That drive has an SCR front end (rectifier) as opposed to a simpler Diode front end. That means there is a firing circuit for the SCRs. They do this to get rid of what's called a "pre-charge" resistor circuit in the drive that prevents damage to the components from the capacitor charging current inrush when you first energize the line terminals. In a pre-charge resistor circuit, there is a current limiting resistor in series, which is shorted out after a few seconds (or they use an NTC resistor). The alternative is to use SCRs instead of diodes, and ramp the voltage into the caps to prevent the inrush. But as I said, SCRs then require a gate firing circuit instead of a resistor shorting circuit, so it tended to only be done on larger drives.
What has changed in the latest generation of drives is that the drives are all made now with "IPM" devices, "Intelligent Power Modules", which are a single potted device that contains all of the transistors and all of the diodes or SCRs, along with all of the firing circuits for them in one monolithic block. So for the VFD mfr, since the firing circuit fro the SRCs is already there, all they need to do is interface with it on the mP comm bus.
What that means here however is that something is interfering with the command signal to the IPM telling the SCRs to fire, which means the DC bus is not getting voltage through the rectifier. What is causing that is the only actual mystery here. Eaton (or Vacon who actually make the drives) might be able to tell, but would need the drives back for forensic analysis.
What YOU need to do first is to see if you can duplicate this when on a different power system, to eliminate a noise or interference issue as the cause. So remove one of the drives and take it to a different building to power it up. If it still does this, the drive is defective. If not, there is something in the local power system causing the SCRs to not fire. My guess would be something on the line that is causing severe line notching. For SCRs to fire correctly, the system must determine the zero-cross reference point of the sine wave coming in. If that sine wave is very noisy, it can be filtered, but if there is severe line notching, the zero crosses of the notches can "fool" the sensor circuit of the gate driver and it shuts down to prevent misfires and catastrophic failure.
If it does work fine elsewhere, look around at any other nearby machinery for something with an old (or really cheap) DC drive. They were notorious for causing that. Some other older linear power supply that is starting to fail might also do that, even and older lighting ballast on its last legs.