Incident Energy at Load Side of VFD

KlintMann

New Member
Jan 24, 2019
2
1
1
31
#1
Does anyone have any suggested best practices for figuring incident energy on the load side of a VFD? For example, we have a rolling mill where one VFD in a motor room will make a ~65' - ~100' run to a panel with local disconnects for each of the four or five motors the single VFD drives. I'm trying to figure out what the most reasonable method for calculating incident energy would be.

I could assume the VFD is not there and model the system with just the breaker feeding the VFD, or maybe assume the maximum through fault current of the VFD would be some multiple greater than starting current say 10x?
 
#2
In most sites where I have performed arc flash studies, drives were set to default. This means that regenerative functions were disabled. This makes a difference to the model. If you do not have the regen function, then you have options, otherwise model the drive.

There are several papers that show semiconductors will switch to the non conductive phase almost instantaneously (<5ms or so). There is no real case to model these, however, the standard IEEE 1584:2018 requires that these be modeled. You can force the SCCT on the load side of the VFD to a negligible value.

Regarding the option of not modeling the VFD and modeling the motor - i would strongly recommend against that. The motors will not only provide a false fault contribution but will also negatively influence the bus X/R ratio. This will lead to erroneous arc flash results and failed equipment evaluation.
 

KlintMann

New Member
Jan 24, 2019
2
1
1
31
#3
In my case both modeling the drive (provides a extremely low arcing fault current) and considering the SCCT to be negligible both lead to very small incident energies well under 1.2 cal/cm^2. And both of those cases match my anecdotal experience of modern drives being very fast and very sensitive.

I appreciate the sanity check!
 
Likes: Zarheer Jooma