Buddy-check
Hi all,

I have a quick question about using "buddy-checks" to verify equipment maintenance procedures and return to production readiness. We use a buddy check system, in which an experienced technician verifies the work as the work is being completed. This process seems flawed to me. I was wondering if others had better procedures. In particular, if the person who performed the work called out items from a checklist and the experienced tech verified completion (if it can be seen) there would be some structure. Is this called challenge and response as opposed to do and verify (by oneself)?

Tim
5 Replies

Hi Tim,

Interesting question. In Japan when they are training a new worker they use a technique called “point-and-say” where the trainee first points and says what is the work item they will execute, then they point and says as they do the work. These activities are monitored by a “water beetle” who works with the trainee until they demonstrate a mastery of the task. I hope this gives a helpful point of view for you!

Best regards,

Greg

Norm Howe
33 Posts
Any kind of double checking or verification will, at best provide 80% assurance of catching errors. Plus, it costs money. So, in my opinion they're ok if you have nothing better in place. Best is to get creative on ways to constantly drive possibilities for error out of the system (Poka Yoke, automation, etc.). Only after you've done everything you can to re-engineer the process to prevent errors from occurring do you use verification.

Hi Greg,

Thanks for the response. That's an interesting twist on what I've been thinking. It bears further thought.

Tim

Hi Norm,

We completely agree. I try to avoid traditional buddy checks as they are costly in terms of the extra work, but also there is wasted time waiting for a buddy. It can also be perceived as disrespectful.

I'd like to find a method that minimizes the above while we're working on automation and poke-yoke. I think we can deal with the disrespect issues by communicating well and listening.

Tim

Dear Norm,

I question the often-quoted 80% number for human checks. If each person involves catches 95% of all failures then a two-person system will fail to catch 0.25% (5% X 5%) and the system will then be 97.5% effective. This is a reliability principle that drives redundancy in design. But, it infers independent, random errors. When the instances of failure are constrained, and the rational sub-groups of the types of failure are known and supplemented by using digital sensors, then the improvement can be made to be more effective. I think that the 80% number came from a different time with a different design and production paradigm than is available today.

Best regards,

Greg