As described in my last post, disruptions due to defective parts were the single largest cause for our highly variable and long manufacturing cycle-time.
I discovered the true impact of these defects early in my manufacturing days after spending two days on the manufacturing floor trying to resolve a problem with a machine in final test. The defective part itself was very simple, a quartz window, installed in the wall of a vacuum chamber, and sealed using an o-ring. There was a high degree of variability in part sizes (all within specification as we later discovered), and some quartz windows would seal while others would leak, allowing air to enter the chamber due to inadequate o-ring compression between the quartz window and the vacuum chamber.
Installing the window was simple and quick. However, pumping the chamber down to the required level of vacuum took almost 6 hours. It took another 15 to 30 minutes to set up and test for a leak. If we found a leak, then we would have had to vent the chamber and bring it back to atmospheric pressure, remove the window, find and install a replacement, and repeat the entire pump down and test process. After losing 48 man-hours of time, we nearly missed the ship date for the product.
Then, in a casual conversation with one of the test technicians, I discovered that this was a routine problem. Apparently, everyone on the floor knew about these defective windows. The standard practice was to pull another from stock if the first one failed. Bad units were simply discarded. We had scrapped nearly forty windows to date! Yet, no one had informed the design engineer or the supplier that there was a problem with the part. To make things worse, we had no data from units that had failed previously. At some point in the past, someone had created a rule that if a part cost less than $100, it should be scrapped without processing a quality report. Each window cost only $60 and all failed windows had therefore been discarded.
Resolving just this one problem saved us an average of 16 man-hours per machine. This single part was a revelation for me. How many more such issues did my team know about? Why had these issues not been documented?
It turned out that there were many more such issues. Over the next six months, we resolved over a thousand such issues.
How did we uncover all of these issues? By simply asking the technicians! In a daily meeting on the manufacturing floor, I would would ask each employee to provide me with a list of their top issues. Everyone had to give me at least one issue without worrying about whether it was trivial or complex. Then we would classify the issue as either Safety, Quality or Productivity related. Issues were then ranked as High, Medium and Low based on frequency and severity of disruption, and were resolved in the order of priority, with safety always going to the top of the list.
By resolving the quality problems with the highest impact, we were able to make a 40% improvement in cycle-time. Reducing our cycle-times freed up capacity in terms of human resources, factory space, assembly fixtures and test bays. As a result, we were able to consolidate two factories into one. By reducing the number of defects discovered during the manufacturing process, we had reduced cycle-time, improved productivity, freed-up capacity, and reduced our asset base.