Having doubt in defining ports of a directional coupler

I am trying to design a UWB directional coupler in HFSS and facing a lot of difficulties.Some one please help me with these doubts and error messages
1,do we need to define a radiation box
2,what type of port should be used;lumped or waveguide
warnings
1,Adaptive Passes did not converge based on specified criteria.
2,ast sweep setup, process hf3d: Poor convergence in seeding data.

1. probably not. what is your topology? without radiation boundary the outer surface will be PEC, so if that has impact then ou decide whether you need radiation boundary.
2. depends on topology. waveport if the fields at the port can be assumed to have specific mode. CPWG is often better lumped over wide band.
1. may be setup or geometry related. check port setup as well.
2. Fast sweep is best for narrow sweeps, use interpolating for wideband.

I am trying to design a slot coupled directional coupler with two substrates and 3 conduction layers,middle one is the ground plane with a rectangular slot
../imgqa/eboard/EM/EM-0kg5u2ochk1.jpg

You need an air box. use vacuum material. As it stands, the design traces are shorted against the outside PEC. If there's no air box defined, HFSS invisibly assigns the outer surface as perfect conductor so your signal lines are shorting with that. Make the box big enough to limit interaction with the top cover of the box. rule of thumb is about 5-6 trace widths or substrate thicknesses above traces.

I like your model, prince. You need an air-box assigned as Radiation Boundary. In the old days I assigned vacuum instead of air. But the software and computers are so fast now I always assign air. If your microstrip is already determined close to 50-Ohms you can just make those rectangles into lumped ports. If you want wave ports you need to do like reidinstransit said and also pull your radiation box in so 4 of the edges contact your board's perimeter.