"another question, Is this high vibration caused by looseness between shaft and bearing inner race?" Unquote
Was there a problem or potential problem seen, concerning the installation of the bearings, but was it decided that "it" was no threat, during the "assembly in March 2016"?
If there was something noticed during assembly, it very well could be increasing looseness.
"during the balancing work in march, i found so many weights attached from previous balancing work, we ignored and continued with balancing(adding another weight) , and could not lower the 1x than 5mm/s ." Unquote
Too many weights are a historical indication of balance problems, IMO.
Did you test to see if the fan is running at or near a critical frequency. Balancing is rather difficult, at best, when running at its critical frequency. Many weights sometimes is an indication of this. An easy test is to monitor, in live mode with low Fmax and low resolution, the 1x amplitude of the fan as it coasts down.
But a normal stored tach light coastdown tracking of amplitude, rpm, and phase is better.
I see that maybe the motor has the highest amplitude at 1x (on the point "1" H.
What was the motor Drive End readings?
Which bearing position on the fan did you use during the balance job?
What was the axial NDE fan bearings amplitude?
Excessive looseness, as will some other "things", will hinder the balancing procedure. Were your phase readings erratic or steady? Amplitude reading? Steady or varying?
What are the sidebands off of the running speed and running speed harmonics? They appear to be less than 1/3x and more than 1/4x.
You spoke of seeing 1/3 sub harmonics. Which data set do these show in?
1/3x, with harmonics of 1/3x, are a good indication of bearing to shaft looseness IMO.
Thanks and Have a Great Day,