Recently a customer of mine asked what they should use as their ideal cycle time in a particular manufacturing workstation. More importantly, they wanted to know which average was most indicative of a performance measurement, or benchmark time for a given work task. Would it be their mean (what we know as average), median or mode time? Surprised to hear that there are three possible methods of calculating average? Don’t be. While each method provides a different answer when determining averages, only one should be used to set your benchmark cycle time in a given production workstation. Only one points to the ideal performance measurement and only one should be used to track variances from task-to-task and from day-to-day. So, which one is it?
Finding Your Benchmark Cycle Time
No ERP or MRP system is intuitive enough to resolve the issues of high cycle times. Your system can reconcile inventory, reconcile production volumes and providing insight into the variances in manufacturing times, but in order to set your ideal time in a given work station, you have to be willing to witness production happen in person. Only after determining your ideal time can you then track the variances from these benchmarks. Now, by no means am I implying that every single operation should take the exact same time. That’s simply not possible. However, if you set your ideal time in a given work cell, you can then track your variances within your ERP or MRP system afterwards.
Using the Mode Time as the Perfect Cycle Time
Again, the three methods of determining an average include the mean, median and mode cycle times. The “mean” is simply the total of all sample times divided by the number of iterations or number of work operations. The median involves taking the entire sequence of numbers and placing them in order. Next a simple calculation is used to determine the median number within that sequence. However, the mode is the number that occurs the most often within the sequence.
It is the mode that isn't impacted by any highs or lows in the variances within the sample portion. It is simply the one cycle time that occurs the most often and consequently, it is often the one that sees the least amount of downtime. This is why it might be ideal to use it as your benchmark time, one you measure all future performance against and one you continually strive to attain.
The three approaches to calculating averages are explained below in addition to a graph that came from the aforementioned post.
A. Mean: Total up all the sample cycle times and divide it by the number of operations. In the table above, this means taking 51 minutes and dividing it by the 20 samples. The mean time of 2.55 is the LIGHT BLUE straight line.
B. Median: For the median, we lay out all the numbers from the sample portion in sequence. We then use the answer from the equation below to isolate the median time.
- Median: {(n+1) / 2}
- N= sample size: which in our case is 20 operations
- Median: {(20+1) / 2} = 10.5
2.15, 2.15, 2.3, 2.3, 2.3, 2.3, 2.3, 2.3, 2.45, 2.45, 2.45, 2.5, 2.5, 2.5, 2.5, 2.55, 3.25, 3.25, 3.25, 3.3
C. Mode: Again, the mode is the cycle time that occurs the most frequently in our sample. This would be the 2.3 second time which is encountered six times in this sequence. The mode time is the RED straight line on the graph, and is represented by the red numbers in bold below.
2.15, 2.15, 2.3, 2.3, 2.3, 2.3, 2.3, 2.3, 2.45, 2.45, 2.45, 2.5, 2.5, 2.5, 2.5, 2.55, 3.25, 3.25, 3.25, 3.3
The video above is a summary of the three methods of calculating average: mean, median, mode. This comes from the post: Cycle Time Tracking & Variance Analysis in Excel for Small Manufacturers
The intention here isn’t to imply that you should never encounter downtime. If all you do is analyze a given workstation, and isolate the causes of work stoppages, then you’ve already set the table for increasing production throughput. I’ve often been accused of trying to take a “perfect world” approach to manufacturing, or to try and install a process where there are absolutely no delays. That’s simply not the focus of this exercise, and it simply isn't possible to have no lost time. Instead, it’s to analyze a given workstation, take sample cycle times and isolate the most ideal time when no work stoppages are encountered. That’s it. As you eliminate the causes of delays, you’ll begin to see your mode times occur more frequently. The idea is to get as close to perfection as possible, as many times as possible. That effort will produce results.
Comments