Did you ever get an answer?
For CPWG, we are using a waveguide port that is electrically shorted around the perimeter - making it like a PEC metal tube.
The bottom edge of the port is on the top edge of the ground plane (i.e. the gnd-plane below the CPW trace, making it CPWG).
The top of the port is about 20 PCB heights above the CPW trace. That way, the sim can include fields in the air way above the CPW. At 20 PCB heights, the fields at the port's metal boundary is >60 dB below the fields between the CPW and the GND plane. If you don't have a Gnd-Plane, (i.e., it is CPW, not CPWG), then I would assume the port should go 20 PCB heights below the CPW, (i.e., not just 20 heights above it.)
We got the TDR to work to look at line impedance. This required the sim to be done with "No Check" and to run for a specified time duration, which was the round trip time for a wave going through the dielectric.