One of the problems we have is that our job is not the typical situation that most solar installers face. Most people who get solar don't have any in the first place, and so the solar installer can take a year's worth of PG&E electrical usage data and come up with a good estimate about how the homeowner can eliminate 80% of their PG&E bill by installing solar. At which point, the homeowner is thrilled because the estimate shows them paying PG&E far less. In our case, we already have a solar PV system, so our bill tells the installer little about our electrical use. We still draw about 2000 kwh from the gird, but we pay nothing for that power because PG&E credits us around 3x for power generated on summer afternoons what we pay for power at night, when we are recharging our converted plug-in Prius, and in the winter, when we have lots of lights on and the furnace runs. We never use up the surplus credit, though we have been coming closer since we converted our 2008 Prius to a plug-in hybrid. We also are also primarily interested in completely offsetting the carbon footprint from our electricity usage, not in eliminating or reducing our electric bill, though of course that should follow if we completely offset the electricity usage.
In addition, we are having some new electric appliances installed, and are planning on buying a Nissan Leaf electric car next year. So we need somehow to plan a system that will offset our current grid draw and, in addition, offset the usage from the new appliances and the electric car if we want to eliminate our carbon footprint. Estimating the usage from the appliances is straightforward, but with the car, the electrical usage is likely to be somewhat more unpredictable. However, I came up with the following projected usage requiring offset in addition to the power provided by our current 2.5 kw system:
- 2000 kwh/year - current grid draw
- 360 kwh/year - HRV system used for 6 months in winter when the windows are closed, 24 hours per day
- 633 kwh/year - on-demand electric backup hot water heater. This was calculated by using the historical energy use from our gas-fired hot water heater, correcting for the (in)efficiency of gas-fired tank hot water heaters, and assuming that the backup heater would be needed for 6 months in winter.
- 2000 kwh/year - Nissan Leaf charging. This was calculated assuming 8,000 miles per year at 0.25 kwh/mile. We typically drive our commuter/around-town car about 8,000 miles per year and 0.25 kwh/mile is a standard figure that a reasonably aerodynamic electric car should achieve (it is slightly more than what Nissan assumes, since the Leaf has a 24 kwh battery and is supposed to get around 100 miles on a charge).
My original plan was to add a 30% margin to that to allow for power usage growth in the future. For example, if we revisit the decision not to install a geothermal heat pump, or even maybe go for an air source heat pump, we would then remain carbon neutral without having to install new panels. However, I found out near the end of the design process that PG&E won't allow solar installers to put in more panels than will completely offset the homeowner's electricity use, without good reason. Theoretically, this would not have allowed us to offset the new appliances or the Nissan Leaf because they weren't on our bill last year. REC and the other companies bidding on the contract kept sending bids for 18 panels, or 26 panels, until finally the REC engineer explained the problem. When I told them about our situation, the REC engineer accepted my explanation - that we were buying an electric car - as sufficient justification for a larger system. However, they would not let me include a 30% margin. This is unfortunate, because I was planning to offset the carbon generated by our gas hydronic furnace, gas stove, and new gas fireplace from the portion of the 30% margin that we didn't use. I guess we will just have to continue to buy carbon credits for that, as we are doing right now.
I also originally requested Sunpower E19 modules which generate 315 watts/panel, thinking that more watts per panel would reduce the number of panels and therefore the amount of space on my roof taken up by the PV system. However, Christine pointed out that the E19 was actually larger than standard modules, around 10" wider, though the same length. So, in effect, we would end up using about the same amount of roof space with the E19 as with a set of standard panels that generate 235 watts/panel, the maximum currently available in a standard-sized residential panel. REC also advised not to go with Sunpower, since they are a name brand and they charge extra for the brand. They instead recommended the Kyocera 235 watt module, the KD235GX-LFB. This model produces exactly the same amount of power as the Sunpower E18 series (which is standard sized) but is cheaper.
