Analyzing Year One
After installing solar panels on my house, I got down to the fun part: cheering for sunny days and making charts and graphs. Because what good is an experiment (hypothesis: solar panels are a good investment) without data collection and charts and graphs? I have solar panel production data, utility purchase data, and utility sale data, all at 15-minute intervals, since January 1 2017. This lets me answer a bunch of questions:
- How quickly are my solar panels paying for themselves?
- How well did my pre-install modeling match reality?
- What is my self-consumption ratio?
- What is Time of Use billing worth?
- How valuable would a big battery be?
- Could I go entirely off-grid?
- What about the environment?
So how quickly are my solar panels paying for themselves?
That would depend on how you choose to assign value. For the moment, let's assign zero value to avoided pollution, added roof protection, environmental advertising of having PV panels on the roof, or support of my local economy. And let's assign zero cost to the production of the wire, conduit, panels, racking, and inverter. We'll focus purely on the economics for the moment.
The most favorable assumption is that, without solar panels, I would be on flat-rate billing at about 13.6c/kWh. Then the value of the solar panels is my total usage times that flat rate, minus my actual costs. By that measure, the panels have been worth $1273 in 2017, putting me on pace for a 10-year payback period on my more-expensive system design, or closer to 9.5 years without the added cost of anti-rodent armoring.
The least favorable assumption is that the solar production is only worth what my avoided costs would be on flat-rate billing. That is, every kWh produced is worth that $0.136 which I would otherwise have had to pay to my utility company. By that measure, the panels have been worth $882 in 2017, on track for a nearly 15 year payback period.
That higher value assumes that my ability to switch to Time of Use billing is solely due to having added solar panels, and therefore all of the value provided by ToU billing should accrue to the solar panel project. While I don't currently have the data I would need to rigorously assess that claim, I believe that with just the electric car, I would likely have been only slightly past the cross-over point where ToU billing would save me money. And therefore, some small part of the ToU savings should be credited to the buying a Nissan LEAF project. However, the electric car has been a fabulous idea purely on economic grounds without needing the cheaper night-time electricity provided by ToU billing; accordingly I'm willing to let all the ToU value accrue to the PV solar project. So I'm calling this one a 10-year simple payback, in line with my and my installer's estimates.
How well does the yield model match reality?
In 2017, my panels yielded approximately 6.5-6.7MWh of electricity, versus a model estimate of 7MWh. My inverter monitoring reports two sets of numbers: 15-minute interval power, and daily energy. These numbers do not quite line up. On any given day, the energy yield implied by the power data is generally within ±3% of the daily energy yield reported. Summed over a month, the energy yield is consistently reported to be 2-3% higher than the power data would imply. I'm not sure which of those two numbers is more accurate. For the year, that means 2017 yielded 92.9-95.4% of the energy predicted for a typical year, depending on which set of numbers you trust.
The difference between the model and reality will be due to some combination of model inaccuracy, yearly variations in weather, and/or snow shading. The model assumes, for simplicity, that the panels are never covered by snow. Given the shallow tilt of my array, snow does cover the panels in winter, and I can only readily clear off half of them. For the Apr-Oct period, yields were 93.5% of the model prediction; for the rest of the year, only 90.1%. The model over-predicted yields more during the winter months than it did during the summer months, which gives some idea of the snow losses I'm suffering.
Looking at the monthly data, February and September were nearly 15% sunnier than expected, while the rest of the year was less sunny than expected.
What about self-consumption?
I have no particular financial incentive to maximize self-consumption; indeed, I'm actively disincentivized to do so, since the optimal scenario consists of me selling all power I generate on-peak, and buying it all back off-peak, pocketing the difference. Instead, I have done almost nothing to affect my usage patterns, with the result that I'm immediately consuming about 37% of the energy I generate, selling the rest to the power company.
Rate structures make or break residental PV
Self consumption data indicate that I'm only using about a third of my solar power when it's generated. To get any value out of the rest of it I need to either sell that power, requiring some sort of net metering or feed-in tariff, or store it myself for later use. My local utility supports net metering, where I can sell energy at the same price I would have purchased it for. This lets me treat the utility as a large zero-cost battery (zero marginal cost; I'm still paying about $20/mo in fixed charges for my grid connection) that I can put energy into during the day for use that night, or during the summer for use during the winter. In fact, I am currently allowed to combine that with Time of Use billing which increases weekday mid-day prices and decreases weekend, night, and holiday prices. That way, during the day when I'm producing an excess of energy I can sell it at a high rate, and puchase it back for less money overnight. After switching to a ToU rate, the average price I was paid for my energy rose by over 50%, from just over 13c/kWh to over 20c. Selling energy from my panels for 20c during the day and buying it back at night for 7.5c to put into my car is great!
