Supplemental Off-Grid Solar Heat Pump

[beltim]

Hello all,

I've been doing some thinking about ways to get into the solar space as a hobby and to learn, with hopefully a useful output.  I briefly considered getting a small solar panel and using it to charge rechargeable batteries, but after laptops and phones, my household doesn't use much battery power.  Plus, that sort of operation would either require being at the solar panel to charge, or charging using a secondary battery, greatly increasing the cost.  What I'd really like is a small system that can automatically make good use of solar power, without tying it to a grid.  This could be used for renters, for example, as the system could be disassembled and moved.

This led me to considering a supplemental HVAC system.  A few solar panels (~1kW) should be sufficient to power a small DC window unit or split heat pump.  Such a system that could be programmed to provide heat in winter, and air conditioning in summer, could take off some of the load of a home HVAC system.  Knowing that extensive battery use would kill the economics of such a system, I'd like to avoid batteries, but I suspect a few batteries may be required in order to even power delivery when the solar panels produce almost but not quite enough power.  I have read @Syonyk 's report on his off-grid setup as well as threads on Solar Resources and @Bakari 's off-grid system.

From my research, the two largest problems with this idea are automation and cost.  First, automation: I don't know at what point programmability enters such a system.  Can a charge controller be programmed to start a heat pump once the output from the solar panels reaches a certain minimum, and turn off once the output falls below a certain minimum?  Are there heat pumps that can change heating/cooking capacity based on received power?  Is there another component that can be added to the system to accomplish this goal?

Second, there is the issue of cost.  One off the shelf package that I've seen that would accomplish what I'm thinking of can be seen at https://www.hotspotenergy.com/DC-air-conditioner/DC-AC-Complete-Systems.php for about $4k all-in.  At ~$4.50 per watt, this is expensive, but half the cost is a highly efficient DC heat pump.  From that, here's some rough math.  First, the assumptions:
The system will be used for heating 5 months and cooling 5 months. 
The system will be placed in an area that receives on average 4.5 hours of sun per day.
Because average power draw of the heat pump is about 500W, I will use that as the effective power of the panels.
The heat pump is more effective as AC than a whole home unit (EER of 20 compared to 12).
Any heating or cooling done by this unit would otherwise have been done by the whole home unit.

Amount of energy saved in cooling months:
5 months * 30 days/month * 4.5 hours/day * 500W * (20 / 12) efficiency gain = 562 kW*hr of electricity saved by using solar
 Amount of energy saved in heating months:
5 months * 30 days/month * 4.5 hours/day * 500W  = 338 kW*hr of electricity saved by using solar
Total = 900 kW*hr per year

If home power costs 13 cents per kW*hr, that saves $117 per year, which gives a payback period of 34 years, which is beyond the lifetime of many if not all of the components.  It seems that the efficiency gains of the DC powered heat pump are outweighed by the high cost of the unit, which is probably about twice as much as an AC powered unit and sufficiently large inverter would cost.  Is this why it was so difficult to find DC powered heat pumps?

If such a system were actually implemented in an owner-occupied home, the excess power from the panels could be diverted to hot water heating to make this a far more economical system.  As is, if the system lasted 20 years, with one change of batteries, I calculate the total cost to be about $0.27 per kW*hr – economical in high power locations, but two to three times the price in low-price electricity locations.

I know that solar power posts always miss something, and I have only recently begun reading about this, so I welcome people pointing out errors or gaps in my thinking, as well as suggestions to improve the economics of the system.

[Ripple4]

you don't talk about the inverter, that is where the secret sauce is. all your concerns seem to stem from needing to learn about this part of the system. while I'm not a very experience in solar, I have installed one larger PV system, 5200 watt-power that is off-grid and its is economy minded, search for my posts.

In your case if I can devine what you might want, 1) Economy: a payback in only a few years 2) reduce your eletric bill by displacing home loads from grid 3)without a battery at all, 4) and also don't want to be official and get a interconnection agreement with the utility. So taking all that in I suggest look at grid-tie inverters with Zero-export, or also called grid-limiter. it will act like a grid tie inverter but will not create a out-flowing current as seen by your meter, or that's the idea. it has a current clamp that goes in the breaker panel and will provide as much power as it can up to the point that it exceeds the usage at a given moment, I've not used one so I cannot say how well that works. they make a allegedly 1000w that has plenty of YouTube videos.

