Many companies measure machine or process utilization, throughput, and quality, but fail to consider the combined effect of these factors on overall performance.
Overall Equipment Effectiveness (OEE) is a simple percentage that shows the ratio of actual equipment output to its theoretical maximum. OEE factors in equipment availability, speed performance, and quality, and is based on the premise that all production losses on machines and processes can be measured and quantified.
OEE is calculated using a simple formula:
The Availability accounts for unplanned downtime losses. It is equal to the actual machine/process running time divided by the total available time. Planned downtime events such as lunch breaks are not part of the OEE calculation.
The Performance accounts for speed loss. It is equal to the ratio of number of parts produced over the measurement period (shift, day, etc) to the theoretical maximum number of parts that could be produced if the machine or process ran at its highest possible speed. The performance calculation is simple when the machine or process produces one part per cycle. However, the performance calculation gets complicated when a machine produces multiple parts per cycle, and even more complicated when a machine makes varying numbers of parts depending on the job being run.
LETS has the unique ability to calculate an accurate OEE for operations where multiple parts are made during each machine cycle - even when multiple jobs making differing numbers of parts-per-cycle were run during the sample period. This makes LETS ideal for metal stamping and injection molding operations.
The Quality is the ratio good parts to total parts produced.
Traditional OEE comes up short in many applications. While the Availability and Quality metrics can be universally applied to all machine types, difficulty arises when the traditional Performance metric is applied to processes such as metal stamping where the true “ideal production rate” is more dependent on the tooling than the machine.
When the machine is capable of running faster than the tooling, the resulting OEE percentage is artificially low. This low percentage belies the true capacity of the machine while it runs that job, and lacks the resolution necessary to precisely gage the impact of process improvements.
Additional difficulty arises when a process makes more than one part per cycle, or requires more than one cycle to make a part. The former causes the Performance metric to be artifically inflated, while the latter greatly reduces it.
The errors inherent in traditional OEE can be manually factored out on a job-by-job basis, but this task becomes extremely difficult when trying to summarize OEE over a longer period of time. In addition, the calculations required to properly "weight" jobs of varying length are very complex.
LETS "fixes" OEE by doing the following:
During an 8-hour shift with 1/2 hour for lunch and two 15-minute breaks, a machine has a maximum availability of 7 hours (420 minutes). If there were 82 minutes of unplanned downtime during the shift, then the machine would've actually run for 338 minutes. The availability would be calculated as follows:
Availability = 338 minutes / 420 minutes = 80%
Running at full speed, the machine is capable of producing 6000 parts/hour (or 100 parts per minute). However, the three jobs that ran during this shift were not designed to run at the machine's maximum speed. The first job (which ran for 2 hours and 18 minutes) produced 4 parts per cycle. This job’s Ideal Rate is 75 cycles per minute. The theoretical maximum number of parts that could have been produced by this job was 41,400. The machine actually made 38,665.
The second job (which ran for two hours) produced a part every 6 cycles of the machine. The Ideal Rate for this job was 50 cycles per minute. The theoretical maximum number of parts that could have been produced by this job was 1000; the machine produced 950.
The final job (which ran for one hour and 20 minutes) produced one part per cycle, with an ideal rate of 80 cycles per minute. The theoretical maximum number of parts that could have been produced by this job was 6400; the actual number of parts made was 5900.
During the 338 minutes of running time in our example, the machine could have theoretically made a total of 48,800 parts for all jobs combined, but produced an actual total of 45,515. The performance percentage is calculated by dividing the actual number of parts by the theoretical maximum:
Performance = 45,515 actual parts / 48, 800 possible parts = 93%
Out of the 45,515 parts produced, 830 had to be later scrapped. The quality percentage is the ratio of good parts to total parts, and is calculated as follows:
Quality = 44,685 Good Parts / 45,515 Total Parts = 98%
The OEE for this example is:
Availability (80%) x Performance (93%) x Quality (98%) = 73%
You can see that although the component measurements - 80% uptime at 93% of maximum throughput with 98% quality - indicate a super-efficient process, when taken together as OEE, the process is really only 73% effective. There is still room for improvement.