Rob's Experience with Solar Power

Not All Power is Equal

Since 2002 I have benefited from Net Energy Metering (NEM). My first solar system in 2002 was on an NEM-1 rate protocol. Under NEM-1 all electricity was equal. You were charged a premium for any excess you used. Thus the goal was getting a large solar system and forcing your total usage as low as possible.

PG&E began throwing cold water on that idea about 10 years ago when they started charging minimum amount each month for power - use it or lose it. Thus in months when I generated an excess I was charged the minimum fee. I still got credit for my excess power, but the fees added up.

In 2015 I found I could no longer stay on the NEM-1 system. I had just leased a Volt that could easily suck in 10kWh every day. The old rate system was simply not viable and I had to go to Time of Day (TOD) rates. Fortunately PG&E offers an EV-A rate. This was very favorable to charging my car and it had an interesting effect on the solar generation. The other time of day rates have limits on usage or high off peak rates that make them impractical for electric car use. With EV-A offers a very low off peak rate at the price of crushing peak rates and introduction of a weekday rate period.

The real thing thing that distinguishes the EVA rate from the E-TOU-A/E-TOU-B rates is the tradeoff of a steep Peak rate for a discounted Off Peak rate. This generally works in my favor. Five days a week my solar generation is a combination of Part Peak and Peak. Only on the weekend is most energy lost to Off Peak.

Several years ago PG&E convinced the state to end the NEM-1 rate protocol. I would be grandfathered for 20 years (until 2021), but would eventually be on a NEM-2 TOD rate system that is less favorable to solar than EV-A. So in the end moving to EV-A was to my benefit. It definitely is with the new system

How My System Performs

Now let's get to how my system has performed. I will present this multiple ways. Since in the previous section we learned that all power is not equal, I will present the purchased vs consumed comparisons by dollars. In my first table I will present the monthly results in raw kWh for each rate period for each month

Note that the first April time period begins on 3/27 (when I received by Permission to Operate). The other time periods run from about the 13th of the previous month to the 13th of the listed month.

A Summer Day

In this section we will go into detail on a particular summer day to show the interplay of electric car charging, solar generation, and AC use. I selected Friday June 22, 2018 since I took a long trip the previous day, used the car on the 22nd, and the maximum temperature was 89°F so the AC was required. While it was hotter on Saturday, being a weekend distorts the discussion. In 2018 I did not find any long stretches of hot weather when I was also using my car so it does show a case of where the AC really struggled, but one has to go with the data available.

First let's look at the temperature profile. It was relatively cool overnight. It started warming in the early afternoon, but was not hot until the peak electric rates kick in. The temperature hit its max at 4 PM. At that point it started cooling, but outside > inside prevailed until at least 9 PM. The inside temp at that point was also increased by radiation from the attic.

Finally let's examine how the bill for the day was determined and how it compares with what I would have been billed without solar. For a detailed look at the day here is the spreadsheet

So that was a hot summer day.

• The car and AC dominated the Off Peak period
• A significant portion of my solar generation is consumed behind the meter and therefore not visible to PG&E
• The solar system sold power during most of the Part Peak giving me a large dollar credit. This continued into the peak
• The AC usage (on this day) was in the early evening as the high outside temperature eventually overwhelms the house's insulation
• As 11 PM comes and the next Off Peak begins the car and the AC dominate

What Ifs

Models of Different Alternatives

Since I was allowed to turn my system on in March 2018 I have maintained a spreadsheet of hourly power purchases/sales as reported by PG&E and hourly solar generation as reported by SunPower. This vast wealth of data allow me to model several alternatives that I could realistically face. This section will talk about the various model results. Not to bury the lead, but the most favorable outcome for me is the current NEM 2.0 system with my current EVA rate schedule. This should not be a surprise since my choice of solar system was based on my having an electric car and for NEM to time shift the credits (not kWhs) to make up for costs that occur at night when the electricity is cheaper.

Current System

The first model is real life. These are the costs I am actually seeing. The cited figures here and in later sections include Non-Bypassable Charges (NBC) where applicable but do not include taxes. I also have minimum costs but these were ~$300. Thus any model that produces less than a $300 charge just means that normal billing becomes irrelevant and my final bill will be $300

The following table provides some scale to my electrical use.

No Solar

In this model I do not have a solar system. Instead I am buying power directly from PG&E. Under this scenario there is no NBC because the meter always runs forward. Calculating the no solar scenario is actually quite easy. Since I know how much power was generated and how much I bought it is simple math to determine how much I used.

Reduced NEM

At the present time I can sell power to PGE for almost the full retail price. It is not the full retail since I am charged NBC on each kWh sold. In this scenario I assume a Nevada like scenario where PGE drops the sale price to 12¢ during all hours.

I also ran the numbers assuming NEM was 50% of the current rate.

Both of these include NBC which might be eliminated in a reduced NEM scenario. NBC accounts for about $100 of my annual bill.

