Sol-Ark 12K shutting down on grid failure

[Ampster]

Why would limiting the current to 100A fix anything?

I don't know but if it works you have a solution that works, at least until you can get more information to confirm if there was voltage sag on your battery pack. Perhaps there was some other parameter that was changed concurrently with changing current settings. Perhaps you can increase the current settings incrementally to see if when problem reoccurs?
EDIT: Following my post, @RCinFLA gave a good explanation of what may be happening.

 

[RCinFLA]

Sorry for being dense but I'm still very confused:

1. If my loads are ~2 kW when the grid fails, that's only about 40A out of the battery (confirmed by the inverter and the battery display). Why would switching over to the inverter cause a spike near 150A? Any motors should not experience startup current spikes in this situation, right?
2. Why would limiting the current to 100A fix anything? If there are in fact motor loads spiking large currents over 100A, wouldn't capping the total current in fact cause those loads to shut down altogether?

I did put an inrush clamp meter on the battery and it only recorded 3A which obviously doesn't make sense. The same Fluke meter accurately measured steady state loads (or at least matched the battery display).

As for the resistive heating loads, those can be managed. I'm just trying to understand the inverter/battery interrelationship at the moment.

Thanks for the continued help.

When grid collapses, the inverter which is running in parallel with grid, will momentarily overload up to full surge capability of inverter before releasing pass-through relay disconnecting inverter from grid. Surge can last up to about 8 msecs with corresponding short spike in battery current of a few hundred amps.

The actual surge depends how the grid collapses, abruptly or an overload slump, like a tree branch brushing power lines. Also depends where in the sinewave cycle the grid collapses. An abrupt grid drop right at the peak of sinewave of grid is the worst case instant surge, although a grid overload AC voltage slump lasting a few cycles can drag out the high load current on inverter before it releases pass-through relay.

An inverter set for zero back feed export will not prevent this inverter loading. There is a finite response time for inverter to recognize the situation and release pass-through relay.

Pass-through relays take a beating due to grid glitches and opening their contacts under near peak inverter AC current capability.

 

[afblock]

When grid collapses, the inverter which is running in parallel with grid, will momentarily overload up to full surge capability of inverter before releasing pass-through relay disconnecting inverter from grid. Surge can last up to about 8 msecs with corresponding short spike in battery current of a few hundred amps.

Okay that’s helpful. But then what is the discharge limit setting in the inverter actually doing? And why not apply that setting during spikes but allow higher discharge rates for steady state loads?

 

[afblock]

Perhaps you can increase the current settings incrementally to see if when problem reoccurs?

Yes, I am going to try this.

 

[RCinFLA]

Okay that’s helpful. But then what is the discharge limit setting in the inverter actually doing? And why not apply that setting during spikes but allow higher discharge rates for steady state loads?

Any AC current limits relies on reading CT sensor and firmware readjusting inverter. There is a minimum response time for this to be done. Inverter is dealing with 60 Hz which is a limiting factor on how fast things can be detected and responding to.

There are hardware limits on high frequency currents in high frequency transformer, MOSFET's, and IGBT's to help protect them to their hardware damage limits.

SolArk has a large bank of HV DC capacitors that absorbs some of these variations but it cannot totally insulate operation from a grid drop out.

 

[Quattrohead]

Okay that’s helpful. But then what is the discharge limit setting in the inverter actually doing? And why not apply that setting during spikes but allow higher discharge rates for steady state loads?

This could be due to the inverter ramping it's output up slowly at the cost of a slower switch over time instead of ramping up quick with minimum switch over time.