2.2 Data Collection Review

Data collection and analysis for OEE is sometimes thought of as good in theory but not in practice. The arguments against it use excuses such as "We have too many different products" and "Our process is changed for different style outputs." In these situations, the best approach is to step back and review the boundaries of the system. Start where materials are input into a systematic flow with an expected product or subassembly for the next factory step. This transformation step is often linked with others in a series of steps that have few if any fixed buffers. The process has an expected flow or cycle time.

OEE is appropriately applied to bottlenecks, critical process areas, and high expense areas. An appropriate test is to ask, "If the effec tiveness of this transformation step is improved, will the bottom line be significantly impacted?’ If the answer is yes, then putting effort into generating true OEE and driving improvement is worthwhile.

As an example, I once observed a work center that successfully used OEE on the shop floor as follows. The company was highly automated; it used shop floor computers to gather much of its information. Its Equipment Performance System (EPS) collected not only the various downtime causes and frequencies, but also run time and speed monitoring. From this database, the company could easily compute OEE for each product.

Essentially, the company picked a standard process that represented its most common product. This product-process format was used as the benchmark for OEE. Because the format was used so routinely, significant production history was available. Furthermore, the product was manufactured on all of the work center’s different equipment flowlines. Next, the work center defined how other formats and sizes with the same product should compare with the benchmark process. This comparison generated an OEE coefficient. The comparison was repeated for different product families and formats as well as for different process setups. The information gathered was valuable when communicating with superintendents and plant managers about capability questions and the impacts of different product mixes. It also provided the yardstick for shop floor crews to use when examining their real time productivity on shifts.

This plant had the advantage of having automatic data monitoring and information feedback for nearly all the products it produced. However, at the very minimum, plants can simply gather the information for each product run, usually manually from cycle counters, run hour clocks and other measuring devices. Simple chart recorders can be extremely valuable because the frequencies and duration of events can be easily captured and analyzed.

Figure 2-1 provides a form that lists the minimum information that should be gathered.

This information collected for each product run can quickly form the database to begin examining OEE and to start driving productivity improvements. For example, comparing start/stop time vs. run time measures efficiency, start/stop cycle time vs. run time measures speed information, and units vs. transferred output measures quality. Comparing input materials vs. units produced captures waste and inventory information. Comments from the crew leader help cross-functional teams work

on root cause elimination of limiting problems. One goal is to understand the actual functions that have failed, as well as th e actual equipment and technical problems. Another goal is to reconcile the actual output with the computed OEE, confirming that true OEE is being captured.

A decision must be made about how to handle re-work. In many processes, manufactured items cannot be transferred or shipped with out being re-worked first. (In such cases, the first effort of bottleneck has failed. OEE for that manufacturing time is zero.) Re-work efforts can fall into the following three categories.

1. The re-work can be completed off-line using non-critical equipment. It may even simply involve re-packaging and can be completed manually. In either case, the rework does not impact the bottleneck system. It becomes a manufacturing cost decision. OEE of the bottleneck does not change. However, the measure for factory units produced should note how much re-work was finally transferred so that reconciliation between OEE and first pass yield can be determined.

2. The re-work can be completed online at a time when the equipment was not originally scheduled for production, perhaps on weekends or overtime. As with the first category, this work essentially is completed with off-line equipment and, again, it becomes a manufacturing cost decision. As before, the first pass yield number needs to be identified. This type of action should be identified when examining the TEEP metric; it involves activity on a key asset that would have conflicted with regular production had it been scheduled.

3. The re-work must be completed online, competing with regular production time. In this case, the re-work material should be looked at as new input material. The time, speed, and quality factors should compute into the current OEE. A note needs to be made so that the incoming inventory is adjusted appropriately now that waste has been turned into good units.

Consider the following example:

Assume that 100 percent OEE (running at ideal speed with no downtime and no quality losses) for a production area is 100 units per hour. Normal production has been running at 75 percent OEE (75 units per hour).

During week 1, the work area ran production for 168 hrs and produced at a normal rate for 160 of those hours. However, during 8 of those hours, the product was placed in the wrong colored boxes, creating 800 units of re-work. In sum, for 8 hrs, OEE is zero and for the remaining 160 hrs OEE is 75 percent. The week's report would indicated an OEE for the area of 71.4 percent, calculated as follows:

During Week 2, a holiday week, the area worked 144 hours including the re-work. The equipment ran normally. However, the 800 rework units had to be manually fed into the system. The time for this rework took 12 hrs, resulting in only 780 good units. Because 780 units in 12 hours averages 65 units per hour, the equivalent OEE is 65 percent for those 12 hours. The rest of the production for the remaining 132 hrs was at a normal OEE rate of 75 percent. The week's report would indicate be 132 hrs at 75 percent and 12 hrs at 65 percent, yielding an OEE of 74.2 percent.

The overall OEE for the two week period is 160 +132, or 294 hrs, at 0.75 percent, 8 hrs at 0 percent and 12 hrs at 65 percent. This yields a combined OEE of 72.7 percent.

In general, good data collection is a key requirement for successful OEE strategy. The success of any factory is greatly affected by how effectively accurate information is collected and analyzed.

2.3 Practice Production Report

The spreadsheet in figure 2-2 follows and provides a sample 40-hour production report. It includes many different types of interruptions that illustrate the different OEE categories. Assume this area has a normal waste rate of 3.5 percent and that it produces finished units at the rate of 4 per minute (ideal or theoretical rate). Each column of the spreadsheet represents 10 minutes of calendar time.

Each event is identified with a letter and a brief description. The height of each shaded area represents the rate at which units are being produced, with each row representing 2 units per minute. Thus, an area 2 rows high has an expected rate of 4 units per minute. By summing the shaded areas of production, the number of units produced can be determined (see section 2.5). The units produced for the experiment represented by the block following letter W are excluded from this number.

The analysis that follows computes OEE and TEEP for the specific 40-hour time period in the spreadsheet. Do not confuse the production report for a weekly report. (If the 40 hours did represent the planned production schedule for a week, OEE would remain the same, but TEEP would be computed on the basis of 168 hours.)