November 15, 2023
Table of contents
While it is possible to determine the power mix for an individual consumer (the consumption mix) on purely physical grounds, it is not practical nor useful to do so. The power mix for the customers of a provider, taken as a group, (the procurement mix) is much more practical and useful. Unfortunately, California consumers are only provided the CEC mix via the Power Content Label, which involves creative accounting and magic. Our legislators have apparently caught up on that, and are trying to fix the situation, via SB 1158.
Barbara has solar panels on her roof. At a given instant in a sunny summer afternoon, those panels produce 7 kW, and Barbara consumes only 1 kW. At that instant, Barbara’s mix is 100% solar. Furthermore, she exports 6 kW and her two neighbors connected to the same transformer at the top of the pole benefit from that:
Both Alice and Cora consume 2 kW, are served by Barbara’s export, and have a 100% solar mix. There are still 2 kW going across the transformer to other consumers.
At night, Barbara’s panels produce nothing, and all three are served entirely from the grid, with whatever mix comes across the transformer.
At a given instant in an overcast day, Barbara may generate only 3 kW. With the same consumptions as before, Barbara is still 100% solar, but now exports only 2 kW. Those 2 kW go first to Cora, because Cora is physically downstream from Barbara, and Cora is still 100% solar too. But there is nothing left for Alice, so she get whatever mix is provided by the grid at the transformer.
All it took to build this picture is to measure the power at each house and at Barbara’s solar panels. We can use the same method to determine a power mix at each point in the the grid.
In general, the grid is made of power lines that connect generators (e.g. a dam with a turbine), interconnections (e.g. where Alice or Barbara are connected, or where there is a substation), batteries (in a broad sense, so including pumped hydro) and loads (e.g. houses). By measuring the power transiting on all the lines, we can compute the power mix on each one:
Note that the power mix at a point varies over time. We described the computation at an instant, and power (if any) goes in only one direction on a line. It can however flow the other way at another instant.
The power mixes we have measured for Alice, Barbara and Cora have some interesting, sometime counter-intuitive, properties.
First, they reflect all of and only the physical world. Remember that our interest in power mixes is related to GHG emissions and sustainability. Those are concerns about the physical world, and so having an accounting that directly matches the physical world is a priori the best way to go.
Second, Alice’s mix does not depend on whether she opted for SVCE (her local CCA) or for PG&E, since those entities did not appear in the computation of the mixes (and a fortiori, of whether Alice opted for a basic plan or for a premium plan).
Third, the grid topology has a great influence. Alice and Cora have different mixes in overcast days, only because they are connected at different positions relative to Barbara. And it is not just the topology near the “end” of the grid, or near rooftop solar panels that affects the mixes: the relative locations of power plants and their connections to the grid have similar impacts, not to mention the limits on how much transmission lines can carry and the control exerted by the transmission operator.
Unfortunately, this accounting implies to measure power at many different points. While there are meters at each home, and there may be enough meters on the transmission network (from the power plants to the substations), there are not enough meters on the distribution network.
We also can get non trivial differences between customers, such as Alice and Cora, for completely accidental reasons. That makes it very difficult to act. Even if Alice could convince PG&E to connect her downstream from Barbara, whatever Alice would gain would be at the detriment of Cora.
By analogy with Barbara, we can consider all the SVCE customers as a group:
Here is an hypothetical physical world that captures this point of view:
For each load serving entity (LSE) such as SVCE, all the power sources it procures from are connected together, and all the customers it serves are connected together. If a power plant serves two or more LSEs, it would be split in as many virtual plants, one for each LSE. Then all the LSEs are connected together at a single node at the top of the picture, “the grid”. When SVCE procures more than its customers consume, the SVCE mix is exactly what SVCE bought, and the surplus (with the same mix) is sent to the grid. The mix of the grid is the sum of the LSE's surpluses, weighted by the size of the surpluses. When SVCE procures less than its customers consume, the grid mix enters in the composition of SVCE's mix.
The good news is that we already measure what’s needed to compute all the mixes, because those measurements are necessary to establish payments between generators and consumers.
SVCE gets detailed data in five minutes intervals from CAISO (see slide 65 of the presentation to the September 2022 SVCE Board meeting, presumably in support of agenda item 1f on the consent calendar). That includes the amount of energy procured, and covers at least the PPAs, for which the mix is known.
SVCE receives from PG&E the measured consumptions for each SVCE customers (to establish the part of the bill related to generation). Furthermore, PG&E keeps the consumption details at least at the hourly granularity (you can download that via the “green button”).
