My first professional job after graduating from college was a Junior Industrial Engineer tasked with time study duty and implementing methods improvements. Over the course of several years, I completed thousands of time studies with my trusted stopwatch. Since those days, I have learned several different approaches to setting the standard time like using the average time method, using the lowest repeatable time method and using the shortest observed time method. So which method should you use set a standard time?
My IE training in establishing a valid time standard was drilled into me to capture a repetitive method, rate the work pace, record the proper number of cycles, red circled non-standard times and take the average time as the standard. For example, if 10 cycles were recorded as follows in seconds (50, 53, 49, 47, 49 54, 46, 51, 65, 52) I would have red circled (not counted or excluded) the 65 second time and calculated the standard time for this task as 51 seconds (rounding up to the largest second). Also, I would have added a PF&D allowance (Personal Fatigue and Delay) boosting the time even higher but lets leave it out for this example.
When taught how to do a time observation to set the standard time in a kaizen event, the method was a little different. The observed work was set to a repetitive method and the lowest repeatable time was established as the standard time. Using the same times recorded in the above example, I would establish the standard time as 49 seconds.
Following the teachings of Taiichi Ohno, he believed that the best standard time is the shortest observed time. Again, using the same 10 time observation above, the standard time would be set at 46 seconds. Ohno stated that we should focused on the method used to complete the task in shortest time , trying to repeat it exactly and eliminating the causes in the other observations that prevented the operator from repeating the shortest time.
So which is the standard time in our example, 51 seconds, 49 seconds or 46 seconds?
Let’s look at taking the average observed time method resulting in the 51 second standard time. From the range of timed observations, we see that variation in the process exists. This variation could be from anything from dropping a screw, part mis-alignment, adjustments, location of parts, a miss feed, etc. Regardless of the cause, this variation is non-value added or a waste. So why would we include waste in our standard? What if the range was larger? Would we still accept this method? With our lean approach we should not accept this waste in our standard so it appears that using the average is not the best method.
As for the lowest repeatable time method, is this method of setting the standard time any better? In our example, it removes allowance for some of the waste however elements of waste would still be accepted. In our example, it is slightly better. But what if, based on our observations, the lowest repeatable time was above the average? In this case, it would include more waste in our standard. Again, this may not be the best method.
That leaves us with using the shortest observed time method as the best standard time. Some may argue with Taiichi Ohno saying that this one shortest recorded time is too strict, not repeatable, a fluke occurrence or even a small miracle. But instead of arguing against the merits of using the shortest time, we could put the same efforts into doing exactly what Taiichi Ohno tells us. Determine the reasons why the other times failed to reach the shortest time and eliminate them. Our efforts will be rewarded with the shortest time as the easiest time that can be repeated. Isn’t that core in our lean principles, seeing and eliminating waste?
5 comments:
Great blog Mike. This topic comes up lots. When I have my Six Sigma hat on I get worried with Ohno's method due to fear of loose measurement systems. Perhaps instead of simply accepting the one lowest time we should verify our measurement system and measurement process is robust. But I do like the aggressive approach going with the lowest number gives us. It says, "let's aim for the best and quit being scared." Nice blog - as usual.
Not to address the correct method of calculating standard time, one source of genius in lean is to train shopfloor teams to conduct their own time studies. An IE can assist them until they acquire enough skills, but it's much different when a shopfloor improvement team decides to analyze their process than when the proverbial shirt-and-tie IE stood by menacingly with a stopwatch.
Thanks for the comment Ron. Yes, It is an extremely aggressive approach that pushes us to the edge of our comfort zone. But the concept is the absolute elimination of waste.
Hi Karen,
Way back when, in a galaxy far, far away, I used to be one of those shirt-and-tie Industrial Engineers that trolled the manufacturing floor looking for jobs to time. We may have been menacing to some however I tried to be open, honest and cooperative with any person I timed to set a fair and accurate standard. Once we started down the lean path, we did teach the ways of the jedi timestudy masters to all kazien team members to set their own standard time. The shopfloor teams did extremely well with this new skill.
One way we might answer what is the best way to set a time standard is, first, to define what a "standard" is. As an industrial engineer early in my career, we defined a time standard as "the work pace that can be maintained by an average, well-trained employee all day long without undue fatigue while producing acceptable quality work." To this,like you Mike, we added a personal and miscellaneous allowance of 15%. And, like you, I won't go into that now.
Using the above definition of a standard, the time to complete a task can be seen to be influenced by three basic factors: 1) skill, 2) effort and 3) method. This means that any variation in the times an engineer may find in a conducting a time study could be the result of any (or combination)of these three factors. It's not only method which is the focus of Lean.
In reviewing the approaches you outline for setting standards, a very powerful one seems to be missing ...performance leveling. The weakness we all experience in using actual times from a time study to set a standard is that we are left with guessing at what a fair and accurate time expectation should be. As you outline, is it the lowest repeatable time, the average time or shortest observed time? There are even other approaches including upper quartile and modal time.
Using the above definition of a standard, it's only coincidental, then, that the true standard is any of these. This is where performance leveling comes in. With performance leveling, you observe and capture the actual time each employee takes to perform a task. This time is then adjusted (i.e., leveled)to reflect 100% performance. For example, if an employee takes 5.00 minutes to perform a task during a time study and you access the performance level of the employee at 85%, the time for the task at 100% would be 4.25 minutes (5.00 X .85 = 4.25). Using this approach eliminates the guesswork. However, that said, it takes specific training to learn to effectively access employee performance level.
In the above definition of a standard there is also a key assumption...that everyone is following the same method. This assumption aligns perfectly with Lean in that Lean suggests standardizing work first.
This also gets us back to another fundamental issue with setting time standards. What's the basic purpose of determining the time to perform a task? If it's going to be used to evaluate the performance of an individual employee there are only two standard-setting approaches that result in the accuracy needed: 1) leveled time study (described above) and 2) pre-determined time systems. Without going into detail, for those who are unfamiliar, PDTS are dictionaries of pre-established times for performing specific cycles of activity already leveled at 100% performance and are +/- 5% of true accuracy. These are powerful tools not only for establishing standards but, because of how they are applied, in streamlining methods. Two examples of popular PDTS used in manufacturing today are Method Time Measurement (MTM) and Maynard Operation Sequence Technigue (MOST). There are even ones especially developed for white collar tasks including Advanced Office Controls (AOC), Modular Arrangement of Predetermined Time Systems (MODAPTS).
I've gone on too long so I'll quickly summarize my opinions. If the point of establishing a time standard is simply to get a reasonable understanding of the time it's currently taking to perform a task or process so that improvement efforts can be launched, "average time" should be all that is needed. However, if the time will be used to measure individual performance then a more robust and accurate approach is definitely called for.
Post a Comment