I also work in oil and gas. The machines we have move around a lot from project to project, offshore to land and back and forth etc.
Im monitoring the machines, no control via ignition but I do have means of remote parameter viewing and adjusting etc. I monitor pressures, temps, component run hours, filter hours, motor start stops and lots more stuff trying to build trends. Somewhat of a crude ML to ultimately replace parts before history tells us a part will fail.
Biggest challenge was keeping it all connected either via site WiFi or cell modems. So every machine is connected via MQTT back to a cloud based ignition server and a VPN network. Same VPN network also lets me remote access the machine / PLC’s, apply updates etc.
As I mainly monitor everything In hours run the messages are sent every hour, just the basic stuff. But I also programmed ignition to be able to send a MQTT message to the machine on a button push, to have the machine send a more in-depth MQTT transmission when requested. This was the biggest factor in keeping data overhead low. Perspective was a game changer as the operator can now open the app, scan the QR code on the machine, view past data or request more information if they are having any problems with the machine etc.
As with every project it starts with a simple idea and it quickly grows arms and legs.
A good and recent example of the benefits to this. I recently had a hydraulic pump motor fail after 400 hrs, only a small 5hp motor. We typically change it every 2k hours so this one caught us off guard. Closer inspection we seen that it had been stopped and started way more then usual, the operator was stopping and starting the motor every few minutes, over a span of days. Turns out they didn’t like the noise so they would stop the pump every chance they had. That’s a brief summary but benefits of capturing data that at the time seems irrelevant, is valuable later on.