I have the Schneider XW-Pro 6848 6,800 watts at 120/240 volt split phase. And I now have a usable 36 KWHs of Chevy Bolt battery modules wired in a 14S12P arrangement of 60 amp hour cells LG cells. That makes 720 amp hours at a nominal 51.8 volts. It's about 58% of a complete Chevy Bolt battery pack. Yes, I am following the recall news.
Since you want to run your pool heater during the day, I would just add more solar with microinverters. That is by far the most efficient, to use the power as it is being produced. Charging is 95% efficient, at best, and inverting is also about 93% efficient, and the batteries lose a little as well. I lose a full KWH when I move 10 KWH from solar to peak rate evening time here.
The problem with just adding Microinverters is that he stated his 6kW load is intermittent - than means when the heat exchanger turns off he be blasting an additional 6kW to the grid.
Aside from that likely taking peak export power beyond the terms of the NEM agreement, there is a safety issue - when the utility approves a NEM agreemebt, that is in the context of reviewing upstream capacity of all transformers - if you push more peak power out to the grid than you agreed you would, it can cause damage and possibly even a safety hazard.
We can argue about whether your NEM agreement gives you the freedom to put out a square wave of power at agreed-upon peak versus the usual solar ‘hill/mountain’ around midday, but pushing out 6kW above your peak limit without review/permission first is a very bad idea.
And my rates go from 17 cents a the cheap time to 43 cents at the peak rate. Moving that 10 KWH in theory saves me 10 x 0.42 - 10 x 0.17 = $2.50 cents a day. I figure I average moving about $2.00 a day for 300 days a year to account for days of weak sun or other reasons I can't push the full 10 KWH in the peak window. I may actually do better than that. But that still means I am saving $600 a year by time shifting the power. But I spent $3,000 on the inverter, and probably something like $5,000 on the batteries. Maybe a little more with the cases, BMS units, and wiring etc. So taken totally separate from the solar power, the battery system will take 15 years to pay off. But I am only expecting the batteries to last about 10 years. So yup, at the peak to off peak rate here, it won't pay off, but it sure does cut down on the cost of having backup power.
Yeah, the batteries are the killers, aren’t they? I was planning a Hybrid install much like yours (Magnum PAE instead of Schneider Conext) but the infinite path to break even as well as the cost/complexity of rewiring my mains panel into a critical loads panel scared me into looking for another alternative.
But if you take the entire cost of my solar installation into the equation, and figure on what I am saving, then my system pays off in about 11 years. My Enphase solar PV system has already produced 16.9 MEGA watt hours. And it has just been running a bit over 2 years. In the full year of 2020, it made 7.9 MWHs. If I just figure that half of it was used at peak, and half off peak, so we average the value of the electricity, that is 30 cents per KWH x 7,900 = $2,370 per year chopped off of my electric bill, but that is with the battery doing the time shifting. Adding up everything I spent on the solar panels, inverters, installation, and my battery system, I spent $26,000 over a span of 2.5 years. Divide by the 2,370 per year = 10.97 years to cover the whole cost of everything just from electric bill savings. Production may fall off some, but I also expect So Cal Edison to jack the rates more. So that may end up a wash.
I’ve got a 4kW Microinverter-based system under NEM that was covering 100% of consumption until the TOU peak hour changes. That ended up costing me ~30% of my generation credits on an annual basis and added another ~5% to the cost of my consumption on an annual basis. So I need to add another ~1.4kW of production if I want to continue to cover my electrical bill.
I would have had to modify my NEM agreement to increase my peak output from 3.5kW to 4.75kW and would be subject to new rules adding taxes that I wanted to avoid.
So the DC-coupled GTIL system I added is to offset self-consumption and preserve generation credits so that I essentially consume 35% less from the grid (and keep my grid-tied generation credits where they were before the TOU changes.
I spent only $420 on 1.14kW of new panels and $500 on two GTIL inverters, plus another ~$500 for racking, wiring, and combiner boxes, so let’s say $1500 all-in or ~$1150 after tax credits.
I’d probably be charged over $250/year without that new system so on purely the solar generation front, the break-even is not bad (<5 years).
The rub is that to make the system work, I need to store up unused PV power until it is consumed, and that requires a battery and a charger, which cost me $1500 and $250 (or a total of $1350 after tax credits).
So the whole new system has cost me $2500 after tax credits and break even is back up to ~10 years (assuming the 3 cheap Chinese boxes I purchased last that long
).
I built this ‘platform’ with a plan to expand it once we get an EV (~2023).
If I add a single 500W panel (costing $200-250) to the DC array, it’ll offset another ~500kW of consumption annually which will allow me to preserve enough additional generation credits to drive ~1700 miles (worth $100/year).
So as my consumption increases, break even drops under 10 years.
This grand plan only works out if these cheap Chinese boxes last, but I figure my worst-case downside is to purchase higher-quality more expensive products when they crap out (this market is evolving so fast right now).
But back to the OP, I don’t think he can use Microinverters without exceeding his NEM cap so charging a battery at 20% / 95% cost = 21% at night then draining it to power his 6kW load during the day at 93% efficiency translates to ~77.4% savings (or at least 75% savings even if we throw in another 10% loss cycling energy through the battery).
The 6kW output peak is going to put him up into the higher-cost inverter category but between a hybrid such as the SolArk (or a 240V Deye if his heat exchanger takes 240V, as I suspect) or a Schneider Conext if it provides the charge-time control he is seeking is going to be the cheapest way to achieve what he is after.