Dear yaseen
My best clue is that the convergence is happening due to the large value of Y+ for the blade walls (The k-omega model requires a Y+ less than 1)

I couldn't use inflation to reduce Y+ because this is a hex mesh in the outer domain as shown and it gets all messed up with inflation.
And I tried making the inner domain structured too, to be able to reduce Y+ using bias of edge sizing and I still couldn't.
If you have the exact same case then you better head for ICEM CFD as it's more suitable