Some of our readers have expressed an interest in the application of Statistical Analysis Tools for OEE. We have also reviewed various texts and articles that have expressed opposing views on the application of Statistical Tools with OEE data.
Our simple answer is this:
- At best, statistical tools should only be used on unique or individual processes. Comparisons may be made (with caution) between “like” processes, however, even these types of comparisons require a thorough, in-depth, understanding of product mix, customer demand, and potentially unique process considerations.
- At worst, it may be worthwhile to use statistical analysis tools for the individual OEE factors: Availability, Performance, and Quality.
Negative trends may have Positive returns
While certain top level improvements can be implemented across the company, such as quick die change or other tool change strategies, their very integration may (and should) result in changes to the existing or current operating strategy. A quick tool change strategy may provide for substantially reduced set up times and, as a result, operations may schedule shorter and more frequent runs. In turn, the net change to OEE may be negligible or even lower than before the new change strategy was implemented.
A drop or decline in OEE does not necessarily translate to negative financial performance. From a cash flow perspective, the savings may be realized through reduced raw material purchases, reduced inventories, and subsequently lower carrying costs that may more than offset any potential decrease in OEE. As we have discussed in previous posts, OEE should not be regarded as a stand alone metric. It is important to understand the financial impact of each of the OEE factors to your bottom line. We’re in business to make money and Cash is King.
Scope of Analysis – Keep it Simple
As the scope of the OEE analysis increases from shift, to daily, to weekly, to monthly summaries, the variables that affect the end result are compounded accordingly. Extending statistical techniques to OEE data across multiple departments or even company wide introduces even more sources of variation that make statistical modeling unrealistic.
While the application of statistics may sound appealing and “neat”, it is even more important to be able to understand the underlying factors that affect or influence the final result in order to implement effective countermeasures or action plans to make improvements or simply to eliminate the source of concern.
Regardless of the final OEE index reported, someone will ask the question, “So, what happened to our OEE?” Whether an increase or decrease, both will require an answer to explain the variance. If there was a significant increase, someone will want to know why and where. This improvement, of course, will have to be replicated on other like machines or processes. If there was a significant decrease, someone will want to know why and where so the cause of poor or reduced performance can be identified and corrected.
Ironically, no matter what the result, you will have to be prepared to supply individual process OEE data. So why not just review the primitive data and respond to the results in real time? While we recommend this practice, there may be certain trends relative to the individual factors that can be statistically evaluated in the broader sense (Quality – PPM, Labour – Efficiencies).
Statistics on a larger scale
We would suggest and recommend using statistical techniques on the individual Availability, Performance, and Quality factors of OEE. For example, many companies track labour efficiencies relative to performance while others measure defects per million pieces relative to Quality.
While most people readily associate statistics with quality processes, many operations managers are applying statistical analysis techniques to a variety of metrics such as run time performance, performance to schedule, downtime, and setup times. Maintenance managers are also analysing equipment availability for Mean Time to Repair, Meantime Time Between Failures, and equipment life cycle performance criteria.
As one example, we have conducted and encouraged our clients to consider statistical analysis of production data. Tracking the standard deviation of daily production over time can reveal some very interesting trends. These results will also correlate with the OEE factors. Where the standard deviation is low, the increase in production is reflected in other metrics as well including financial performance.
A final note
Lastly, OEE is a tool that should be used to drive improvements. As such, the goal or target is forever changing whether in small or large increments. Another SPC solution for OEE that may be a little easier to understand and execute is as follows:
- System – Define and establish an effective system for collecting, analyzing, and reporting OEE data – preferably in real time at the source.
- Process – Understand and establish where and how OEE data will actually be collected in your processes and how it will be used to make improvements.
- Control– Establish effective methods to control both systems and processes to assure OEE is and remains a truly integrated metric for your operations.
We strongly recommend and support “at the source” thinking strategy. Quite simply, we prefer points of control that are as close to their source as possible whether it be data, measurement, or product related. Quality “at the source” (at the machine in real time) is much easier to manage than final inspection on the dock (hours or even days later). Similarly, OEE managed in real time, at the process or machine, will serve the people and the company with greater control.
We welcome your feedback. Please leave a comment or send us an e-mail (firstname.lastname@example.org) with your questions, suggestions, or comments.
Until Next Time – STAY lean!
We respect your privacy – we will not distribute, sell, or otherwise provide your contact or other personal information to any outside or third party vendors.