Another option is to use a home battery to increase self-consumption. At current electricity and battery prices, this is not economical where I live. However, increasingly large parts of Australia are experiencing a combination of factors that are leading to large home battery installations: electricity prices are high, generous feed-in tariffs are expiring, and battery prices are falling. At a few cents per kWh feed in tariff, versus a few tens of cents per kWh electricity price, many Australians are investing in home batteries for purely economic reasons. Hopefully this helps drive down battery prices enough that they become a reasonable option where I live too.
What about a whole-house battery?
Not yet. Maybe soon, though.
A recent Bloomberg New Energy Finance report says average Li Ion (various chemistries) battery pack costs are down to about $273/kWh as of 2016, with dramatic price reductions in the recent past, and in the predicted future. Costs are predicted to drop to $200/kWh in 2018 or 2019, $150/kWh in 2022 or so, and $100/kWh in 2026. Based on those numbers, large home battery packs will soon become economical for electricity price arbitrage or load shifting (essentially the same thing from an economic standpoint). This leads me to believe that the current utility rate structure cannot persist for much longer: if my utility is offering me ~7c/kWh electricity at night, and will buy it back for ~21c/kWh during the day, then I can profit from a battery with a cycle cost below 13c/kWh: assuming a not-quite 90% round-trip efficiency, that 7c/kWh to get energy into the battery becames about 8c/kWh for energy coming out of it, plus the amortized cost of the battery itself. Suppose you can get 1000 full-capacity charge/discharges out of the battery during its lifetime (this would likely be composed of more like 1500-2000 partial-capacity charge/discharge cycles). Then a battery costing below $130/kWh used in this way will pay for itself; one costing less than that will generate a profit.
In other words, the current utility rate structure, with a peak-to-off-peak price spread of 14c/kWh, cannot continue for more than about a decade. Shortly before it becomes economical for me to do so, the utility will be strongly incentivised to provide its own battery storage to reduce the system demand peaks which drive up prices. And then the price differential should begin to drop, or net metering will be discontinued or heavily modified, or some other disruption to the rate structure will occur to protect utility profits.
Could I go off-grid?
Absolutely! It would be a terrible idea on purely economic grounds, though. I would need to overbuild my solar system and install a large expensive battery to cover my winter energy demands, and those assets would be wasted in the summer. My electricity demand is relatively flat throughout the year, peaking slighly in summer and winter and falling off in spring and fall with no heating or cooling demand. To match that demand curve, I should install my solar panels at a very steep angle - about 75 degrees - which sacrifices rather a lot of summer time generation to raise the winter minimum. Until solar power becomes really quite shockingly cheap, that's a bad idea.
Until then, the power company provides a nice alternative. The $230/yr in fixed charges that I pay my local utility can be thought of as a rental fee for a seasonal-storage sized battery. That is, for a few hundred bucks a year, I can harvest a ton of energy in the summertime and get it back in winter. Furthermore, paying for access to the grid gets me access to net metering and ToU rates, which more than pays for the fixed charges.
Or you could subtract the fixed service charge from my total annual bill to find out what my net energy costs have been. I've purchased about 1.9MWh for about -$40 in 2017. With ToU rates, the power I sell is mostly on-peak and the power I buy is mostly off-peak; the difference between those prices is big enough to let me buy quite a bit more than I'm selling, and still end up ahead at the end of the year. With 2-3 more panels installed, or with milder weather that reduces my annual consumption, my total yearly bill would be negative.
What about the environment?
Wisconsin's primarily coal-powered electricity generation fleet produces about 630g/kWh of CO2, which is actually quite a bit better than I would have expected. Still, that means every year I'm avoiding burning almost a ton of coal (and quite a bit of natural gas as well), keeping roughly 4 metric tons per year of carbon dioxide out of the atmosphere in the process. At nearly two years old, my system is almost net-energy-positive, meaning I've produced almost as much energy as was consumed by the production of the racking, panels, inverter, wiring and conduit, etc. used in constructing the system. This analysis is of necessity approximate as I don't know exactly how many pounds of aluminum, glass, copper, etc. went into producing the system. However, based on its overall type, the system should have a short energy payback period of approximately 10% of its useful life.
Any way you slice it, we're off to a great start!
Continued in part four, expansion.