Then i asked the million dollar question of how accurate they are.
A lot more accurate than they are required to be. However, they can fail at times.
Any suggestions on how to do it on my own ?
By installing a sub-meter, of course. The question would be: what kind of sub-metering equipment do you have? A PQ meter would be nice.
At the minimum, you will need some means to measure the voltage, current, phase angle, and a stopwatch. A large, steady load helps a whole lot.
You can get an approximation by using the average of several measurements taken at different times. For each measurement, count the number of dial rotations (simulated by the display of an electronic meter), and the elapsed time in seconds. You usually want a time of at least 30 seconds or more to cut down on stopwatch errors. If a full turn of the dial takes forever, you may want to use a fraction of a turn that equals the amount of patience you have as it may take many minutes if the load is small. You will need to measure each phase. Take the voltage, current, and pf readings at the same time you are counting the dials turns.
The meter secondary constant is labeled on the front of the meter (the Kh constant). The primary constant multiplies the secondary Kh by the CT and PT ratio to get the primary constant.
Then compare these two:
meter kW = 3.6 x Pri Kh x # revolutions / seconds
validation kW = sum(V1 x A1 x pf1 + V2 x A2 x pf2 + V3 x A3 x pf3)
As you can see, it would be much easier with a single PQ meter instead of a voltmeter+ammeter+phase angle meter.