building on that, there might be other reasons to want off grid for real, perhaps grid down backup power, mobile applications like a bug-out kit or RV, maybe pure learning. if its off grid its got to have a battery of one kind or another. I know a lot of the people out there use lead batteries, but I would caution against that. lead batteries have poor round-trip efficiency, think of it this way, you put $1 in the 'bank' and only get 80 cents back its terrible interest rate, you would never do that with money. Which is to say lead batteries lose 20% of the energy you try to store in them. But since nothing in nature is ever wasted, the lead battery turns some of that that 20% of wasted energy into explosive hydrogen gas, Yay! whereas with lithium we are talking 98% round trip efficiency, its amazing. I suggest reading about lithium iron phosphate. 1) this type of battery is more fire safe than lithium ion which are themselves safe, both need a BMS and active balancing. 2) LIFEpo4 products are taking off on amazon and eBay, there are plenty of 12v, large AH batteries at a cost that are within a few multiples of Lead. 3) they get thousands of charge cycles so they don't need replaced every few years like the 900 cycles or so of lead, which would kill your payback. 4) you can use 80-90% of the capacity of a Lifepo4, so that 100ah battery will give you 80-90ah, whereas to come close to multi year life with lead, it can only be discharged to 50% so that 100ah battery only gives you 50ah of storage.

The automation you mention can be built into higher end inverters, for instance many higher end off-grid inverters also have a battery charging feature that will recharge the battery AND power the loads in a bypass mode from a generator. so instead of a generator, use the grid, and set the battery charging to zero, bam off grid and no loss of convenience or switches to flip. this is how I did it in my home. starting AC units with off-grid is a problem, even a small 8kbtu window airconditioner can demand 130+amps when started, so unless you've got an inverter that has some weight, its not going to start it. my central AC takes 155amp/240 to start, so the meager 40amps peak current of a brand name inverter cannot touch it, even if it has enough power to run the 9amp/240 draw once its on. my solution is to bring in another kind of automation, the Arduino. So many home hobbyist problems are solved with the 'if I could measure this and have this happen' capability of the lowly $2.50 single board computer. the plan is to monitor several current and voltage levels of the system, and once the AC has started on grid to wait 20 seconds and then switching the live load over to solar. its more than a little tricky to do that and not a simple matter, but possible.

[beltim]

you don't talk about the inverter, that is where the secret sauce is. all your concerns seem to stem from needing to learn about this part of the system. while I'm not a very experience in solar, I have installed one larger PV system, 5200 watt-power that is off-grid and its is economy minded, search for my posts.

In your case if I can devine what you might want, 1) Economy: a payback in only a few years 2) reduce your eletric bill by displacing home loads from grid 3)without a battery at all, 4) and also don't want to be official and get a interconnection agreement with the utility. So taking all that in I suggest look at grid-tie inverters with Zero-export, or also called grid-limiter. it will act like a grid tie inverter but will not create a out-flowing current as seen by your meter, or that's the idea. it has a current clamp that goes in the breaker panel and will provide as much power as it can up to the point that it exceeds the usage at a given moment, I've not used one so I cannot say how well that works. they make a allegedly 1000w that has plenty of YouTube videos.

I didn't mention an inverter because as designed, my system is entirely DC.  That eliminates the need for an inverter, reducing power losses, and increases HVAC efficiency.  I wonder if that also limits my options for automation (I hope not!).

Your divination of what I want is partially correct - 1, 2, and 4 are dead on.  I don't mind having a battery, and it is probably necessary based on my design – either for startup of a heat pump or to smooth out power flows – but I don't want to run down the batteries every day, nor do I care about being able to use power from this system in the absence of photons.

The big thing that you missed is that I don't want to tie it to a grid.

[Ripple4]

While I’ve not done it, I’ve seen YouTube videos of people changing out the hermetically sealed alternating current compressor in a miniature fridge with ~$150 Danfoss 12VDC herm compressors to use in a RV. With the limitation of only wanting DC power, one low cost option might be to find a 12VDC hermetically sealed unit big enough for a window AC unit and swap it out.

[beltim]

I've discovered one solution.  A product like the Viktron BatteryProtect allows for customizable low voltage disconnects.  For lead acid batteries, there's a nice, well-known, and distinct conversion between voltage and charge state, so I could program the circuit to turn on at, say, 90% of battery capacity, and off at 80% of battery capacity.  This means that the system runs whenever there's sufficient power output from the solar panels to charge the battery above 90%, but prevents the depth of discharge from every exceeding 20%.  The batteries thus act as more of a power stabilizer than a battery, and should allow for a very long battery life since they're not being discharged very much.

