Skip to content

The Changing Economics of Electricity Supply Reliability

What happens when Californians take reliability investments into their own hands?

(Today’s post is co-authored with Duncan Callaway)

It’s easy to take your power supply for granted … until it’s gone. Here in Northern California, recent power outages have demonstrated just how disruptive it can be when the grid goes down for an extended period of time. 

Not again!

So far this year, millions have had their power supply interrupted, many for days at a time. As of now, there are no more public safety power shutoffs (PSPSs) planned. But we’re not out of the proverbial woods. Warming temperatures, drought, and millions of dead trees have drastically increased wildfire risk in California. Keeping power flowing during wildfire season is becoming an increasingly risky business.  

It’s estimated that it will take a decade or more to make the investments needed to provide grid power safely and reliably through wildfire season. As our household comes to terms with this new normal, we’re weighing our options. Everyone’s talking about solar plus storage. Home-generator manufacturers’ stocks are soaring. This week’s blog digs into the changing economics of electricity supply reliability.

A sleek and not-so-sleek backup solution

Reliability for Sale

Before we dive into the details, keep in mind that these calculations depend a lot on how much electricity a customer uses and how much of that consumption they want to maintain in an outage. We’re going to use our household as an example, but your mileage may vary.  

As for us, we are below average consumers. Our most important appliances include: Our espresso shrine, our internet and laptops, our refrigerator. Add some lights and the occasional laundry and we’re at just under 4 kWh per day for these important loads. What would it cost to keep these services running when the grid goes dark? We’ve looked into two options:

Option 1: The Shiny Object

The Tesla PowerWall + Solar PV

As a lot of people learned during the recent California blackouts, a solar photovoltaic (PV) system without storage won’t supply your loads during an outage. Folks that have PV on their roofs do it because they want to reduce their environmental impact and because net metering is a really good deal (for them), but not for reliability. 

This is where storage comes in. Storage inverters can “island” a house, so the lights stay on during an outage. If you couple that storage with solar, the PV can replenish the energy you draw from the battery. In our analysis, we just considered the incremental cost of a storage system because the private economics of PV pencil out on their own for many households.  

There are a number of home energy storage systems out there. But we couldn’t resist using the sleek Tesla Powerwall for our test case. Here in California, Tesla is offering a $700 battery discount to anyone affected by wildfire outages. Given our electricity usage, one Powerwall battery (13.5 kWh capacity) would be more than enough. Tesla only sells the Powerwall direct to customers with PV, so to get our system quoted we also priced out the smallest PV system Tesla offers – 3.8 kW – which is all we’d need. 

The Powerwall battery would cost us about $6,700 out of pocket. To maximize the returns on this investment, we’d make sure to fill the battery with sunshine during off-peak hours and sell back to the grid at peak prices. Assuming a round-trip battery efficiency of 90%, this arbitrage play would earn us almost $400 per year on our current E-6 rate. However, we’ll be pushed onto the E-TOU rate in 2022 which will reduce the arbitrage earnings to only about $60/year. What a difference a rate makes!  

Once we’ve netted out the arbitrage value of storage (over a 10-year horizon on the new rate), we are left with an annualized storage investment cost of about $725 (using a 4% discount rate). That works out to $150 per day – if we assume the grid will be down 5 days a year – to stay caffeinated, refrigerated, illuminated, and charged. This feels like a lot to us. To put this in some perspective, we’d need a value of lost load (VOLL) of around $40/kWh to rationalize this cost. That’s over a hundred times the retail rate, and several hundred times more than the marginal cost of grid electricity.  

Option 2: The Workhorse 

Honda generator

The second alternative we considered is a gas-powered generator. No clever arbitrage plays. No sleek storage gizmo in the garage. But $1,200 gets you a 2,200W generator. Add a $500 transfer switch and you’re good to go.

Using numbers from the Honda website, we worked out that the generator is about 6% efficient. (Ouch!) Assuming a gasoline price of $4/gallon, it would cost us just under $2 to generate one kWh. Feeling conscientious, we add $0.80 per gallon to account for the pollution impacts.

Taken together, this workhorse option works out to less than $50/day (assuming 5 outage days/year). We can rationalize this with a VOLL of $13/kWh. That’s much lower than the battery, but still much higher than retail prices.

