1. The Problem

In recent years, the technologies available to capture and calculate OEE have become more widely available. Coupled with an associated drop in the magnitude of cost to deploy these technologies - from home-grown solutions to out-of-box applications - many manufacturers are beginning to deploy these systems. With the charge being led by Food and CPG, many Discrete manufacturers are now anxious to reap the benefits of this type of system. Even the Pharmaceutical Industry, whose cost of manufacturing is typically a negligible component of overall cost of goods sold, is rapidly adopting OEE as a key metric. As an aside, at a recent Pharmaceutical OEE Benchmarking seminar, co-sponsored by Q-mation, Inc., the conclusion reached by the firms present was that typical current OEE is about 30%, with world-class achievable goals for OEE set at about 65%. That would be equivalent to doubling your manufacturing capacity while adding NO additional equipment!

For years, many manufacturing companies have been measuring the Efficiency of their Lines and Work Cells in such a way as to “mask” many of the causes of lost efficiency. As a logical result, management focused on the numbers being reported, and not on what is included or excluded from the measurement. This is one of the reasons OEE (Overall Equipment Effectiveness) is such a valuable measure. It allows a company to look at all sources of lost time and lost production. For example, in the past, company management would succumb to the current rhetoric, and manufacturing managers would have to pick up the banner and run with it. Buzz-words like “zero-defect” became popular, and when forced to comply, manufacturing plants did see Quality Rates improve, but at a significant cost to overall throughput (Performance), and machine uptime (Availability). Now, manufacturing plants can tie Quality, Performance and Availability into one metric. The plants can now report their Overall Equipment Effectiveness (OEE), without masking a major downturn in any particular area.

Again, the OEE concept is not new, so what is new ? The ability to:

• Accurately capture data, at high resolution (minutes, seconds) without distracting operators.

• Report metrics in a timely fashion to both management and front line operators so they can actually do something to improve performance.

• Easily and securely share data over the internal plant network (intranet).

• Deploy cost-effective system (6 to 9 month ROI) without incurring costs that require approval from the Board of Directors (under the radar).

As a result OEE has emerged as the leading approach for accurately measuring true plant productivity.

 

2. What is OEE exactly ?

 

By definition, OEE is the mathematical product of Availability, Performance, and Quality percentages:

 

OEE = Availability Rate x Performance Rate x Quality Rate

 

Availability Rate is associated with Downtime Losses (non-production) like:

 

 

Performance Rate is associated with Speed Losses resulting from:

• Running a production system at a speed lower than the Theoretical Run Rate for that SKU on that Line / Machine / Work Cell.

• Short stop failures such as jams, overloads, running out of components, or other faults that can be cleared without maintenance intervention (many lines have 1000 or more short stops per week which results in a massive reduction in output).

Quality Rate is associated with Yield Losses resulting from:

• Material Rejected

• Material Reworked

Obviously, each of the components detailed above will result in some loss of productive operating time. As shown below, TOTAL PRODUCTIVE TIME is merely a fraction of the TOTAL AVAILABLE TIME.

 

 

In the theoretical world, a line exhibiting 100% OEE would run at 100% of its theoretical rate, never shutdown during scheduled hours and produce perfect first-pass quality. The only time the line would stop was when there was literally nothing to produce. In the past, manufacturers would plot Quality, Uptime and Throughput separately because, quite frankly, they could not find a reasonable way to correlate all three. In the example below, it looks like our Quality is steady, and our Uptime and Throughput are climbing – so why are the plant numbers not reflecting the improvement ? Unfortunately our Quality graph is produced in shipping, and does not represent first pass quality off the line.

 

 

 

If we accurately captured Availability (Uptime), Performance (Throughput) and Quality (First Pass), the graph would look more like the one below, where our OEE – THE PLANT’S REAL PERFORMANCE – is flat to down for the year.

 

 

If we can’t see it, we can’t impact it. If our charts in the hall are a month behind, we can’t actually do anything with the data they report.