Does that make sense?

Incidentally, a lot of charge controllers have low voltage disconnects, which in principle should serve the same function, except they usually have far too low a power distribution for my purposes.

[wienerdog]

What type of heat do you have now?

[beltim]

What type of heat do you have now?

For comparison purposes, let's say heat pump.

[beltim]

It looks like a AC unit plus inverter is more cost-effective.  A 12,000 BTU window heat pump can be found for ~$750, and I found an Energizer 2000W power inverter for $200.  The AC heat pump turns out to be just as efficient as the DC unit.  That saves $1k on the system compared to what I wrote above.

[Syonyk]

Well, since I was mentioned by name...

Relevant background reading here (my blog, trying to offer a good amount of required background for discussions like this):

https://syonyk.blogspot.com/2018/05/why-typical-home-solar-setup-does-not-work-off-grid.html

And, also related, https://syonyk.blogspot.com/2018/05/so-you-wanna-go-off-grid.html

The main point you have to understand from the first link is how solar panels behave - constant current, up to a point defined by the normal illumination, then the voltage collapses.  Without storage, you can either run them far below maximum power, or you don't have a stable system - and very few devices like their power supply collapsing under them without warning, then oscillating back up to full voltage.

What I'd really like is a small system that can automatically make good use of solar power, without tying it to a grid.  This could be used for renters, for example, as the system could be disassembled and moved.

Such things don't really exist, because they're staggeringly expensive for the small amounts of useful energy they produce.

Even full off grid power systems, like I have, aren't cost effective compared to grid power.  If I get lifetime per-kWh costs on my system down to $0.25/kWh, I'm doing quite well - and grid power out here is $0.10/kWh.  I think I'm sub-$1/kWh now, based on power delivered, but that's after 4 years of operation.  They're not cost effective for a variety of reasons, though panel costs have come down a good bit in the last few years.  It still doesn't really change the economics of the system that much, because panels are an increasingly small fraction of the cost of a standalone system.

This led me to considering a supplemental HVAC system.  A few solar panels (~1kW) should be sufficient to power a small DC window unit or split heat pump.

"Should be" is not a particularly strong engineering statement.  Panel output is rated at STC conditions - which are a set of conditions that are very useful for testing and grading cells in a manufacturing facility, and have roughly no chance of being met in most actual conditions.  So that "1kW" of nameplate will produce less in any realistic condition, and far, far less in winter conditions.

I have 2280W of panels hung on my main office array.  I rarely see anything past about 1600W input, though this is partly because they're not aimed ideally at the moment.  In the winter, on a bad day, I'll see less than 75W - peak.

Depending on the unit, you also have to handle the startup surge.  My window unit (which is a non-inverter type) pulls about 700W running, but pulls close to 2kW to start.  If you can't meet that startup surge, you can't run the unit.  My small backup generator would be able to run the unit, but it can't start the unit, so I can't run the unit on generator.  Not a big deal, but it's something to consider.  I don't actually know the momentary peak draw, but it's significant.

From my research, the two largest problems with this idea are automation and cost.  First, automation: I don't know at what point programmability enters such a system.  Can a charge controller be programmed to start a heat pump once the output from the solar panels reaches a certain minimum, and turn off once the output falls below a certain minimum?  Are there heat pumps that can change heating/cooking capacity based on received power?  Is there another component that can be added to the system to accomplish this goal?

Unless you're loading the panels up, there's no way to tell what the current available power is.  The way you do this is to load them up, monitor the voltage, and see when it starts to drop - a MPPT sweep is a common term for this scan.  But you can't just look at the voltage and guess.  You might be able to put a normal light sensor near the panels, model them, track panel temperature, and guestimate that way, but that's a pretty custom setup - there's nothing off the shelf I'm aware of that will do that.

Then, to make use of it efficiently, you'd need to be able to vary the compressor power consumption, rather rapidly, or you'll collapse the voltage if a plane flies over and shades a panel.  It's not impossible, but it's also more of a "custom engineering" project than an off the shelf unit.

Second, there is the issue of cost.  One off the shelf package that I've seen that would accomplish what I'm thinking of can be seen at https://www.hotspotenergy.com/DC-air-conditioner/DC-AC-Complete-Systems.php for about $4k all-in.  At ~$4.50 per watt, this is expensive, but half the cost is a highly efficient DC heat pump.

