feeder/transformer loading

[jcraft]

we have an outdoor substation at our plant. It has a 4160v fused cube on it that feeds a section of plant. it runs from this substation to a huge splice box inside the plant and picks up 4 - 500kva 1200a 4160v to 240v transformers. we have been adding a lot of loads on this feeder over the years and we want to make sure it is within the limits of this feeder. the best way I have heard to do this is monitor the sec. of all the transformers(at 240v sec.). The question I have is this. We have an Amprobe DM11 datalogger that will log rms current, rms peak, rms average etc. the best way I can think is do a test on all 4 transformers and then add up the currents to find how much is flowing on this feeder. I am not sure if I should be using the peak current value on the logger or if I should use the average. looking at the profile so far there are a lot of short duration peaks. In the end I want to make sure the wire size is okay and the main fuses in this outdoor sub are not going to blow with all the additions. I guess to make a long story short( yeah right) when doing these load tests do you look more at the average current flowing in a line or at the peak. I guess another example would be if one of our 1200 amp rated transformers is showing an average current of 900 amps but it is showing a lot of short ( 15 second or less)peaks around 1250amps are we still safe to say that we could load this transformer more and disregard the peaks. thanks for the help sorry for the long post

 

[DanDel]

Overloading is based on the continuous current, not peak currents which are usually caused by intermittent motor inrushes.
The best way is to monitor at the source for about one week. This should give you a good idea of the actual continuous load. This doesn't sound practical in your case, so individual monitoring at the separate transformers will have to do.
If you can only monitor one at a time, try to determine if the loads are consistent over time (not counting the inrush peaks), otherwise estimate as conservatively as possible from the data, and add the these values together.