My original calculations, including the 30% margin, came up with a total system size of 6.85 kw DC. However, I failed to account for shading and losses due to AC conversion. When REC calculated the size of the system, it was 7.050 kw DC without the 30% margin. In either case, the number of panels is 30. As a practical matter, this is probably the maximum amount we can fit on the part of our roof that gets sufficient sun year round. You can see the problem in the picture below, which was screen-copied from the satellite photo of our roof from Google Maps (taken before the solar thermal system was installed on the east roof):
The next decision we needed to make was about inverters. An inverter changes the electric current from DC to 60 Hz AC (usually around 440 volts), which is compatible with the grid. The conversion process is relatively inefficient, losing around 12-15%, which is why REC's design calls for more DC power than my initial estimate. Our old system has a centralized inverter that sits on the side of our house, here you can see it:
When we had the old system installed, centralized inverters were the only technology available. Centralized inverters require the solar panels in each row to be wired up in series. The output side of one panel is connected to the input side of another and the entire system is then manually balanced so that the voltage seen by the inverter is the same for each row. The problem with this design is that shading, dirt, and irregular aging of the panels can cause the voltage from one or two panels in a string to drop, either transiently or permanently. Like a string of Christmas tree lights in which if one light goes out, they all go out, the entire string of panels is then cut out and the system loses power from all the panels in the string. The result is that the system can lose somewhere around 15-20% of total power output during a shading incident or permanently if the problem is due to irregular aging.
Recently, new technology has been developed to alleviate this problem. There are two different approaches:
- Microinverters - rather than having a centralized inverter, each panel has a separate microinverter on the panel and the electricity coming off the panel is AC instead of DC,
- Maximum Power Point (MPP) balancing - the electricity coming off the panels is still DC, but a control unit on the panels adjusts the voltage and current so that the inverter always sees the right combination to maximize power from the array.
MPP balancing systems use a centralized inverter, but each panel has a small control unit on it that measures the voltage and current. The control unit constantly adjusts the voltage and current so that the inverter sees exactly the right combination to maximize the energy generated by the system. The MPP balancing system can also determine whether the panel is experiencing accelerated aging or otherwise needs replacing, and report this through a Web interface (microinverters have this ability too). The capacitors in the control unit don't have to be very large and therefore don't need to be electrolytic. So MPP balancing systems typically have a longer warranty, usually 20 years. Still not the same as panels, but better.
When I was originally considering 315 watt panels, I discovered that nobody made a microinverter which handled more than 230 watts per panel, so I requested an MPP balancing system. Residental MPP balancing systems are not parametrized by the panel power rating but only depend on what the centralized inverter can handle. Tigo is one manufacturer of MPP balancing systems. Because MPP balancing systems are used more in large, commercial PV installations where ease of installation by relatively untrained personnel is not an issue, very few residential installers have much experience with them, so they either won't bid them or they have lots of reasons why such systems aren't as good as microinverters. I requested that Christine call up Tigo and another MPP balancing manufacturer and ask them for recommendations about solar installers that know how to install MPP systems.
After I found out about the increased panel size for the 315 watt panels and decided to go with the smaller 235 watt panels, I briefly considered switching to microinverters. But in addition to the issue of the short warranty, microinverters may have a thermal problem. The temperature under a solar panel in the summer is usually very high, maybe well above 100F. The inversion process itself generates heat, and the electronics become more inefficient as the temperature rises. Since a centralized inverter can be positioned where it is in the shade (as it is on our house), a lower temperature can be maintained. MPP balancing systems therefore have the potential to be more efficient in summer, when the PV system is generating the most energy. Because they don't handle such large amounts of electricity, they don't generate as much heat from the electronics. As a result, I requested that we stick with an MPP balancing system. REC has experience installing Tigo MPP balancing systems, which is another reason why we selected them.
REC's final projection was that the system would offset 101.65% of our electricity use and 90.04% of our bill. It's impossible to offset 100% of the bill because PG&E charges around $7/month for the privilege of connecting up to their system. Here's a table of REC's projections about our power use, solar production, current (actually projected) utility cost without solar, and projected utility cost with solar:
The chart shows that we only need to pay for electricity in January and February. In the other months, only PG&E's connect charge appears. The calculation isn't strictly accurate because the estimated usage is the same for each month. Since we'll only be using the on-demand hot water heater and HRV from October through March, the usage for those months should be higher, while the usage for the months April through September should be correspondingly lower. I recalculated the estimated usage and estimated utility charge with solar and came up with the following table:
My assumptions were:
- The 993 kwh/year usage from the HRV and the backup hot water heater were removed from the summer months and distributed evenly over the winter months,
- The 165.5 kwh/month thereby gained in the summer months were sold to PG&E for $0.28/kwh, the summer afternoon net meter tariff,
- The 165.5 kwh/month used in the winter were bought from PG&E for $0.10/kwh, slightly higher than the current PG&E winter tariff.