A significant chunk of the cost is battery, and those batteries, used heavily, are unlikely to last more than 3-4 years.

I'm not going to rip their numbers apart, but I'll simply observe that I think they're a load of trash, and I wouldn't believe them for a second.  They're the sort of numbers put together by people who don't actually have any hands on experience with year round off grid operation, do a tiny bit of research, and then throw together numbers that look decent.  They don't reflect any sane reality.

Amount of energy saved in cooling months:
5 months * 30 days/month * 4.5 hours/day * 500W * (20 / 12) efficiency gain = 562 kW*hr of electricity saved by using solar
 Amount of energy saved in heating months:
5 months * 30 days/month * 4.5 hours/day * 500W  = 338 kW*hr of electricity saved by using solar
Total = 900 kW*hr per year

You have clear skies, all winter long, and are going to be aiming your panels to track the sun every day?

I guarantee your panels won't produce nearly the power you think they will in actual conditions.  Minor cloud cover impacts energy produced quite a bit (which is why I'm pretty badly overpaneled - to handle cloudy days, hazy days, etc), and winter conditions vary, but on a bad winter day, you'll produce absolutely nothing.  My office system can produce (really, consume - I'm consumption limited most of the time) 10kWh a day in halfway decent conditions, but on a bad winter day, I'll be lucky to get 0.3kWh out of the sky.  It's not a big deal as I've got the backup generator, but if you assume full rated panel production all year round, you're going to be very, very disappointed.  Cut your production numbers in half, and you might get close to annual production, though PVWatts would give you a more accurate estimate given your site.

Is this why it was so difficult to find DC powered heat pumps?

They're a weird, niche little product that doesn't have a good reason for existing, so... yeah, they're expensive.  I expect a lot of the market is actually to the RV folks, because they seem to largely not care about cost.

Inverters got good about a decade ago.  It's now far cheaper and easier to have a large inverter and use AC loads.  As soon as anything requires the inverter, you're paying the idle penalty (somewhere in the range of 30W on my 2kW continuous/6kW peak unit), so may as well use the cheap stuff designed for grid power, and save money on wiring.  DC systems just don't make a lot of sense for any realistic use case.

If such a system were actually implemented in an owner-occupied home, the excess power from the panels could be diverted to hot water heating to make this a far more economical system.  As is, if the system lasted 20 years, with one change of batteries, I calculate the total cost to be about $0.27 per kW*hr – economical in high power locations, but two to three times the price in low-price electricity locations.

Yeah, that sounds sane.  Plus or minus some, but you're certainly in the right ballpark.

...as well as suggestions to improve the economics of the system.

It's not possible with current technology.  You're missing a bit, but the stuff you're missing would just drive the system costs up.  The analysis you've done is mostly correct - it simply doesn't make any financial sense to do it.  Which, I'd add, is why what you're looking for doesn't really seem to exist.

=============

4) and also don't want to be official and get a interconnection agreement with the utility. So taking all that in I suggest look at grid-tie inverters with Zero-export, or also called grid-limiter. it will act like a grid tie inverter but will not create a out-flowing current as seen by your meter, or that's the idea. it has a current clamp that goes in the breaker panel and will provide as much power as it can up to the point that it exceeds the usage at a given moment, I've not used one so I cannot say how well that works. they make a allegedly 1000w that has plenty of YouTube videos.

In most areas, you can't legally do that.  If you've got a power source on the home wiring, even if it's "zero export," you still have to meet NEC regulations, and almost certainly still have to have an agreement with the power company - because the system could backfeed the grid.  Even if it's not designed to, if the sensors malfunction, it could backfeed, so they get to know about it.

YouTube videos on solar are collectively godawful advice, mostly put together by morons, and frequently involve non-UL listed equipment that cannot be interconnected to any sane grid legally.

The nicest thing I'll say is that 90% of what YouTubers say is fine, 10% is incredibly hazardous advice, and it takes an awful lot of experience to tell the difference.  Some of the "slick" well produced videos have some of the worst advice.  But anything found on YouTube can safely be assumed to be somewhere between wrong and actively hazardous advice.

Also, most landlords don't allow renters to make modifications to the electrical system or structure anyway.

...like the 900 cycles or so of lead, which would kill your payback.