 

3. Automated Collection of Data and Calculation of OEE

 

With modern software tools, the information available in the PLCs can be leveraged to produce sophisticated real-time reports that allow manufacturers to fully understand all of their sources of lost productivity and to motivate the plant team to continually optimize OEE. The graphic below provides an example of the types of reports that can be generated by these systems. The data is collected automatically, in real-time, and provided securely throughput the facility in a simple web-browser.

 

 

 

4. Back of the Envelope Calculation of OEE

 

Given that many manufacturing companies believe they are running at an Efficiency of 85% to 90% (which is true, the way they currently measure efficiency), it is very helpful to get an idea of the true potential for improvement by performing a rough, Back of the Envelope Calculation of the current OEE. This calculation will lack the detail (breakdown of Availability, Performance, and Quality and the detailed reasons behind all the stoppages) but it will provide a good idea of where your production lines really stand at the moment. This back of the envelope calculation is easy to perform.

• Select a Typical Line - Select a Line, Work Cell, or Machine to perform this calculation. Select one for which you will have accurate production numbers.

• Determine the OEE Calculation Time Period - Select a period of time that is long enough to account for any major, periodic, Availability related downtimes that occur. For instance, if your line runs continuously for 2 days and then must be stopped for a CIP (sanitation) for 4 hours every 3rd day, then run your calculation over the 3 day period.

• Determine Time Not Scheduled - During the OEE Calculation Time Period, you need to determine how many minutes the line was not scheduled to be used for any productive purpose (changeover, sanitation, PM, etc. are productive purposes). This is the time that you had no production requirements. For instance, if your plant only works 2 shifts, then 3rd shift time would be looked at as “Not Scheduled.” On the other hand, if 3rd shift was used for Preventative Maintenance, then this would be viewed as scheduled time.

• Determine Theoretical Rate - You will need to know the real Theoretical Rate of the Line based on the equipment specifications for each SKU run during the OEE Calculation Time Period. This is the rate the equipment was to provide when purchased, not the rate at which operators may currently be running the equipment. This can be in any units (cases/hr, units/min, feet/min, lbs/hr, etc.).

• Determine Good Units Produced - For each SKU, you will need to know the quantity of Good Product Produced in units equivalent to your Theoretical Rate units (i.e. be consistent with your units, cases, individual units, pallets, etc.).

If all the SKU’s run during the OEE Calculation Time Period have the same Theoretical Rate, then you have all the information you need to complete the calculation!

 

Current Average Reported Line Efficiency	88%
OEE Calculation Time Period	4,320 minutes (3 days)
Time not Scheduled	1,230 minutes
2 shifts per day, over 3 days - the line should be unused a total of 24 hours (1440 minutes), but reviewing time cards, with overtime, the actual time not scheduled is 20.5 hours
Planned Production Time	3,090 minutes
Planned Production Time = OEE Calculation Period - Time Not Scheduled = (4,320-1,230)
Theoretical Rate for SKU #1	500 cans per minute
single run for the entire 3 day period
Good Units Produced	994,980 cans

1st pass yield during Planned Production Time - actual data from production reports, it is not calculated. If the production reports are in cases, then you must convert to cans to match the units of the Theoretical Rate.

 

In a slightly more complex case, let’s assume that 2 different products (SKUs) were run during the three days, each with a different theoretical rate. In this case, we need to know the Planned Production Time, the Theoretical Rate, and the Good Product Produced for each SKU. The OEE calculation is then weighted based on the planned production time of each product, for example:

 

 

5. Leveraging OEE Improvements to Generate Business Gains - Theoretical

 

Let’s extend the example of our Canned Food Plant. Let’s say that the plant had an initial OEE of 64.4% prior to implementing an automated OEE and Downtime measurement system and the associated improvement initiatives. Relatively minor improvement has significant impact:

• Availability improves from 79% to 83% by reducing setup times (small investments in tooling, establishing best practices, and training).

• Performance improves from 82% to 86% by identifying and resolving the most (5) most serious causes of short stops.

