I’ll start out the analysis by characterizing my home loads as well as the power sources. Interestingly, as I progressed on the journey, I ended up with not just two, but three different power sources!
HOME LOADS
I have a pretty good handle on my base loads and typical usage loads year round with my solar and PG&E net monitoring. While I don’t have individual circuit or appliance monitoring, I do have 10 years of gross and net consumption logging.
Load Profile | Day (16h) | Night (8h) | Daily total |
Normal | 733w | 400w | 15 kwh |
Base | 400w | 400w | 10 kwh |
Critical | 100w | 100w | 2.4 kwh |
My base loads currently run about 400W at night when everyone is sleeping. Over 24 hours that’s about 10kwh/day, and that’s what I confirmed when we were away on vacation last month. During the day, we typically use an additional 5kwh, spread out fairly evenly over 16 hrs. So that’s an additional 333W on top of base loads, for 733W. This is excluding any EV charging, which we would avoid doing at home during outages.
These are fairly modest loads, so the important part of this section is to caveat that I am afforded the privilege of even contemplating different options because of my mild climate and reliable natural gas for heating and cooking. Specifically my locale has a 7-month heating season and 2-week cooling season (with no central A/C). That leads to fairly constant loads over the year - if anything, we use maybe 1-2 kwh/day more during the winter months due to the furnace (5 hrs x 120W ECM blower) and more lighting hours.
So during an outage, if we want to live as normal, we need 15 kwh/day. But if we want to conserve to stretch our fuel, we can still power the house but be sparing in our usage, and get closer to 10 kwh baseload profile.
Finally, for the zombie apocalypse or SHTF, we can contemplate limiting consumption to only critical loads - turning off all circuits except the essentials, I think we could get down to 100w average or 2.4 kwh/day. Critical loads include:
-furnace (600wh)
-refrigerator (1200wh, 2x inverter compressor)
-tankless gas water heater (240wh, ignition only)
-internet modem (100wh)
-aquarium circulation pump only (240wh)
POWER SOURCES
Gas/Propane Generator
As documented earlier in this thread, I settled on the A-iPower 7100 inverter generator with 5700W peak because of the 240V output for the BPTM whole house transfer meter, and ability to run only on propane. It can handle any of my whole-house loads except the 240V electric oven and the EV charger at full output, neither of which I’d be using in an outage.
Now people like to say fuel generators (gas/propane) are the cheapest and best for extended outages, because you have an unlimited supply of fuel available. But that is true only to a certain extent, once you have to resupply, there’s a certain inconvenience factor that has to be factored in. Since I’m running only on propane for lower maintenance, I’m limited to my 2x 20lb BBQ propane tanks before I have to go out to refill. So making that 40lb of propane stretch longer is important for convenience (and during the zombie apocalypse, it may be absolutely essential).
The 100% 5700W peak is important to mention, as the power efficiency is directly tied to the load on the generator. Now the only formal spec to work with is 0.36 gal/hr of gas per hr of runtime at 25% load (1425W), which is an efficiency of 11.7% (1 gal gas = 33.7kwh energy).
Generator Fuel-to-AC Efficiency:
10% load - 5%
12.5% load - 7%
25% load - 11.7%
50% load - 15%
66% load = 16%
80% load = 17%
How to build a profile of the efficiency across a wider range of loads? Fortunately, there is an active community of solar/battery off-gridders who have to top off their batteries during winter and clouds using gas generators. Enough of these have done efficiency tests, that I can find 11-12% efficiency at 25% load is pretty typical of decent inverter generators of various sizes that matches mine. While I didn’t find any direct load tests with my A-iPower, a range of Yamaha, Honda, Champion inverter generators yielded pretty typical efficiency curves going up to 15% at 50% load, leveling off to 16-17% efficiency around 80% load.
On the lower end, not as many data points, but enough to suggest about 5% efficiency down at 10% load - basically most of the energy is just being used to spin the motor and create waste heat, with only a small fraction being used for loads. I think for non-inverter (i.e. modified sine-wave) generators, the efficiency starts to drop off rapidly below 50% load, so at least I had the better type of generator for my loads - I could have gained efficiency if I’d opted for the smaller WEN generator (3500W), as I’d be operating at a higher load, but probably that would’ve only been 1-2 percentage points at best, and that being a non-inverter generator, only if operating above 50% load.
Off-Grid Solar Panels
OK, so this was the most unexpected part of the journey. It was not in the plans at all, but somehow I’ve ended up with 800W of portable off-grid solar panels. First, as part of the Anker Kickstarter, they offered a few bonuses to get their portable panels at a huge discount, so I opted-in just to do some experimentation. And then I won a top prize in the Facebook user community for doing some initial testing of the Anker Solix and sharing/learning with the other early adopters.