Haven't been near a modern solar use lead acid datasheet in a while, have you?  Any decent pack will last far, far longer.

There's also the seasonal aspect - in the summer, with properly sized panels, you basically don't use the battery bank most of the day.  My office can be into float by 10AM during the summer, and I actually don't even run the absorb cycle every day in the summer to be easier on the batteries.  In the winter, you need the batteries more, so having the extra capacity is nice (though lead does have somewhat less capacity in the cold - on the flip side, you don't have to keep them warm like you do with lithium).

But I don't really want to head off into the weeds of batteries.  You're not wrong, but you make the usual set of claims that turn out to not matter as much as people like to make them out matter in off grid systems.  None of it changes the economics of the system much either way, and I'd rather not have a conversation with insurance adjusters if a home built lithium pack burns a house down.  There's a reason my office is separated from the house, and the fact that I do lithium work in there is a major factor.

You're right about automation being a DIY project for this sort of thing, though.

It looks like a AC unit plus inverter is more cost-effective.  A 12,000 BTU window heat pump can be found for ~$750, and I found an Energizer 2000W power inverter for $200.  The AC heat pump turns out to be just as efficient as the DC unit.  That saves $1k on the system compared to what I wrote above.

Correct.  And you still need batteries either way.  And more panels.

But I still doubt you'll make the math work to save you money on this project.

[beltim]

Well, since I was mentioned by name...

Thanks for stopping by, and being generous with your thoughts.  I read a good chunk of your blog before starting this thread.

I think I muddied the waters by saying that I wanted to avoid batteries.  I meant that I didn't want to run a system where the batteries were drawn down every day, or to have backup storage sufficient for backup power in case the grid goes down, or other common uses of batteries.  Instead, I want to use batteries to smooth out power delivered to the heat pump, in order to avoid sudden swings in power provided by solar panels.  On cloudy days, I expect that my design wouldn't run the heat pump at all.

This led me to considering a supplemental HVAC system.  A few solar panels (~1kW) should be sufficient to power a small DC window unit or split heat pump.

"Should be" is not a particularly strong engineering statement.  Panel output is rated at STC conditions - which are a set of conditions that are very useful for testing and grading cells in a manufacturing facility, and have roughly no chance of being met in most actual conditions.  So that "1kW" of nameplate will produce less in any realistic condition, and far, far less in winter conditions.
I have 2280W of panels hung on my main office array.  I rarely see anything past about 1600W input, though this is partly because they're not aimed ideally at the moment.  In the winter, on a bad day, I'll see less than 75W - peak.
Depending on the unit, you also have to handle the startup surge.  My window unit (which is a non-inverter type) pulls about 700W running, but pulls close to 2kW to start.  If you can't meet that startup surge, you can't run the unit.  My small backup generator would be able to run the unit, but it can't start the unit, so I can't run the unit on generator.  Not a big deal, but it's something to consider.  I don't actually know the momentary peak draw, but it's significant.

First, I wanted to set up the system such that the batteries and inverter are capable of providing the startup surge.  However, it appears that real-world results indicate my calculations are off, both in terms of peak production and sustained production.  I used a variety of calculators to suggest that I was in the 1600 kWh / kW year area of the country.  The 4.5 hours per day was simply dividing that number by 365 days to get an average amount of hours of power per year.  Is this not a reasonable way to calculate average generation?
The peak generation is more worrisome.  If your peak production is just 70% of listed capacity, what's your average?  Adding one additional panel to my design to have listed power ~2x the sustained draw of the heat pump is one thing, but if I have to double the number of panels because the yearly production is only, say 500 kWh per kW panel for the year, that changes things a lot.  Come to think of it, that changes the economics of all solar projects a lot.  Doing some research suggests that the current leveled cost of energy for current residential solar projects is about $0.16 per kWh, which is 2-3 times higher than what the payback calculators all suggest.  Interesting...

From my research, the two largest problems with this idea are automation and cost.  First, automation: I don't know at what point programmability enters such a system.  Can a charge controller be programmed to start a heat pump once the output from the solar panels reaches a certain minimum, and turn off once the output falls below a certain minimum?  Are there heat pumps that can change heating/cooking capacity based on received power?  Is there another component that can be added to the system to accomplish this goal?