Finally, it is reasonable to assume that whatever is needed to balance production and demand is also well tracked, since SVCE needs to buy/sell it.
The bad news is that SVCE does not care to compute a mix along these lines, or at least does not care to publish it.
This seems very strange given that the mix, and even more so the companion tracking of GHG emissions, should be a key metric of the progress of SVCE toward fulfilling its mission (“providing carbon free [...] electricity”), and that SVCE is a public agency.
This LSE procurement mix accounts for the excess energy generated by Barbara (it is included in what the LSE procures) but not for the energy generated and consumed by her right away. We can compute her individual procurement mix by combining the mix she gets from the grid and the (100% solar) mix she gets directly from her panels. And we can also account for a battery at Barbara’s house, just as we did for a battery on the grid. In fact, that amounts to treating Barbara and her panels, battery etc. as a subdivision of her LSE.
By law, California LSE’s must make an annual “Power Source Disclosure” to the California Energy Commission (CEC), listing all the wholesale purchases (including fuel type, net MWh, plant id), and the MWh of retail sales. The CEC provides a template as an Excel spreadsheet (e.g. for 2021) that the provider fills and returns to the CEC. From those data, we get a “Power Content Label” that lists: the proportion of energy from renewable sources, the proportion of energy by fuel source, and the average CO2e emissions per MWh; those emissions are actually computed in the spreadsheet, so we get a chance to understand the details of the computation. We also have the CEC’s Power Source Disclosure Resources for Retail Suppliers that provides guidance on how to fill the spreadsheet.
The Power Source Disclosure must be certified by the provider. For Alameda Municipal Power (AMP), that certification comes from the city’s Public Utilities Board, and we can see the report prepared for that purpose, which contains the full Power Source Disclosure. I have reproduced this PSD using the CEC template, so that we can explore their numbers (there is one additional tab, labeled “Eric Muller”, for that purpose).
Here are the important numbers:
| MWh | % total | % reported in PCL |
|
| Wind | 20,277 | 6.02 | 6.20 |
| Solar | 169 | 0.05 | 0.10 |
| Renewable hydro | 2,493 | 0.74 | 0.80 |
| Biomass & biowaste | 77,096 | 22.90 | 23.40 |
| Geothermal | 162,267 | 48.19 | 49.30 |
| Large hydro | 67,038 | 19.91 | 20.40 |
| Natural gas | 7,351 | 2.18 | 0.00 |
| Total bought | 336,391 | ||
| Energy used by the city | -7,154 | 2.13 | |
| Losses | -14,122 | 4.19 | |
| Energy sold to customers | -304,140 | 90.41 | |
| Energy unaccounted for | 11,275 | 3.35 |
Magically, the 7,351 MWh produced by burning natural gas have disappeared from the Power Content Label, and the GHG emissions on the PCL (117) similarly ignore the emissions from burning the natural gas (they would be 126 otherwise).
The explanation can be found in the CEC faq:
Why does the PSD Annual Report template adjust certain net specified procurements?
Retail electricity suppliers typically buy more power than they need to meet their retail demand (also called retail sales). They need to, among other things, cover for their own electricity consumption as well as transmission and distribution losses. The Power Content Label is required by statute to only reflect electricity serving retail sales, which means some power may be left off the label. For consumer transparency, these adjustments to net specified procurements are shown on schedule 1 on a PSD Annual Report to indicate that a retail supplier has additional power sources that are not represented on its Power Content Label.
The AB 1110 rulemaking updated the rules for this adjustment. Under the new rules, retail suppliers apply their low-GHG resources (eligible renewables, large hydro, and nuclear power) to retail sales first. If the retail supplier procured more electricity than it needs to serve retail sales, then its fossil fuel resources are reduced to account for the difference between net specified procurements and retail sales of an electricity portfolio. All natural gas resources are proportionally reduced first, then coal and other fossil fuels as needed (if total adjusted procurements still exceed retail sales, then all remaining fuel types are proportionally reduced).
We can indeed see this method of computation at play in the spreadsheet for AMP: there is an excess procurement of 336,391 - 304,140 = 33,551 MWh (sheet Schedule 1, cell N10). The “Net MWh Procured” for the natural gas buys (rows 46, 47, 58, 74) is 7,351 MWh, so that is completely erased: the “Adjusted Net MWh Procured” is 0 for those rows. And since there is still some excess, the adjusted net for all the other rows decreases a bit. Similarly, the 117 lb CO2e/MWh only reflect the biomass and geothermal emissions.