What’s an Energy-Geek Family to do?

At $50/day, the gas generator looks like the winner. But here’s where the behavioral economics in our household get interesting. 

Having worked through these calculations, one of us is inclined to keep it simple with flashlights and some extra ice in the freezer. There are mounting concerns about the air quality implications and fire risk of a growing number of backup generators firing up when the grid goes down. up. Given the financial costs, hassle costs, and the idea that we’d be contributing even incrementally to local air pollution, a generator doesn’t seem to be worth it.  

The other remains enamored with that sleek Tesla Powerwall. Energy storage is the future. And when we finally replace our beat-up Prius with an electric car, we’ll be willing to pay more for reliability. Plus, wouldn’t it be fun to host the neighborhood Powerwall Potluck when the grid goes dark?

One of our takeaways, then, is that valuing reliability is a tricky thing when customers take matters into their own hands. The technology we use to keep the lights on has other attributes that could affect the path we privately choose. 

The Changing Economics of Reliability Investments

Setting aside our intra-household deliberations, these calculations also have us thinking more about the big picture. For larger consumers, or a commercial business, these back-up solutions could make more sense. Our back-of-the-envelope calculations suggest that if the generator is used to its full capacity 5 days a year, it pays for itself at a VOLL of well under $5/kWh. That’s half of what others have used to value the electricity consumption lost in an outage (see Catherine’s excellent blog on this topic).  

If lots of customers start to make these investments, some thorny concerns could arise. The going-forward social cost of power outages could be significantly reduced if the most valued load has been backed up behind the meter. Would that make utilities less inclined to invest in costly grid upgrades? If the answer is yes, this could leave customers that don’t have the capacity to invest in backup power in the lurch. Put differently, increased private investments in reliability could have equity and justice implications. 

On the efficiency side, it’s interesting to compare the potential costs of proposed grid reliability investments against our back-of-the-envelope backup calculations for decentralized solutions. Investments in grid-level improvements are measured in terms of many billions of dollars per year. It seems possible – even likely – that some of these investments could be cost-effectively postponed or avoided with targeted investments in behind-the-meter storage and back-up generation. 

Should private consumers be relied on to make the most socially beneficial behind-the-meter investments? Should the deployment of these decentralized assets be coordinated in a centralized way? We’re starting to see a mix of both as retail suppliers launch programs to steer these investments in ways that can help meet reliability requirements and address equity concerns. We’ll be following these investments with interest, even if we don’t end up investing ourselves, because they have implications for everyone.

Keep up with Energy Institute blogs, research, and events on Twitter @energyathaas.

Suggested citation: Callaway, Duncan; Fowlie, Meredith. “The Changing Economics of Electricity Supply Reliability” Energy Institute Blog, UC Berkeley, November 18, 2019,


20 thoughts on “The Changing Economics of Electricity Supply Reliability Leave a comment

  1. There is a third option (The Semi-Shiny Object?): our batteryless grid-tie inverter includes a separate AC outlet that can be energized when the grid is down and the sun is shining.

    The outlet provides up to 2000 watts, assuming the photovoltaic panels on the roof are producing that much energy. That is enough to cool down the refrigerator during the day, and charge cell phones, laptops and flashlights.

    The AC outlet was a no-cost option when our PV system was installed, and there are no batteries involved. It does require manual intervention and running some extension cords in the event of an outage, and enough electrical knowledge to keep total load under the 2000 watt limit.

  2. Meredith and Richard,
    I almost hate to say this, because the TOU arbitrage under PG&E’s new EV2-A rate is even bigger than what you have under your grandfathered E-6 rate (and unlike E-6’s does not disappear in the winter), but starting July 1 of this year Residential customers with (non-trivial) energy storage are allowed to be on the EV2-A rate even if they do not own and register an EV.

    The EV2-A rate has a 4-9PM peak, unlike the original EV-A rate’s 2-9 peak or E-6’s out-moded 1-7 peak, so depending on your load shape and the size of your PV system E-6 still might work out better (again, for you). But you sure can’t complain about the rate differential being too small, for which Richard seems to have temporarily and uncharacteristically forgotten his basic economics training on marginal costs and benefits 🙂

    By the way, in our household we are exploring an even lower-cost, limited solution – buying a $150 300W 36V to 110V inverter to run refrigerator (for a while) and gas water heater (which still needs a fan and electronics), or computer system, all from our 36V, 13Ah electric bicycle batteries. You may have a smaller 36V weed-whacker battery you could plug into it, though that wouldn’t get you up to 300W.