So net is that I’ve ended up with $2000 worth of off-grid solar panels for a tiny fraction of the retail price, that perfectly match the solar inputs of the Anker Solix. They’re designed for camping and offgrid - so they’re expensive but high-quality, waterproof. I’ve tested them and gotten 75% of STC on a sunny day, comparable to rigid mono panels.
Now you would think these mobile, adjustable-tilt panels would be ideal for a power outage, esp with a supplied 30-foot cable I could place them in good sunlight. Well for my house, basically in the three months around winter solstice, the neighboring oak trees shade my entire yard and driveway except for two hours a day. On the roof, I can do slightly better, but between wind, rain, and clouds - the opportunities to harvest the sun are pretty much not worth it.
The other three seasons, I can get good sun in the driveway, or even better on the roof - and since my 4kw rooftop array will be offline during an outage, I can even drape the portable panels right on top of my best-performing rigid panels for each time of year. Since I have my Enphase per-panel logs for the entire year, I can reliably predict how much power I can harvest.
So my projection is that I can reliably get 4 kwh on a sunny day, over a 6-8 hour period, during 9 months of the year. But since not every day is sunny, I’ll do all my analysis both with and without solar. 4kwh/day is not huge, but not insignificant during an outage, especially with my base loads. It does require the power station to be functional, but who says this is not a “solar generator” - with solar panels it is absolutely one!
Battery Power Station
I picked the Anker Solix F3800 from the emerging crop of portable/home power stations because of the 240V compatibility with the BPTM transfer meter - my other options would have been the Zendure Superbase V or 2x EcoFlow Delta Pro (Delta Pro Ultra was not announced til after I’d picked the Anker). I decided on Anker because of the Kickstarter discount, and the expectation of quality, reliability, customer support and longevity (the company, not just the product) - so far they’re living up to expectations.
It’s a nominal 3.8 kwh battery - not sure if 100% is usable, haven’t figured out if they’ve reserved top and bottom buffers yet, but will use 3840Wh as the baseline for analysis. There were some early Youtube load tests, but they were on prototype samples. It’s also 6000W (9000W surge), so can handle even higher loads than my gas/propane generator - it has no problems charging my Tesla at 25A, if I desire to do so, the battery capacity just doesn’t last too long doing that.
Now power stations are also far from being 100% efficient. Anker does not specify any roundtrip efficiency figures like Tesla Powerwall does, but even if they did, I wouldn’t necessarily use them, as those figure are based on several key assumptions. What I have done is some basic testing to try to model the overhead losses for power outages scenarios.
On the discharge side, there is overhead to run the basic electronics, and additional inverter losses anytime the AC is on. Based on some cursory testing, I think I’ve narrowed these to 10W for just powered-on (or with DC output on), and additional 60W for powering the 6000W inverter, regardless of load or 120V/240V.
On the charging side, we’ll assume the battery is full before any outage starts, but once the 3.8kwh is depleted, it must be re-charged from the other two sources - generator or solar. The inverter is bi-drectional, so will also have the 60W inverter overhead when charging from AC (120 or 240) as well. However, AC is not the only way to charge it, and necessarily for some of my scenarios, I need to be able to charge from the DC (solar) inputs. I’m using the following charging efficiencies:
Solar DC to DC input to battery: 100% (there are some losses, but I’m factoring the 4kwh/day from solar above as the net to battery after small converter/charging losses)
Generator AC to AC input to battery: 96% (this uses the internal inverter with 70W overhead factored into 1800W charge rate)
Generator AC to DC input to battery: 93% (via Chargeverter)
Because the Solix F3800 has only a single bi-directional inverter, it cannot charge from AC while outputting 240V to the house. Therefore, the only way to re-charge from the propane generator is by using a hybrid solar inverter, “chargeverter” or switched-mode power supply. The EG4 Chargeverter is a rather new product, but gaining huge following among the DIY crowd for charging solar batteries from a generator - inside is simply two 98% telco-grade rectifiers that can take even dirty generator power and turn it into clean DC power with overall demonstrated 93% efficiency. Cheaper would be 120V or 240V switched-mode power supplies, depending on quality could be as low as 80% efficient or less.
Next up, will be using one, two, or all three power sources together in different usage modes, and looking at the efficiency and convenience of each mode. Basically the overall efficiency of each mode results in how long my 2x20lb propane tanks will last me in hours or days, before I have to go do a refueling run. The convenience will be tradeoffs just how often I have to refuel, against whether I can run household as normal, or have to cut back to base/critical loads, plus other potential hassle factors.