Who Done It? A Variation Investigation
Process variation is running rampant in your factory and you've been called-on to solve this mystery. You are passionate about finding these variation culprits because know that controlled processes yield fewer defects and require less quality control effort.
So how do you find the variation culprits in your factory? Well you can find some good clues by studying your process data and investigating the 6M’s of Manufacturing. Let’s take a look at our suspects one by one.
Suspect 1: Measurement
You want to know how much of the total variation comes from the measurement system. When spec tolerances are tight, like in precision machine shops, even well-trained inspectors using precision gages can be responsible for a high degree of measurement error. Solve this mystery by doing a Gage R&R study. It’s like doing a DNA test at a crime lab. A Gage R&R study will determine if your measurement system is a variation culprit or not.
Suspect 2: Manpower 
Operators are often guilty of crimes against the process. They can be responsible for process over adjustments, under adjustments, needless adjustments, human error, inconsistent behaviors, and poor machine set-ups. All these things can create shifts and drifts in a process. Solve this mystery by looking at the process data over the course of a week or month. A time-series graph is like a fingerprint analysis. Often you can identify operator induced process shifts by looking at data trends.
Suspect 3: Material
Raw material lot changes can dramatically affect process variation. In food processing for example, raw ingredients often come directly or indirectly from farms. The source location and growing conditions can have a large impact on process variation. We need to look no further than potato chip processing for an example of this. When a new crop of potatoes hits the fryer, the process quickly changes. Here again, a time-series chart can provide you with some variation clues. Just inspect the data trends for a shifting process during raw material lot changes. Stable processes require consistent raw materials.
Suspect 4: Machine
Your production machines are always viable suspects. If equipment is poorly designed or poorly maintained, it should raise your suspicions. To investigate machine suspects, complete process capability studies. Whenever a study yields a Cp result of less than 1.33, then you should consider machine design upgrades and implementing Total Productive Maintenance (TPM). These activities will put your machine-caused variation on house-arrest.
Suspect 5: Mother Nature
We all know that Mother Nature can occasionally be a major disruptive force, but when it comes to a manufacturing process, her affect can be very subtle. Often temperature and humidity changes can impact a process and cause variation. If she is a suspect in your investigation, collect production room temperature and humidity data and compare it to your process data using a correlation analysis. If there is a statistical correlation, then you will want to install environmental controls.
Suspect 6: Methods
Methods always conspire with other suspects in the crime of variation. But they can be "turned" to aid the investigator. Methods are a common thread connecting all of the variation suspects together. The lack of good training programs, test methods, material specs, maintenance procedures, work instructions and quality systems can be the cause of variation but the reverse is also true. Poor methods are guilty conspirators but good methods can help to imprison variation.
The Police Line-up: After your initial investigation, you will have a short list of probably suspects. You may even have a variation conspiracy on your hands. But you need more conclusive evidence before an arrest can be made. The final damning evidence can be found via a multivariant analysis. Here you include your top suspects in a controlled process experiment. This type of DOE* will clearly show you who is guilty of process variation and whether they acted alone or not. A multivariant analysis is like conducting a police line-up. It will point you directly to the variation culprits.
*Design of Experiment
Stay connected with news and updates!
If you want some weekly T4T wisdom coming straight to your inbox for your reading pleasure - look no further! Join our mailing list to receive the latest blogs and updates.
Don't worry, your information will not be shared.
We hate SPAM. We will never sell your information, for any reason.
