The process of adding sensors is rarely done as part of a well-planned diagnostic process. More typically, sensors are added as part of a philosophy of the end justifying the means. Too often, sensor placement is just a band-aid solution to a bad design. This is not to say adding sensors isn’t important, but the emphasis must be on optimizing sensors, in order to avoid the many pitfalls.
Sensors can reduce expected Fault Group Size
Reducing ambiguity is why sensors are usually added. However, measuring ambiguity isn’t as simple might be expected. Fault Group Size alone does not factor in a timeframe. That is, do you care if all Fault Group Sizes of 2 or more will occur 20 years in the future? You might, but then again you might not for the purpose of trying to add more sensors.
Sensors add Unreliability to a System
Fairly obvious, but also commonly overlooked, is that more sensors means more failures. A bit of a catch-22, of course, but the trick is finding the right balance. This is why sensor placement is a trade-off between reliability and maintainability (or availability).
Sensors produce False Alarms
Most false alarms come from sensors. With an increase in false alarms comes an increase in Maintenance Downtime. To what extent do added sensors add to the fault group size being reduced?
The right balance between adding more sensors and redesigning the system to take better advantage of existing sensors varies with each project. With the right approach to diagnostics from the beginning, it will be clear when it’s time to add more sensors. Ensuring that standard systems engineering practices (up-front, top-down design) prevail over mad sensor placement (after-the-fact, bottom-up design) is the best way to stay on track.