    • The issue on peak/off peak differential is how Ramsey pricing is to be implemented, as the EPMC scalar is Ramsey pricing with an assumption of constant elasticity across customers. SCE implemented its TOU pricing using a different understanding of that application.

  3. Our historical resilience system is an $80 inverter from Harbor Freight. It attaches to our car battery, and delivers us up to 800 watts of power — enough to keep the fridge and freezer going during an outage. But you have to run the car engine. Using a 134 hp Honda Civic engine to drive an 800 W inverter is not a good load match. During a 5-day outage, we went through six gallons of gas running this a few hours in the morning and evening, and using lights and charging phones while it was on. We might have produced a total of two kilowatt-hours. But if it’s only five days every ten years, that’s not a lot. $80 for the inverter, $20 for gasoline, a total of $100, or $20/day. But not continuous power — just enough to keep the ice cream frozen and the milk cold.

    But now we have an improved model. The same inverter can attach to the coach battery of our Kia Niro EV. The Niro has a small 12V “coach battery” for running the lights and radio, and a much larger 64 kWh 356V “traction” battery to take the car 293 miles (rated range is 239 miles, but apparently we’re better than average). With the car turned “on” but sitting in the garage, the inverter draws on the coach battery — up to 800 watts. When the coach battery drops to 50% charge, the car automatically recharges it from the traction battery. So, in theory, we have access to a 64 kWh battery, as long as we either don’t drive away. If we can drive somewhere else to charge the car, maybe we can go more than 64 kWh.

    We’ve not tested this in a 5-day outage yet, so I don’t know if it will recharge the coach battery 20 times to keep things running. I’m watching the web to see if someone else tests this first — to see if there are disastrous results on the car battery.

    Nissan offers a “Vehicle to Home” package with the new Leaf, designed to provide exactly this kind of resilience.

    But neither the Niro nor the Leaf will do this without eventually relying on a source of supply — which a PG&E customer (or one in Puerto Rico) might need. An off-grid PV system can do this:

    If you don’t already have an off-grid rated PV system, maybe that’s where “microgrid in a box” may come in: a shipping container that contains the solar panels, batteries, inverters, and controls systems to let you be independent. See

    We’ll stick with the grid for now. It’s only eleven cents a kilowatt-hour here in Olympia, plus $7.44 per month. BUT, if the utility raises the fixed charge — things could change.

  4. The first step we too [in S Asia] was get large blocks of ice for keeping food refrigerated, and hand fans [which functioned as mosquito/fly swatters], and kerosene lanterns. Went through university with these backups. Then came circuit breakers and UPS which kept only a few lights and fans going. Diesel generators. Solar w storage.

    Looks like the US is moving backwards – or forward to energy savings.

  5. “At $50/day, the gas generator looks like the winner…The other remains enamored with that sleek Tesla Powerwall. Energy storage is the future.”

    Meredith, there is no evidence from either physics or economics suggesting energy stored in batteries could possibly provide a reliable supply of grid electricity. None. The public has no understanding of how much it would cost to provide a single day of electricity to California from batteries (multiples of our state budget), not including the extra clean generation necessary to charge them. Batteries would need to be replaced every 10-12 years.

    Currently, virtually all grid-scale battery installations in California are sitting next to gas plants. Why? They permit gas generators to avoid the expense of building sufficient generation to efficiently meet peak demand. Instead, they charge batteries at night to be able to shave a tiny slice from the top of daytime peaks. Batteries waste at least 15% in resistance and bi-directional inversion losses, making gas generation 15% dirtier than it is already. And of course ratepayers are charged for that extra, wasted gas.

    The push toward “distributed energy resources”, once homeowners discover their solar panels don’t work at night, can only result in increasing adoption of home gas generators. The losers are all of us when it becomes no longer possible to upgrade society’s supply of electricity en masse, when dealing with climate change is entrusted to self-interest. As evidence of a disaster in the making I need only point to Wal-Mart, the largest corporation in the world. The company’s $800+ billion/yr in revenue has been achieved not by providing the best product selection, nor the best possible service, but by sacrificing all other considerations to offer customers the lowest possible price.