Unless you're loading the panels up, there's no way to tell what the current available power is.  The way you do this is to load them up, monitor the voltage, and see when it starts to drop - a MPPT sweep is a common term for this scan.  But you can't just look at the voltage and guess.  You might be able to put a normal light sensor near the panels, model them, track panel temperature, and guestimate that way, but that's a pretty custom setup - there's nothing off the shelf I'm aware of that will do that.

Then, to make use of it efficiently, you'd need to be able to vary the compressor power consumption, rather rapidly, or you'll collapse the voltage if a plane flies over and shades a panel.  It's not impossible, but it's also more of a "custom engineering" project than an off the shelf unit.

Isn't much of this addressed by using batteries to level power delivered to the heat pump?  The way I've designed the system, there's still a hard cutoff when the batteries drop below a certain level, but hopefully only once a day unless the weather changes drastically during the day.

Second, there is the issue of cost.  One off the shelf package that I've seen that would accomplish what I'm thinking of can be seen at https://www.hotspotenergy.com/DC-air-conditioner/DC-AC-Complete-Systems.php for about $4k all-in.  At ~$4.50 per watt, this is expensive, but half the cost is a highly efficient DC heat pump.

A significant chunk of the cost is battery, and those batteries, used heavily, are unlikely to last more than 3-4 years.

If I'm only discharging the battering ~20% per day, my research suggestions that flooded lead acid batteries should be good for 6-8 years.  Is that also based on overly optimistic manufacturer reports?

Amount of energy saved in cooling months:
5 months * 30 days/month * 4.5 hours/day * 500W * (20 / 12) efficiency gain = 562 kW*hr of electricity saved by using solar
 Amount of energy saved in heating months:
5 months * 30 days/month * 4.5 hours/day * 500W  = 338 kW*hr of electricity saved by using solar
Total = 900 kW*hr per year

You have clear skies, all winter long, and are going to be aiming your panels to track the sun every day?

I guarantee your panels won't produce nearly the power you think they will in actual conditions.  Minor cloud cover impacts energy produced quite a bit (which is why I'm pretty badly overpaneled - to handle cloudy days, hazy days, etc), and winter conditions vary, but on a bad winter day, you'll produce absolutely nothing.  My office system can produce (really, consume - I'm consumption limited most of the time) 10kWh a day in halfway decent conditions, but on a bad winter day, I'll be lucky to get 0.3kWh out of the sky.  It's not a big deal as I've got the backup generator, but if you assume full rated panel production all year round, you're going to be very, very disappointed.  Cut your production numbers in half, and you might get close to annual production, though PVWatts would give you a more accurate estimate given your site.

I addressed this before, but it's so important to the overall economics I'd like to revisit it.  I put my info into PVWatts, and I get 1400 kWh for the year per 1kW panel, which is ~88% of what I was calculating above.  Is PVWatts a realistic estimate, or do I need to halve my original projections?

[Syonyk]

Instead, I want to use batteries to smooth out power delivered to the heat pump, in order to avoid sudden swings in power provided by solar panels.  On cloudy days, I expect that my design wouldn't run the heat pump at all.

You can do that, but life is still far easier if you've got a heat pump that can largely follow production on a second by second basis - which I don't think exists.  They're just not designed for it.

Cloudy days mean you won't run it, certainly.  But you're doing cost calculations based on it running every single day, for the rated period of time.  Which, simply, won't happen.  The details depend on your environment - Seattle is a different environment than southern AZ.

First, I wanted to set up the system such that the batteries and inverter are capable of providing the startup surge.  However, it appears that real-world results indicate my calculations are off, both in terms of peak production and sustained production.  I used a variety of calculators to suggest that I was in the 1600 kWh / kW year area of the country.  The 4.5 hours per day was simply dividing that number by 365 days to get an average amount of hours of power per year.  Is this not a reasonable way to calculate average generation?

It gets you a number, but not (IMO) a particularly useful one.  You want to know how much per month you're going to be generating, and then mapping that against heating/cooling degree days to find out how useful the power offset is likely to be.

In the winter, you get very few hours of sun, in the summer you get a lot - but it's also hotter in the summer, so production is less.  At least where I am, various forest fires tend to haze the air up in the summer, so summer and early fall production is worse than spring production.

As an example, if it's 60 degrees out and sunny, I get great production - but I don't need much of anything in the way of thermal management for the house.  Actually, my office needs cooling above about 30F if it's sunny, just because I run a bunch of spare computer equipment to blow off some of the excess into Folding@Home/BOINC, but... point stands, I'm only cooling because I'm running loads.  If I don't have those loads running, I can just work out there all day without any real thermal management.

