Hi Karen! Thank you very much for your reply. I tried your suggestion and used the name of the deep-Nwell layer (DNW) in the "Ground layer name" field, and I finally got Calibre to attach the parasitics of the different deep-Nwell regions (i.e. power domains) to different nodes! The problem, however, is that the LVS names of those regions are actually the local supply nodes and not the grounds, since, for obvious reasons, the deep-Nwells are tied to the positive supply. So for instance, in my example, I am getting the parasitic caps tied to vdd_A and vdd_SW, instead of vss_A and vss_SW.
Following the suggestion of Vern, I took a look in my rule deck and noticed that there is a layer called "psub" which seems to correspond to the (isolated) psub regions inside each deep-Nwell. So I put that layer as the ground layer name and almost all the parasitics are now being (correctly) tied either to vss_A or vss_SW according to the region!
The remaining problem is the parasitics of the local substrates and deep-N-wells themselves, which are still being tied to the "0" node. For example, the capacitances between the vss_SW region (local p-substrate) and vdd_SW (the deep-N-well underneath it) and the capacitance between the vdd_SW (deep-N-well) and vss_SUBSTRATE (the global p-substrate underneath it), are tied in the extracted netlist to node "0" instead of vdd_SW and vss_SUBSTRATE, respectively. I can force all these remaining nodes to be tied together to a net of my choice using the "Ground node name" field, so if I specify "vss_SUBSTRATE" there I get the caps for the deep-N-well regions correctly connected, though the ones for the local p-substrates are now wrong.
I guess this is as far as I can go, and even though not completely correct, probably the netlists extracted this way will be much accurate than the ones I use to simulate, where all the parasitics were tied to the same node (I would normally choose "vss_SUBSTRATE").
...If someone knows an alternative approach to achieve this type of extraction, please let us know!
Thank you guys so much for your help!
Cheers,
Jorge.