Often, Kaizen teams for Lean Six Sigma lack the data and corresponding analysis needed to drive consensus and convince others to adopt the project recommendations. The Industrial Internet of Things (IIoT) facilitates fact-based decisions, a fundamental theme of both Six Sigma and Lean Manufacturing. IIoT provides data and analytics that could improve the effectiveness of Lean Six Sigma programs and Kaizen teams.
Six Sigma DMAIC process and impediments
Usually, Lean Six Sigma Kaizen teams start by creating a value stream map (VSM) of the process they want to improve. The VSM methodology drives consensus for the current state of the process among those who actually do the work (as opposed to a manager or support engineer) and helps them identify sources of waste. When a defect is found to be the problem, the Six Sigma DMAIC (define, measure, analyze, improve and control) process comes into play.
Kaizen teams can be thwarted, however, by impediments that block progress. Common causes of project failure include:
- Lack of a means to gather data for the measure phase.
- Using inappropriate analytics for the problem at hand like statistical analysis assuming a normal distribution curve when it isn’t normal.
- Slipping back to the old way of doing things leads to a weak control phase.
IIoT can provide a means to overcome each of these impediments.
The measure phase involves data acquisition and numerical studies for parameters around the previously defined defect. This activity involves validating the measurement system, including the instrument’s accuracy, and understanding the potential sources of variation. Obtaining this fidelity requires a lot of data points—certainly hundreds, and maybe even millions of samples.
In the early days of Six Sigma (late 1980s and 1990s), teams could easily find processes running at the two sigma quality level. Here, a hundred measurements should contain 31 defects, which was often enough to guide the team to the root cause of the defect. Moving beyond three sigma—a need for today’s teams—requires much larger data sets (tens of thousands to millions).
With the higher sigma quality levels, it becomes impractical to obtain the needed measurements manually. This is because the high costs, labor intensity and extended elapsed time required for data acquisition often scuttle a project. Also, low-quality data overwhelms the true defects. Manual data acquisition has a higher error rate than the process step under examination. Studies performed decades ago to show the benefits of barcode data entry found 10 percent of manual entries containing 80 characters had an error—and that’s back when schools taught good penmanship.
Automated data acquisition is more effective, and IIoT fills this need. Also, IIoT can go beyond process data and add often previously unavailable equipment data that can have significant impact on defect rates.
Analysis goes beyond normal distribution curve
Training programs for Six Sigma nearly always use statistical analysis that assumes a normal distribution curve. This might be acceptable for the typical two and some three sigma projects. Unfortunately, the real-world environment often has patterns that do not follow a normal distribution curve. For example, none of the equipment failure patterns used to determine an asset’s maintenance strategy follow the normal curve.
Particularly with higher sigma levels, the analysis needs to move beyond assuming a normal distribution curve. IIoT platforms provide a broad set of analytics. Some also have a means to quickly identify associations within a data set that contains many parameters (i.e., various I/O measurements over time). This becomes necessary to determine the true source of the problem rather than a parallel effect (causation vs. correlation).
The improve phase uses a plethora of methods to reduce defects—limited mainly by the team’s imagination and process knowledge. IIoT offers a means to continuously monitor the health of a process or equipment and generate an alert when it deteriorates. The most common application of IIoT involves monitoring an asset’s condition for predictive maintenance to prevent unplanned downtime.
IIoT can be used to help assure that the improvement sticks and prevent people from going back to the old way of doing things. The continuous monitoring using IIoT can be programmed to generate an alert when things start deteriorating—early enough so that preventive measures can be taken before defects occur. Alerts managed using ad hoc communications (phone call or hallway conversation) with people that can fix the problem are often lost—humans tend to forget. In the case of predictive maintenance, the recommended automated business process has the alerts sent directly to the maintenance planner who can assess, set priorities and schedule a repair in the enterprise asset management (EAM) system.
Lean Six Sigma programs have advanced beyond traditional manual data collection and simple statistical analysis assuming a normal distribution curve. IIoT offers a means to take these programs to a new level of effectiveness.