So, for your concept, power generated is only valuable if it can go to heating/cooling.  If you don't need to heat or cool, or there's no power available to heat/cool with, you're not generating any value.  Look at the monthly numbers, they're far more interesting.

The peak generation is more worrisome.  If your peak production is just 70% of listed capacity, what's your average?  Adding one additional panel to my design to have listed power ~2x the sustained draw of the heat pump is one thing, but if I have to double the number of panels because the yearly production is only, say 500 kWh per kW panel for the year, that changes things a lot.  Come to think of it, that changes the economics of all solar projects a lot.

Honestly, I don't know.  I'm demand limited, so my daily production is (on a sunny day) controlled by how much I can consume, not what the panels will generate.  And the aiming is currently pretty funky, because I'm overpaneled.

But look at the PTC numbers for your panels (as opposed to STC) for a better estimation of production, or just punt to PVWatts, they get close enough.

Then add in some panel degradation over years, soiling... the "nameplate number" on a panel is quite worthless.  Even wind speed has an impact on production - you get more energy on a windy day because it cools the panels better.  There's no hard and fast number, and if you get within about 10-15% either way of your panel production actual values, it's a good number.  Short of putting a panel out on a synthetic load and measuring, it's just hard to get exact values.

Doing some research suggests that the current leveled cost of energy for current residential solar projects is about $0.16 per kWh, which is 2-3 times higher than what the payback calculators all suggest.  Interesting...

It depends on roughly everything, including local rate and tax structures for production.  Residential solar isn't very cost effective, though.  It's 3-4x more expensive (typically) than large commercial farms.

Isn't much of this addressed by using batteries to level power delivered to the heat pump?  The way I've designed the system, there's still a hard cutoff when the batteries drop below a certain level, but hopefully only once a day unless the weather changes drastically during the day.

Yeah, the batteries help a lot.  Just add cost and maintenance.

Honestly, if you're putting batteries in, you may as well put a useful inverter in, because it's backup power if you want it or not, at that point.  You're not talking about a tiny system when you've got a good bank of batteries.

If I'm only discharging the battering ~20% per day, my research suggestions that flooded lead acid batteries should be good for 6-8 years.  Is that also based on overly optimistic manufacturer reports?

It depends a lot on the batteries and the charging/maintenance profile.

I assume you've run across my 11k word epic on lead acid: https://syonyk.blogspot.com/2018/04/off-grid-rv-lead-acid-maintenance-charging-failure-modes.html

If you buy generic "deep cycle" batteries from Walmart, and cycle them daily, I'd be surprised to see 2-3 years on them.  If you get a good brand name (Trojan or Crown or something) designed for solar/off grid use, and treat them properly (which means enough charging voltage, for long enough, temperature compensated, water them a few times a year, etc), you can easily get a decade on them.  A good solar industrial battery should get closer to 20 years, but you probably don't want to bother with them.  Large, heavy, not exactly mobile.

I addressed this before, but it's so important to the overall economics I'd like to revisit it.  I put my info into PVWatts, and I get 1400 kWh for the year per 1kW panel, which is ~88% of what I was calculating above.  Is PVWatts a realistic estimate, or do I need to halve my original projections?

PVWatts is close enough.  But it won't factor in the days when you don't need heating/cooling.

There's basically no realistic way to make your system make any financial sense whatsoever.  If you want to build it to mess with, have a blast, just don't expect it to save money.

[beltim]

Instead, I want to use batteries to smooth out power delivered to the heat pump, in order to avoid sudden swings in power provided by solar panels.  On cloudy days, I expect that my design wouldn't run the heat pump at all.

You can do that, but life is still far easier if you've got a heat pump that can largely follow production on a second by second basis - which I don't think exists.  They're just not designed for it.

Cloudy days mean you won't run it, certainly.  But you're doing cost calculations based on it running every single day, for the rated period of time.  Which, simply, won't happen.  The details depend on your environment - Seattle is a different environment than southern AZ.

It turns out that company that sells a DC heat pump sells one that can run entirely on solar without batteries and adjusts its output based on solar production.  It also can be plugged into grid AC sources to boost output when solar input is low or when you want to run the system at night.  It's less versatile, obviously, but I wonder if taking out all of the other components makes the economics better.  I'll run some numbers later.

The unit is at https://www.hotspotenergy.com/solar-air-conditioner/