Management of historical vibration data

Hi guys,

Probably a weird -or weirdly formulated- question. Any experience around managing vibration information through Ignition and its historian? I specifically mean storing absolutely all the time-domain information so we can later do sprectral analysis and all the other fancy stuff.

The amount of information generated this way is huge (think of a large turbine or compressor; thousands of data points per second for a single tag), and downsampling it missess the main point -I think- so I’m not sure a “traditional” historian is supposed or even capable of managing it.

There seems to be a reasonable amount of information on tools to analyze this info, but very little on how to store it and manage it long term. Any information on the subject is highly appreaciated (best practices dealing with this kind of info, suitable products, etc).

Thanks and regards.

I haven’t had to do this, but I would use digital audio recording systems, the pro types that use 96kHz sampling rates. Just record all the time. Break the recordings into 10-minute files, or something like that. Use FLAC or some other lossless codec for a bit of compression. (No mp3 or other technology–those use spectral analysis themselves to throw away stuff humans can’t hear, but is important for mechanical systems.)

I reckon you could use Ignition as a front end…

but the rest of the system has to be right. Hardware wise something like NI (National Instruments), like a PXI rack. They have a back end DB system, DiaDem I think, but within your VI in the PXI you would push to SQL via their ODBC sub-VI’s. Make sure you’re sitting down when you get a quote.

Another option may be MQTT based sensors, think I’ve read about use cases similar to this but have no experience with that.

I’ve used NI for 433MHz systems.

I’m not fond of NI. Sorry. I’d buy pro audio gear that uses the AVB standard for data delivery.

Thanks guys!

Just following some keywords in your replies got to companies/systems like which seem to have an idea of the overall vibration data acquisition & managent best practices. They use high-performing InfluxDB as the historian, which I’ve already used with Igntion (Kymera developed a connector), but even there raw data is only stored by exception.

Thansk again; I have a valid trail to follow now.


I’m no expert in vibration analysis of turbines, but here is my 2c:
Relying on raw data transferred via comms protocols like OPC will always provide you a data resolution limited by the speed of your connection.
It may be worth considering a purpose built condition monitoring sensor such as Balluff BCM0002 which pre-processes the data at the hardware layer before transmitting it to SCADA, these sensors can be configured to record RMS data, but also Peak-Peak, standard deviation, min, max values that way you will not miss any spikes or other strange events and you can control the amount of data stored in SCADA based on your sample rate.