Connecting the Dots… Analysis Phase in Complex Issues
What is the situation?
Sometimes we face complex challenges, and the analysis phase becomes harder to complete, thus increasing the difficulty of finding the root cause.
The output (Y) could show variation that we cannot easily correlate to any input (X’s) along a time period (which depending on the process could be hours, days, weeks, or more). For example, in a chemical process case, one issue we found lasted more than 12 months in the analysis phase, before the root cause was discovered. The reason was that it disappeared for months, in this situation, the environmental temperature was a significant contributor, because of this, we believed we had fixed the issue, until it happened again.
How to approach the situation?
The analysis will require the team to focus in the process and to develop a deep understanding of the conditions and parameters used, a TAPS (Team Analysis & Problem Solving) form is recommended to be used for recording information (it is based on 8D, A3, and FMEI forms). This could take hours of discussion and floor shop evaluations (Gemba analysis); additionally, it could need a very detailed review of sample parts or of the process sequence (which could also be applicable to a service, not only to physical products), and customer complaints investigation.
In these complex cases, after finding some possible significant parameters (key inputs) through the use of a C-E Diagram, a Process Map, and the TAPS form, we can run experiments (doing a DOE, Design Of Experiments) to define what inputs (X’s = A, B, C, D, …) are really causing the variations in the process, and by how much, and in this manner evaluate their impact on the output (Y).
Remember to take opinions from the personnel that has had direct experience with the process (operators, inspectors, supervisors, suppliers, etc.), the information and intuition they can provide will help to guide the analysis. The qualitative information obtained should be considered a complement, it will still need to be completed through controlled experiments to get hard facts and data, and there will be a need to replicate the output to confirm the variation changes (Y1, Y2, Y3, Y4, …).
It is important to be able duplicate failures, to confirm that the X’s that were found truly are the drivers, if you can repeat the condition and get the same result (good or bad) for different combinations of parameters, then you have a correct and thorough knowledge and understanding of the process.
A word of caution
Sometimes the experts could fail to agree with the findings and not accepting the results; the key is to have evidence, that through your experiment you can make the problem “turn on and off”, in this way predict the performance of the process.
Real life experience
This situation happened to us, with a wave soldering machine in which we got PCB with green paint (solder mask) detaching in large areas. The engineering expert kept pointing to the supplier of the bare board, but after running some experiments (done during the weekend, to avoid the expert and his pre-existing opinions), we found the flux application amount was the key driver.
Even though we could control the failure in every PCB every time, and remove 100% failure from the process, the expert still argued that the result of our experiment was inconclusive. The expert insisted the machine process was alright and that we needed to talk to supplier; even as he was looking at the thoroughly proven results and could confirm the problem had disappeared in that same lot of PCB that had previously failed.
The lesson learned is that solving new and complex problems takes time and a willingness to challenge pre-existing opinions. So, take care of having good detailed data while developing your DOE and while obtaining your conclusions.
The final stage is to incorporate the findings in the process documentation, which could be work instructions, PFMEA, Control Plan, specifications, etc., to establish permanency for the newly condition discovered.