• Quality improves from 99.0 to 99.2% by identifying the number one cause of rejects and resolving it.

Compounded, these small improvements have the net effect of increasing the OEE from 64.4% to 70.8%! This improvement can have a major impact on units produced:

• Running at 500 cans per minute per line on 3 shifts we see a 10% increase in throughput

 

The results can have a staggering impact on sales and operational costs as well:

• Additional Sales of $6,000,000 / year - At $ 0.50/can (wholesale) the improvements produce an additional $ 23,040 day/line or $6,000,000 / year in additional sales

• Elimination of 78 Shifts / year - A typical 5 day / 24 hour operation runs 780 shifts per year. This OEE gain effectively generates an additional 78 shifts / year.

6. Leveraging OEE Improvements to Generate Business Gains – Real Life

 

A Multi-national Food Company with 65 manufacturing sites in North America deployed a simple OEE / Performance Monitoring system in one of their key facilities that was suffering from persistent mediocre performance. This facility would be used as the Pilot site for OEE Monitoring and as a test site to confirm OEE’s significance within the Corporate TPM (Total Productive Maintenance) program. While there was a high degree of confidence that the pilot would be successful, few knew just how much of an impact the system would have on overall performance.

 

While the system was deployed to monitor 4 production lines, the benchmarking was done on only one of those lines in order to produce conservative results. The values detailed below are actually low in comparison to facility-wide results.

 

• Availability improved from 58% to 74%.

• Performance improved from 52% to 53% - interesting that it was not necessary to run the line any faster in order to achieve dramatic results.

• Quality improves from 98.0 to 100%.

According to the Vice-President of Supply Chain, while they had a handle on the major downtime issues (15 minutes and over), “…we had no clue as to the impact of the minor downtime events, the minutes and seconds…” This system is now the backbone of their TPM initiatives.

 

7. The Bottom Line

 

Avoid the mistakes companies make when deploying a Performance Metrics Program. While they may seem small, the issues multiply, and can actually drive performance in the wrong direction or kill off a valid project before it even starts:

• Don’t underestimate the gains to be had by properly deploying OEE or Performance Monitoring. Most companies expect a 2 to 3 year payback (30% to 50%) when typical payback is in the range of 1 to 3 months (400% to 1,200%). Since these numbers are hard to imagine, they are often discarded as hype. The best rule of thumb is to under commit and over deliver. Agree to a 9 to 12 month payback when you know it can be far greater.

• Don’t assume hand acquired data is accurate or sufficient. Some lines experience more than 1,000 events in the course of a week. Operators are expected to run the equipment, not tally observations. As a result, they miss quite a bit of the most important data – those repetitive, nagging occurrences that rob valuable capacity. Further, hand tabulated data communicates results in a time frame that is not consistent with allowing operators to impact their performance in a timely fashion. There is no use collecting the data if you can’t actually do anything with it other than justify disciplinary actions.

• Start small with the capability of future expansion. You don’t need to spend a great deal of funds to get a project underway. Once the gains are obvious, the system expansion will be demanded as opposed to questioned.

• Get support from senior management. Since a good deployment is less about a vendor sprinkling “magic pixie dust” and more about your facility’s team internalizing the goals and exploiting the data, culture change has a lot to do with your ultimate success. As a rule, people typically invest a lot more of themselves when asked to do so by their manager’s. Also, your company may already be considering expending significant funds for capacity expansion or facility “rationalization”. This type of project will be a fraction of those costs.

• Don’t do it yourself. You build and sell widgets, others build and sell software solutions. Unless you are planning to change your business model in the near future, stick with the widgets. The real issue is that once the initial system is deployed and the gains are evident, you will be under pressure to rollout the technology fairly rapidly. There are few if any ‘home-grown” solutions that work well when expanded. If you take the do-it-yourself path, you’ll risk all the gains you’ve made initially, as well as risk backsliding when system performance degrades. Besides, if the gains are really as significant as we’ve documented, there’s no need to take this route anyway.