Thank you all for the insightful comments.
On its face it looks like a disaster, but when you think about it, it may be possible when both Neutrals are bonded to ground. But this makes it sensitive to any accidental un-bonding errors.
Look at the schematic below, this is what I am trying to do in simple form.
I do not see any closed circuits whereby one inverter may be affecting the other on the AC side (but I am not showing the DC side, and both inverters work off the same battery bank - no problem with this as I've been running like this for years powering two separate AC circuits off these two inverters).
2 inv.pdf (27.11 kB - downloaded 378 times.)
The instantaneous voltages:
V1 =/ V2 in general
Vn1 =/ Vn2 in general (unless N bonded to ground, then Vn1 = Vn2 = 0 with respect to ground)
These are decent quality MSW inverters, with the V-waveform selected to match the PSW parameters, so I'd like to theorize as much as possible before actually turning the switch on for this whole thing.
I've talked to the factory a while ago, and they were going to do a hardware test using their own inverters, but I haven't heard back from them.
This house will be powered by inverters only, no going back and forth between inverters and the grid. For the generator I have a different setup.
I agree there is really no true neutral in this inverter (just the two H-bridge AC output terminals), but I refer to it as Neutral because the wire is white (the other wire is black).
I still need to think about your points made about shorting through the battery or the battery becoming hot, so initially my plan is to proceed conservatively and install two separate Neutral bus bars. Only after the measurements show no potential differences between the two bars, and no other problems at the battery, making all connections at one common Neutral bar.
Even if the testing shows that one Neutral bar will not work, these measurements should give interesting insight how this circuit works, given these specific inverters.