Talk:Predictive failure analysis

Much of the problem of "predictive" failure analysis is that prediction of an imminent failure is not always very accurate. I've love to see a good case-example for this article. Causation of failure is lacking in this article. In my defense, I offer the following example: changes in the oil viscosity and/or color in your car engine might be suggestive that preventative maintenance should be performed, it does not actually indicate when the engine might fail due to that cause. Failure is largely a random phenomenon and very hard to predict with certainty. Failure also needs be clearly defined - If the engine runs but not over 30MPH, is that a failure or degradation? The nature of reliability engineering is why I make this comment. The IBM disk-drive example cited here is insufficient to justify prediction accuracy or a maintenance schedule. --2600:6C48:7006:200:B056:6066:1296:EF0B (talk) 02:14, 11 November 2018 (UTC)