Now I'll give you the method I use
I'm sure there are many others
determine normal and peak operating current from the load
say, 100 and 150 on a 200 A ckt
I use 2x peak, in this case 300
as long as it covers the max ampacity of 200 I am good
so 300:5
now you need the accuracy class
it can be selected 2 ways
based on va burden or loop R
both are basically the same
I'll use R (burden add up loop va make sure < ct rating)
assume leads are 0.1 Ohm
relay is 0.8 Ohm
total R = 0.9 Ohm
2 accuracy rating methods
percent, 0.5, 1, 3% (at 5 A out)
and Class xxx
I'll use class, seems to becoming more common
at xxx volts at 20 x 5 A accuracy is 10% (there is also a 5% class) , ie, 10% at 20 x 300 = 6000 A fault
at 300 A very accurate, at 1000% (10 x fault) < 10%, etc
loop R 0.9 and i = 20 x 5 = 100 A
so V = 90
a class 100 can be used
if greater accuracy is required a C200 or a C100 5% rated can be used
the greater the difference between fault V and class xxx the better the accuracy
if long leads and say R = 0.3 and relay is 1.1 Ohm
V = 140
class 100 no go, need C200
This explains the 2 accuracy rating methods
https://cdn.selinc.com/assets/Liter...ccuracy_SZ_20060606_Web.pdf?v=20160502-161100
this gives good headroom
rated 10% at 100 V at 20 x 300 = 6000
but actual trip is typically closer to 1200% x 200 = 2400 A
or 40 A at the relay, V = 40 x 0.9 = 36 V << c100 rating
so accuracy much better than 10%, guessing closer to 6%