Batteries

Let me first start by saying that my situation is not ideal for storage. First my electric car uses a huge amount of power that I consume mostly at night. My solar system only produces a max of about 34kWh a day in spring. A long drive by itself requires 30-50 kWh. Anything used by the A/C, cooking, dryer, etc is extra. So there is no realistic way of not using a lot of purchased power.

Second, Under NEM the credit received from selling 1 kWh of power produced at on a Weekend at 11 AM will only buy 1/4 kWh at 4 PM. Time shifting is a huge win. On a weekday that same 11 AM kWh buys 1/2 kWh at 4 PM. Time shifting will benefit but less so. Finally any NEM credit for selling any remaining 11 AM power could be used against a purchase at 11 PM of 2 kWh of power. Here time shifting costs me money since I only replace the need kWh with one captured earlier when a higher NEM reimbursement was available (currently). Thus merely time shifting kWh is not always a win. Time shifting my expensive daytime power to charge my car is positively silly. Using excess NEM credit is not.

Finally, I sized my solar system assuming these NEM credits. If my goal was limited to replacing my peak power usage I would have bought a 3kW system instead of a 4 kW.

The storage model attempts to track how a battery will perform during this year of service based how I used my current electrical system. The model uses a 30kWh battery. I have tested other sizes, but this seems to be the sweet spot. The model makes three crucial assumptions.

• I must be able to specify certain hours that the storage system will be in control. During those hours the system will attempt to balance the solar generation and use by charging/discharging the battery. If the battery is full it will sell power. If the battery is empty it will buy power.
• It must be able to be programmed to force charging during specified hours. This is critical during winter.
• The model assumes a battery that can run to 0 and then be recharged with 100% efficiency. Batteries such as this can be obtained in the same store that sells the massless, frictionless pulleys we used in Physics 101 😉.

I have assumed that I can only change the programming once per bill cycle. I have tried some micromanagement and it does improve the numbers, but largely it is a waste of time. In real life changing the parameters, especially winter charging, may be required due to the fallacy of the previous bullet. In the spring because I have so much solar the battery picks up a portion of the car charging (I could have upped my sales, but I tried to minimize my NEM). That actually increases my costs since daytime power has more value. Later in the year I schedule car charging when the battery is not connected.

Results

From these numbers some conclusions could be drawn

• As long as grandfathering applies I am not motivated to add storage. I ran the numbers with NEM 2.0/EVA and I would be paying the minimum (about $200 less than now). Remember though that in addition to the electric cost listed above another $1,000 per year of amortization should be included in the calculation. Now if the state did allow NEM to be modified I might eat the amortization just so I could reduce the amount of NEM sold and the amount of electricity I purchased (and minimize the Utility's gain).
• For others the availability of a generous NEM will also be a disincentive to installing storage.
• Since EV-A will soon no longer be available my analysis has its limitations. I did not pursue a non electric car estimate since I felt it would be too much of a SWAG. However, I can see some general conclusions. A storage user without an electric car and with a properly sized solar system could expect that their bills would be close to the minimum connect charge (for PG&E about 1 kWh of peak sale per day). E-TOU-x also eliminates the Partial Peak period. Daytime and all day on weekends are off peak. Thus time shifting would result in either a benefit or use-equivalent-to-generation event which makes the economics of a battery more favorable. I would expect that they would have minimal NBC and that the generation minimum would be met since in such a case there would be less repurchase of NEM sold power. It is even possible that the extra cost of installing a second meter could be amortized since a smaller solar and battery system would be needed.
• Forcing electric car owners to charge under E-TOU-B would double their charging costs and discourage electric car adoption. Thus even though EV-B requires more up front expense it is preferable to operating the entire house/charger on E-TOU-B.
• Reducing the NEM is not fatal, but it would increase the payback time and thus fewer people may be willing to install solar voluntarily. My payback with the current NEM 2.0/EVA is 11 years. With a 12¢ sale price it increases to 17 years.
• A fixed fee like the former SMUD Grid Access Charge discourages both solar and solar with storage.
• When I ran the numbers for the storage model I was initially surprised by how much power I sold. Then I checked when this happened and realized I was on vacation both times my electricity sales shot up. Since all battery sizes are finite I must conclude that adding a battery does not eliminate the issue of NEM. I did reduce the solar I sold from 4,688 kWh to 1,813 kWh which would reduce the "solar impact". The electricity that I did sell tended to be mostly during the Peak so it could be immediately used. Of course, during vacations my battery tended to always run full and thus power was sold when generated.
• Assuming a fixed 12¢ NEM as in my storage model my assumptions of programing the storage system and then making minimal changes to the programing yield satisfactory results. Winter is when more micromanagement might be required since one has to balance replacing peak usage with the occasional sunny days.
• If I received no NEM as a hypothetical storage owner then I would have to be more aggressive than my current model to make sure I use it all. Someone with a smaller solar system that could consume all of the power they generate might have to write off vacations, but otherwise this might not matter. For my current system having the NEM reduced to zero would likely mean my spending a lot of time in Sacramento.
• Reducing NEM would likely dissuade some people from installing a stand alone solar system, but it would have less effect if the customer also installed storage.
• Even in the worst case solar wins.