The result could be worse: instead of natural gas, AMP could have procured 33,551 MWh of coal, which at ~2,000 lb CO2e/MWh would have produced 67 million pounds of CO2e, and still claim 0% coal and no emissions for coal in the PCL!
I can’t imagine any rational basis for this way to computing the power mix. The city of Alameda is a consumer just like the others, and the electricity it uses is no better and no worse than that used by its citizens (not to mention that the city consumes on behalf of its citizens). The losses in transmission and distribution are inherent to the production of electricity, and there is no basis to say that they are the result of natural gas more that from wind or anything else. Energy consumed by a plant (either self generated or coming from other plants) is just like losses. Even if you see the PCL as a measure of the flow of dollars from consumers to producers (although expressed in MWh), it does not make sense to say that one’s dollars served for the “good” kWh; they are paying for the “bad” ones too.
It would be interesting to know if this quirk came from the legislation that instituted the PSD/PCL, or from the interpretation and implementation by the CEC, so as to avoid a similar nonsense in the future.
The PCL computation has another quirk. From the FAQ:
Why are the GHG emissions of certain firmed-and-shaped products excluded from the emissions intensity of an electricity portfolio?
Retail suppliers can exclude GHG emissions associated with eligible firmed-and-shaped products from the Power Content Label if they were procured under a purchase agreement executed prior to January 1, 2019. This provision provides good faith relief to retail suppliers that entered into firmed-and-shaped contracts prior to the development of retail GHG emissions accounting rules.
Of course, it is probably a good idea to not penalize suppliers who acted in good faith, based on rules then in effect. But that does not change the fact that energy sold under those purchase agreements may generate GHG emissions, and that consumers deserve to know about those emissions. It should not be that difficult to compute a number under the old rules (e.g. for regulatory and time series purposes), and a number under modern, realistic rules for information purposes.
Last year, our family consumed 4,793 kWh. SVCE and PCE tell me that my energy is “clean”. English is not my mother tongue, but I understand that to mean “the generation of those 4,793 kWh resulted in 0 lb of GHG”.
As far as I know, there is currently no electric generator that consumes GHG, i.e. has strictly negative emissions. That would solve a lot of our problems, but the best we have found is 0.
Putting the two together, it follows that any portion of those kWh has to result in 0 lb of GHG. In particular the portion generated in any given hour.
Yet, SVCE admits that there are many hours where there are GHG emissions (but does not care to quantify that in any way) (see SVCE case study, page 4; all the darker than yellow pixels in the heat map correspond to hours where there are GHG emissions).
PCE similarly states “Peninsula Clean Energy’s average hour-by-hour emissions intensity for 2021 was 222 lbs CO2/MWh (compared to 5 lbs CO2/MWh on an annual basis).” (in 2021; see Achieving 24/7 Renewable Energy by 2025, page 7.). So with PCE, my 4,793 kWh generated either 1,064 lbs or 24 lbs. Which is it?
Neither PCE nor SVCE clearly explain that magic. I can only think of buying some kind of offsets (e.g. RECs) and applying them to dirty electricity (i.e. washing that electricity). Or equivalently, buying more clean electricity than needed, reselling the excess without its clean attribute (essentially generating RECs), and applying this clean attribute to dirty electricity.
In any case, the fact is that if we had not consumed those 4,793 kWh, there would be 1,064 fewer lbs of CO2 in the atmosphere, so 1,064 lbs it is. Hard to argue with physics.
And for the sleight of hands to work, somebody else has to be willing to buy dirty electricity. That may be possible today, but when the music stops (when we achieve zero emissions from electricity production), somebody is going to end up without a chair. This acccounting is simply not scalable and sustainable.
The good news is that the California legislature recognized that the current practices to account for GHG are counter-productive, and passed SB 1158 Retail electricity suppliers: emission of greenhouse gases to address that. Senator Becker, the author of the bill provided this argument:
SB 1158 is about measuring outcomes against targets. We’ve set greenhouse gas targets for all of our electricity suppliers for 2030 and later years but we have nothing in place to measure how well they are doing against those targets. How can we possibly expect that to work? This bill fixes that gap. It directs the CEC to establish rules for electricity suppliers to analyze their sources of electricity and report on the associated greenhouse gas emissions so that we can measure progress and hold everyone accountable to doing their fair share to reduce emissions.
Let’s hope that the CEC will implement this bill as intended. And let’s hope that our CCA’s will not wait until 2028 to become accountable.