Hi @pkvoon!
Regarding the mesh I would also make sure to use inflated layers to capture gradient information. Could you explain why exactly you use the conditions you are using at the moment? Do you have reference values?
Just as an information: Prescribing the total pressure at the inlet allows you to get a fully developed profile in all pipe extension, which should (theoretically) match the pressure losses better than a velocity inlet.
As mentioned above and to avoid problems you need to satisfy y+ (Y-Plus) parameters in the near-wall region to use k-epsilon model appropriately.
Best,
Jousef