September 29, 2024
Home battery storage offers multiple benefits, such as savings (in particular in conjunction with solar panels) and resiliency. What is less clear is the CO2 benefit, if any. A priori, the idea is to store energy generated from solar panels during the day, and to use that energy in the evening and at night, to avoid burning fossil fuels. But...
Of course, there is the footprint of manufacturing, delivering and installing the battery at your home, as well as the additional equipment it needs (e.g. an inverter). Let’s leave that one aside for now, and look at operating emissions.
Compare the scenarios where you decide to charge or to not charge your battery. The difference is an additional load on the grid when you charge. And that means additional generation, since the grid must be balanced at all times. If you have solar panels, you don’t turn a button so that they generate more, they are always producing the maximum they can, whether you charge or not, and the same is true for all rooftop solar. So in all cases, charging your battery means that some commercial generator on the grid produces more. Similarly, when you discharge you battery, that means less load on the grid.
The next question is: which generator, and what change to the emissions (increase if charging, decrease if discharging)? This concept has been formalized under the name “marginal emission factor”, which is a quantity that varies over time. Furthermore, the generator that is sollicited may change depending on the location of the additional load: may be there is congestion on the grid that restrict the possible generators, and that gives us “locational marginal emissions factors”.
In California, the selection of generators given the forecasted load is done via markets (i.e. a set of rules and procedures) that attempt to minimize the overall cost to the consumers. Ideally, we would determine the emission factor by running the market with and without the load of charging the battery, and without or with the load served when the battery discharge, and compare the emissions. That would require having access to the market software and to all the data that is fed into it (offers from generators, including prices, quantities, etc.). Even in this day and age, this is considered confidential data (search for “subpeona” here), so we’ll need something else.
There are a couple of commercial offerings that attempt to determine the marginal emission factors by more indirect means. WattTime is one of them, and they kindly offer some historical data. The bottom part of this chart, in orange, shows the marginal emissions in lbs/MWh, for CAISO North:
Smart:
Timer:
If you go through the various months, you will see that the marginal emissions are pretty much bimodal: either the additional energy comes from natural gas with ~900 lbs CO2/MWh, or it comes from an essentially GHG-free source. You will also notice that from mid-June to mid-August 2024, as well as in early September 2024, the margin is pretty much always provided by natural gas: the demand was high enough to absorb all GHG-free sources, and then some. Similarly during the winter.
The top chart shows two modes of operation of a home battery. In the timer mode (pink), the battery is charging from 10am until full, and discharging from 9pm until empty. In the smart mode (blue), it is charging when the marginal emissions are less than 200 lbs/MWh, and discharging when they are more than 800 lbs/MWh. The battery has a capacity of 13.5 kWh, max charge 5 kW, max discharge 11.5 kW, losses 10% (ie Tesla Powerwall 3 specs). The box below show the emissions generated by charging, those avoided by discharging, and the net avoided (negative) or caused (positive) emissions. By design, the smart mode never increases emissions. The timer mode can result in net additional emissions: that happens when the losses dominate, e.g. in July 2024.
The computation of marginal emissions factors is indirect and very delicate. Note in particular that there is no “truth” that could be used to fine tune a model. I am a bit dubious about WattTime’s numbers, for the following reason: During the last 12 months, the minimum generation from natural gas has been about 1.9 GW. Let’s say that is the minimum for the proper operation of the grid, and that when the production from natural gas is above 2.1 GW, it could be reduced. There are literally only a handful of hours below 2.1 GW, so there is practically always natural gas that could be displaced. And since it’s not, that suggests that natural gas is the marginal source pretty much all the time.
Definitely a wild guess, but if it is right, then the July 2024 situation is the norm. And if it is not, then it is just a matter of adding enough batteries to absorb the GHG-free sources, and then the next battery will increase emissions.
It is worth remembering the limitations of marginal emission factors (really, all “marginal” accounting). One is that they work only for small variations from the base case; one cannot take the answer for the first additional battery and apply it to all subsequent additional batteries.
The other is that they assume “all other things being equal”; yet, if enough folks add home batteries, that could lead somebody to build a new utility scale solar generator. There is yet another concept: “long-run marginal emissions”. That computation accounts not only for the change in load, but also for changes to the grid itself (e.g. new generators or transmission lines), whether they are related to the additional load or not. NREL offes some data on that.
Of course, this discussion concerns only today’s situation. There is no question that some form of batteries, with significant capacity, is necessary to reach zero fossil fuels, let alone zero emissions.
The discussion so far was centered on home batteries, but it applies just as well to utility-scale batteries: if you charge with dirty electricity, you don’t decrease emissions. Yet there is a significant deployment of utility-scale batteries. The industry is not known for its environmental concerns, not for its foresight, so there must be some immediate financial interest. But what is it?
Again some wild guesses. If you know better, please let me know.
Batteries can increase their output almost instantaneously, so they are useful to balance the grid. Traditionally, hydro was the mechanism to react in a matter of seconds, but it is becoming less dependable.
A solar generator with a constrained connection to the grid (or in an area of the grid with limitied export capacity) can smooth out its production. Instead of requiring x MW of transmission during the day and 0 during the night, it can require x/2 at all times.
Similarly, instead of building a natural gas generator that produces x MW at night and 0 during the day, you can build one that produces x/2 MW at all times. Besides the lower capital cost, there is a lower operating cost, as starting and stopping a natural gas generator is not free.
Or, equivalently: if you have a natural gas generator that produces x at night and 0 during the day, and you need an additional x at night, then add a battery and run the existing generator during the day, instead of building another generator that would also run only at night.