Shigeo Shingo wrote “The parts will tell you what is wrong if you just ask.” You just have to know what questions to ask. This is remarkably simple in its wisdom, which is the crux of his argument. Don’t make your control system more complicated than it needs to be, and use the part as an integral tool of the measurement process.
In lean manufacturing environments, simple and visual are the action words. Simple yes/no or on/off communications that leave no room for misinterpretation. This eliminates the possibility of inducing errors from indirect communications like measuring or writing down data. Visual as pictorial vs language. A picture of right and wrong, limiting inventory by allowing a limited (calculated based on Little’s Law) amount of space. Or, a simple yellow line on the floor to differentiate aisles from work areas.
The same philosophy applies to maintenance. In today’s complex manufacturing environment, oil is not all the same. OEM’s call out specific lubrication for specific operating conditions and you can shorten the life of a machine if you use the wrong material just as easily as if you have too little or too much.
If the lubrication tech has to gauge the oil in a machine, it is much more effective to have a sight glass calibrated with zones, green for good, and yellow for add one quart and red for add more, rather than have lines marked with the words “Add one”. The communication is clear for everyone not just the tech. It also provides direction for the tech on what action to take and when. If yellow means “add one quart this week and red means add 2 quarts now, the tech can prioritize his time. This is important because it reduces two causes for error. The tech is not tempted to put in the wrong oil if he does not have the right one on his cart, nor is he pressed to make a special trip back to the lube crib to get the right oil. He can schedule the right oil for his next tour.
This also applies in the tool room. While working with an automotive assembly facility maintenance group, my team was asked to look at a small tools repair operation. It seems that these small pneumatic tools used to tighten wheel lugs, drive bolts and thread bosses were failing too frequently and causing rework or production delays. Stations had backups ready but there never seemed to be the right repaired spares and there were emergency orders every day. This was costing several millions of dollars a year in tool cost and lost production.
We looked at the tool repair crib and observed several things:
1. Three technicians were assigned to the crib but only one was there. It was explained to us that these technicians were used as reserves to fill in when there was absenteeism or emergencies elsewhere.
2. The broken tools were piled up in the center of the room in 55 gallon drums with no labelling or organization. "The guys know what tools are what from experience."
3. There was no daily plan as to what or how many tools needed to be repaired. "The guys know which tools are used most frequently on the floor."
It was clear that the organization had no clear indicators on tool demand, labor requirements or performance in the tool crib. Everyone was trying to do a good job but there was no definition of a good job.
We asked two questions, how many cycles are there in the tool life and, does the tool care what station it is sent to (the Zen question). For this discussion, we will focus on one tool category, lug wrenches. Each wrench has a tool life of 250,000 cycles. There are ten cycles per vehicle (5 lugs per wheel, 2 wheels per vehicle). There are 2 wrenches required per vehicle and we also use the same wrenches at 2 other locations on the line. We make 1500 vehicles per day so our demand per wrench is 16.6 days (250,000/10 per vehicle/1500 vehicles). Since we need 4 wrenches that means we need a replacement wrench every 4 days. It takes 2 hours to rebuild a wrench so one tech should easily keep up. Even if you replace 2 wrenches at a time on the wheel station, that brings demand to one every 3 days.
The scheduling problem comes in when you add the mix of different tools and applications. There were over thirty different tools that had to be scheduled. While tools within a category did not care where they went, they did have their own failure characteristics. Just because the OEM said a tool would last 250,000 cycles doesn’t mean every tool will last that long or may last longer.
This could have been a major system scheduling headache. We chose to adopt the lean concept of a supermarket to address the problem. Demand calculations were developed for each category like the wrench example above. Minimum inventory of repaired tools was determined for each category. We used 3 for the wrenches; 2 to cover the 3 day requirement and the possibility of 2 at a time and one as a safety stock to account for variation in actual tool life.
The team then built a set of shelves with cubby holes, two cubbies assigned to each tool category. Cubby number one was assigned to broken tool storage. Cubby number two was for repaired tools. We limited the number of tools that could be put into the repaired cubby using wine racks, with only 3 available slots for the wrenches. Slot one was labeled green, slot two was yellow and slot three was red.
Before the shift started, the supervisor would look at the wall of cubbies. You could see all of them from a single observation spot. If a cubby was condition green (all slots full), it was skipped over. If yellow, that tool and quantity was scheduled for the day. If red, that tool was prioritized at the top of the list. Using a standard wall chart showing all the cubbies, the supervisor would set the schedule for the day adding up the demand of 2 hours per tool. She now knew how many man hours to schedule. The techs checked off lines as they completed tools. With a little experience and refinement, the wall chart was eliminated in favor of using the cubbies directly and the techs scheduled their own work, eliminating the task from the supervisor’s schedule. Emergency tool orders were virtually eliminated and the savings was over $1 million in tool costs plus the added throughput to the line.
For more information please contact us through email at email@example.com.