understand how the Queuing Guy defines utilization.  Who is the Queuing Guy?  He’s Rajan Suri, the founding director of the Center for Quick Response Manufacturing.  His application of system dynamics to manufacturing comes from extensive study of queuing, a branch of operations research that emphasizes how systems behave when there are people or things waiting in line.  Others who study queuing also have really good knowledge and intuition about how manufacturing systems behave.  So, let’s compare how the Queuing Guy defines utilization to what most practitioners believe.

In my interactions with manufacturing companies, it appears that the most common way to define utilization is to compare the time spent producing to the time available for production.  So, a machine that is available for an 8-hour work shift would be utilized 75% of the time if it was producing parts for 6 of those 8 hours.  This common way of looking at utilization causes some problems related to decision making.  If a machine is 75% utilized using this definition, a manufacturing manager may think that there is 25% spare capacity.  He or she may decide to load more work onto the machine.  The problem is that the 25% of time spent not making parts may be used for other things such as machine setup or troubleshooting.  If the machine is “busy” being set up for the next job, then it is not available to produce parts.  Similarly, if the machine breaks down, then it cannot produce parts until it gets repaired.  Because of that, we need to instead use a definition of utilization that is consistent with queuing theory; it is the percentage of time that the machine is not idle.  If any other activity is going on, the machine is considered utilized.

To make this difference concrete, let’s look at an example.  Say we collected data using a machine log and obtained the results shown in Table 1.  The utilization of the machine using the common definition would be 55%, while using the Queuing Guy’s definition, it would be 98.75%.  It is clear from the data (assuming that the data is representative of the machine’s usage over time) that the machine has almost no spare capacity.  In order to create spare capacity, a variety of improvements could be made.  Preventive maintenance may help to decrease the machine downtime.  Quality problems could be addressed in order to decrease the amount of time spent making scrap parts.  A setup reduction project might enable less time to be spent changing over to the next product model.  In the meantime, until improvements are made, it would not be wise to load this machine with more work.  And, it may alleviate congestion if some of the work done on this machine could be moved to an alternative one.

Thus, when the Queuing Guy says that you should plan for spare capacity on a machine; he’s talking about actual idle time, not time spent doing activities other than production.

Table 1.  Machine Utilization Example

Machine Utilization

To learn more about utilization, flow time, and their relationship, see Chapter 3 of Rajan Suri’s book, “It’s About Time.” Pages 77-78 are particularly salient to the focus of this blog.  Also, remember that when implementing QRM, we generally want to take a systemwide view by analyzing capacity for a larger subdivision of the company, such as a cell, a production line, or a complete manufacturing plant. We will address this issue in a future post.