    • Interesting that you know more about this than a whole industry that’s putting it’s money on the line to invest billions of dollars in developing this solution. Having skin in the game makes one much a much more credible source.

      Storage is located next to those plants for the singular reason that the grid interconnection is low cost. There is NO reason that storage needs to be next to a powerplant to charge.

      Centralized management is not the answer to our societal problems. Ask the Soviets.

      • Even more interesting that you think money invested in storage represents a “solution” to anything but the profitability of gas companies. The only skin they have in the game is avoiding investment in new generation to meet peak demand, and convincing renewables activists electricity flies from solar & wind farms to their batteries.

        What incentive would gas plants have to store other generators’ electricity for them?

        Not the first time we’ve heard regulated public electricity represented as a Commie plot:

        “In its November 1934 summary, the FTC documented the ‘propaganda’ war waged against the public power movement dating back at least to 1919. In fact, the industry’s own annual proceedings clearly document that its anti-public power campaign had been active since the 1890’s. In 1906, the National Electric Light Association’s ‘co-operation’ campaign was established in part, to monitor and counter the nationwide public ownership movement.”

        Unfortunately, free-marketers lost that one when Commie Franklin Delano Roosevelt and Co-Bolsheviks Sam Rayburn and Burton Wheeler passed PUHCA in 1935. It was the beginning of 50 years of reliable, affordable electricity service – then gas interests and their renewable besties got their grubby fingers on it and messed everything up.

        • The electricity industry hit its crisis in 1973 when the combination of spikes in fossil fuel prices and hitting the peak efficiency for superheated pressurized boilers caused electricity generation costs to start to increase. Rate stability disappeared and the pressure for other solutions, such as the various deregulation measures in the late 1970s, arose. PUHCA had little to do with utility rate stability, and didn’t protect ratepayers when the underlying technological and economic situations changed.

          • PUHCA had everything to do with rate stability. It prohibited sales of goods or services between holding company affiliates at a profit, preventing the utilities from increasing their cost-based regulated rates by artificially marking up the prices paid by the utility operating companies above what the central purchasing affiliate paid. Now, nuclear plant shutdowns have nothing to do with “economics”, but the renewed ability of U.S. utility holding companies to artificially manipulate gas prices for their electricity susidiaries. Just like they used to do.

            The energy crisis of 1973 was the result of OPEC embargoing oil to the U.S., causing gasoline and natural gas prices to skyrocket. U.S. oil lobbyists recognized nuclear energy, with its abundant supply of cheap fuel, posed an existential threat – so instead of taking France’s lead and building nuclear plants, in 1978 Congress enacted legislation encouraging more domestic oil and gas production (PURPA). Also included were incentives for solar panels, wind turbines, energy efficiency, and other trifles which posed no threat to the petroleum industry at all.

          • If PUHCA created rate “stability” why did rates increase so much in the 1970s and 80s? (I’ve responded to you before with the statistics that demonstrate this point.)

            Having demonstrated that closing Rancho Seco was the best economic choice, I have seen that economics leads often to the choice to close nuclear plants. Your assertion is unfounded and based on ignorance of the cost issues.

  6. Nice article! I think a comparison of a battery and back-up generator economics should also take into account the optionality value, which is likely to be much higher for batteries.

  7. Is the high level (private economics) conclusion: Battery costs need to fall by 66% in order to be cost competitive with home generation?

    What is the service life of a powerwall? And a generator?

  8. “However, we’ll be pushed onto the E-TOU rate in 2022 which will reduce the arbitrage earnings to only about $60/year. What a difference a rate makes!”

    PG&E was taken to task in its 2017 GRC Phase II decision for implementing EPMC scaling incorrectly. That suppressed the peak/off peak differential. (SCE’s 2018 GRC Phase II shows how the rate would look if done correctly.) PG&E’s 2020 GRC will fix this problem and the rate differential will look much more like the previous rate differential. You should rerun your analysis using the peak/off peak differentials that SCE has instead.

%d bloggers like this: