I am interested in starting in the world of Online Monitoring.
What dou you recommend? What brand or system do they match?
I've heard something from IFM, Rozh, SPM, Emersson, Comttest, Twave

a greeting
Original Post

From experience, first decide what you are trying to monitor and why. There is a lot of snake oil out there like monitors that claim to basically automatically diagnose faults, and it depends on what you are after. Take for instance RTD's. They are hands down the best protection money can buy for a motor for temperature protection, BUT the reliability of the RTD's no matter how well they are mounted is much less than the motor they are mounted on. Practically all good monitors use majority voting logic for instance to tolerate failed sensors. And the "failed sensor" detector stuff on these doesn't work that great, either.

With vibration analysis the automatic stuff works about 75% of the time. Ask any vibration technician about the built in algorithmic stuff in his meter and how well it works at diagnosing problems vs. looking at the spectrums by hand. Online monitors used to collect data are like that (they are not very good), no matter the brand or name. Which brings me to the next point. When it comes to vibration monitors in particular make sure that the online analyzer uses a file format that is compatible with your analysis software. That's pretty much a must. Second make sure that the probes can be connected to the standard hand held vibration analyzers you use so that you can troubleshoot the online vibration stuff. Third, make sure it actually uses an FFT analysis. A lot of these things quite literally just take the raw output of a vibration probe (acceleration or velocity) and simply trigger an alarm when it gets above a threshold. Not only is this not very good but if you want that, why are you spending thousands on it? You can buy a probe from a well known and very popular manufacturer (CTC) that outputs 4-20 mA and connect it directly to a PLC or even a simple alarm monitor device and do the same thing for a fraction of the cost.

Next, you've got all those cables run everywhere all over the equipment. I don't know exactly how to explain this but your mechanics will NOT appreciate the fragile nature or the importance of putting all that stuff back. Furthermore because it's outside their normal realm of understanding and dealings, the controls/electrical technicians aren't really going to care or understand a lot about it either. So maintenance of this stuff long term is a huge hassle. It's one thing when a predictive maintenance tech has a tool that fails and gets it repaired as soon as possible. It's quite another for a bunch of installed equipment that is just supposed to work automatically all the time without attention and without your usual crews of technicians that are responsible for maintaining everything else having a lot of involvement. Trust me when a machine does NOT go down because the vibration probe failed, it is just not going to get the work order priority that a pump bearing worn out that shuts a machine down will get.

Next, you've got the same age old problem with predictive maintenance. We do a lot of contract maintenance. I'd say 80% of our predictive maintenance reports NEVER get actually used. By used I mean the report clearly states where the problem is at and what needs to be repaired or replaced, but the customer doesn't bother replacing it until it actually fails. You don't need predictive maintenance if you're just going to run to failure anyways, since those same customers don't even go as far as ordering the parts ahead of time. From our point of view we're always pleasantly surprised that these same customers keep spending the money but mostly they do it because it's some sort of corporate or insurance mandate, not that they actually do it to reduce costs and downtime. And these are not small companies either. Most of them can measure downtime in automotive terms (tens of thousands per hour) and in fact many of them actually are automotive related.

Finally, these things also have all kinds of networking and proprietary crap that comes along with them. It's usually Modbus (4-20 mA) or some flavor of Ethernet communication. Along with that comes a lot of cables and network stuff to maintain. Don't forget the total cost here.

Where online monitoring comes into its own is one of three reasons. First is that the P-F curve is so short that basically you can only catch it by online monitoring (and in that case probably too fast to save any money anyways). In fact centrifuges and turbine generators specifically have the wimpy vibration protection I was describing earlier specifically because it's a huge safety issue to have one of these machines fly apart. Typical rebuild costs for turbine generators are in the range of $10-$100 MM, so spending tens of thousands on online predictive monitoring for them is cheap. Second reason is manpower, especially if the equipment is in a remote location. Think for instance of the cost of sending a vibration technician in a helicopter to a pumping station on the Alaska pipeline as an extreme example but even with some plants the shear number of data points and cost of technicians means that it may make financial sense to put in online equipment even if it's just data gathering for the same vibration technician. This is a pure financial argument where it is pretty easy to calculate the ROI on doing this kind of installation. Third reason is purely technical. In this case I'm thinking of for instance online monitoring of partial discharges. Partial discharge in electrical systems basically comes and goes almost randomly and when it is happening it happens the event is in milliseconds. Detecting partial discharge in it's earliest stages essentially must be done online. But again...there's a pretty long P-F curve so it's still not clear today whether online measurement is actually necessary. I've successfully tracked down partial discharge events prior to failure with hand held meters.

The last consideration is somewhat of a hybrid. Is there an advantage to mounting the sensors themselves ahead of time? In many cases, yes. With PdMA for instance on medium voltage motors it is difficult and time consuming (and must be offline) to set up online monitoring in many cases. Preinstalling the sensors makes it a 30 second effort to collect data. Similarly it's not even safe to monitor partial discharge on cabling inside switchgear without shutting it down. Preinstalling the high frequency CT makes it a plug-in effort to get data. In terms of vibration probes at least machining a flat face never mind mounting the probe drastically improves results and reduces the time needed to collect data. In many of our customers plants with 100-200 data points and climbing up and down equipment it takes almost all day just to collect readings so the little time savers like that quickly add up and cut the customer's costs way down.

Add Reply