Good data is the key to improving manufacturing performance and cutting costs. With good data you can maximize bottleneck utilization, measure OEE accurately, use resources more efficiently and find and reduce waste. The challenge is, how do you capture that good data? Here we’ll explore the importance of manufacturing data collection, the difference between good and bad quality data, and how to improve data quality.

It’s a cliché that bears repeating; improvement starts with measurement. Good manufacturers know this, and they make efforts to understand how resources are being used. It’s why machine monitoring and Overall Equipment Effectiveness (OEE) data collection is so widely used: only with data about what’s happening can improvements be identified and implemented.


Defining “Quality Data”

When people talk about “quality data” they generally mean it must be accurate.  Automatic data collection is touted as the solution to accurate data. It’s true, automatic collection from the shop floor equipment is critical to capturing machine events that would otherwise be ignored and not logged, such as short stops.  Examples include swapping tooling, a drop off in production at shift changeovers and delayed maintenance response to unplanned stoppages. This is very important and a step in the right direction to achieve data accuracy, however, we can still improve.

Let’s take a step back and review what the definition of accurate data really means.  It seems simple, when the machine is running then it needs to report that it’s running, and likewise, when it’s not running report that as well. How about some of those micro stops mentioned earlier such as an automatic tool change.  In this case is the machine in cycle or is it at a stop?  The machine is executing a program, but not producing. 

There are different scenarios where either choice would be the better solution. In addition, there are many other similar scenarios where the definition of a state of a machine is not viewed the same by all of the stakeholders.  Therefore, probably the most important aspect to defining quality data is for everyone to agree on the definitions of the data items being collected. These definitions can vary by machine and by process so it’s not always straightforward.  If used correctly the machine monitoring system acts as a communication tool to help get everyone on the same page as to what the data means.  This is really step one to achieving quality data. Accuracy based on the definition of the activity is not the only dimension to quality manufacturing data collection. Timeliness and real-time access to the data is critical if the data is to let management respond quickly to a dynamic manufacturing environment. Knowing that last week a bottleneck machine was down for an entire shift because there was an issue doesn’t give managers what they need to react in real-time.

However, there are many times when the machine is not running, and the machine has no idea why it’s not running, therefore it cannot be captured automatically.  The operator “chooses’ not to run the machine due to another task, a machine problem not captured by a sensor, or a perceived machine problem. What we’re seeing here is that data accuracy also refers to categorizing downtimes from an operators’ prospective, since it’s the operator that is not running the machine.  An operator data interface that can run on a PC or any tablet such as the Scytec DataXchange ODI screen is critical to capturing this level of data and achieving quality data.

Many times you’ll hear that operator-entered data isn’t ideal since it introduces bias and people interpret scenarios differently.  This is definitely true when doing number-crunching analytics, but understanding why a machine is not running from an operators prospective is just as important, if not more important.  Remember, first and foremost the data collection system is a communication tool.  Identifying that operators have different definitions of similar downtimes is an important step to get everyone on the same page. Operator input may also identify training issues as well as process issues.  If an operator is involved then operator input is critical to achieving data accuracy.   

Improving the quality of data requires exposing the data, discussing the data with the team, and acting on the data.  This process changes the culture of the environment, and as using the data collection system becomes part of the daily routine the scenarios are better understood and the data entered is more accurate.  The system feeds upon itself.

By simply collecting data and using the resulting charts and reports as a talking tool to get everyone on the same page the improvement process has already begun. Data definitions will be adjusted, and the understanding of the collected data will be across the board. This is the process of collecting quality data.  What you will find is that some problems magically disappear, and other previously unknown issue are revealed.  Congratulations, the benefits of an industry 4.0, OEE machine monitoring system are being realized.