DOE: Pathways to Commercial Liftoff: Virtual Power Plants
liftoff.energy.gov
liftoff.energy.gov
> Virtual power plants, generally considered a connected aggregation of distributed energy resource (DER) technologies, offer deeper integration of renewables and demand flexibility, which in turn offers more Americans cleaner and more affordable power.
Calling this "virtual power plants" is deceptive. It doesn't generate power. It just moves the load around in time a bit. There are limits to how much benefit that can yield. At some point, customers will want to heat their water or charge their car.
Much of this is a attempt to deal with the problem of intermittent wind power. Wind power varies about 4:1 over the course of a day. Here's today's graph for CAISO, for the West Coast.[1] And for PJM, for the East Coast[2]. Both had about a 4:1 range today, and for both, it bottomed out about now, in late morning. A huge tie line from California to Pennsylvania would not have helped. Peak wind output today was before dawn, which is the lowest period for load.
At least with solar you know when there's going to be output. But even that can vary about 2:1 from day to day in PJM's area.[3]
The important question to ask is, if the grid has N hours of storage, how often is there a power shortage? What should N be? A few hours? A few days?
[1] https://www.caiso.com/TodaysOutlook/Pages/supply.html
[2] https://dataviewer.pjm.com/dataviewer/pages/public/wind.jsf
[3] https://dataviewer.pjm.com/dataviewer/pages/public/solar.jsf
There is some amount of distributed generation out there as well, in the form of either oil or natural gas backup generators, which can be used similarly as batteries.
Edit to add: This wasn't meant as a disagreement with your point. I don't think the existence of a small amount of generation changes the fundamental equation you pointed out on this.
From my understanding, a DER can be as simple as a smart thermostat that can receive a signal from the utility company to set the AC up a couple of degrees during a period of high electricity demand.
As energy is fungible, saving 1kW by turning off 1kW of less-necessary usage is exactly the same as generating an additional 1kW to cope with demand.
Maybe you live in a better climate, but not getting adequate heating or cooling when I need it most negates the entire purpose of having heating and cooling.
Of course except it's only the same if they're closely connected.
Here in Norway, while energy regions are connected throughout the country, the cables are quite limited between certain areas. So generation here might not help for usage there.
So being able to control usage, or having extra batteries connected near the loads, can be even more useful than extra generator capacity far away.
I imagine Norway is seeing a lot of the issues - and opportunities - that having high EV penetration brings, so a good place to watch for what works and what doesn't.
It also, IMHO holds on to an outdated concept that all generation must be centrally dispatched. There are plenty of concepts that allow distributed dispatch including voltage (which in turn can be controlled based on local conditions).
Finally (and perhaps most importantly), it implies that only devices that export generation are part of the solution. The supply and demand must be balanced in real-time. And the demand side of the equation is often much less expensive than the supply side.
I feel like GGP is the one with the preconception about the name "power plant," while the industry (as shown here) is starting to use it to be this broader term encompassing distributed generation and price-influenced demand, etc etc.
When people thinking about generation they think about something different than reducing demand, but it's critical to start seeing that those can be the same thing, in certain circumstances.
So I agree that the VPP terminology leaves a lot to be desired. I prefer the more generic "DER" terminology, because it is more descriptive; either a generator or a load is an "energy resource", and "distributed" is a good description of what is unique about these particular energy resources.
because you're right, the people that own this stuff (and work on these companies) are used to thinking in terms of large transmission-interconnected facilities.
so my take this move has most to do with how to bill people, and how to collect payments and pay taxes on said payments and so on. I expect that the actual electrical engineering is treated as an implementation detail
> By becoming a part of the Tesla Virtual Power Plant (VPP) with PG&E, your Powerwall will be dispatched when the grid needs emergency support. Through the Emergency Load Reduction Program (ELRP), you will receive $2.00 for every additional kWh your Powerwall delivers during an event.
I should point out that I have a Powerwall, and I have no incentive from my electric utility to use it in any way. Overall, with rebates, it cost slightly more than a standby generator, and but it requires no maintenance, so it's "worth it" for me.
But, $2.00 back on a kwh is pretty nice, IMO.
So if you take all these infinitesimally small sources and add them up, a single entity can be a market player for them as a kind of tiny virtual power plant. Naturally, this is extremely complicated and took years to get through the legal processes and for the various grid operators to get it figured out. It's still a work in progress.
Most of the grid is a fixed cost. The transmission lines, distribution lines and the power plants recover enormous investments whether they are used or not.
The marginal costs are surprisingly low. Marginal costs of fueled (fossil or nuclear) power plants are typically < $0.05/kWh.
Peak load is what drives capacity expansion -- both in wires (transmission and distribution) but also generation. Building all of this capacity to only run for < 1% of the year incurs tremendous cost.
By flattening the (demand) curve (through either generation or load reduction) there are tremendous savings for society.
However, electric utilities are natural monopolies, so investor owned utilities (contrasted with munis or coops) must substantiate investment to their regulators who in turn allow them to make a rate of return on their investments. So, these utilities only increase profit by increasing investment.
[0] https://www.sos.state.co.us/pubs/business/news/2012/20120402...
If you are interested in cooperatives the most prominent public thinker on the topic these days is Nathan Schneider, who is a professor at CU Boulder and one of our advisors. For example look up an interview with him on "Frontiers of Commoning" podcast - "Nathan Schneider on Cooperatives and Digital Governance". It is an excellent listen.
It's good to see the DOE taking this seriously.
[0] Virtual Power : The Future of Energy Flexibility
Networking all of the chargers / vehicles into a VPP, means you can actually defer some of that investment (don't have all the vehicles charge at the same time or at full speed if the local distribution is congested), or provide a valuable service to the grid in the form of demand response (Hey grid operator, this fleet of cars always charges between 5:30pm and 7pm. How much would you pay for 50 Megawatts of lower consumption?). So basically it makes it cheaper to charge that fleet of EVs from the grid infrastructure and operation perspective if you charge them smarter and make the whole system smarter.
The VPP operators use this as an additional revenue service for the actual asset. The grid is overall cheaper, and the VPP operator and asset owner get a cut of those savings.
Boo fucking hoo. Grid operators have made absurd profits by running grids that hadn't seen an inspection crew for decades, to the tune that they regularly cause deadly wildfires. PG&E is particularly egregious here - the 2018 Camp Fire was likely caused by a broken C hook dating back to the line's construction in 1921 [1].
Sorry, I can't sympathize with grid operators.
[1] https://hackaday.com/2020/09/17/closely-examining-how-a-pge-...
Like software reliability, additional 9s of service get progressively more expensive. Rolling blackouts on the hottest summer days are BAD and have costs associated with them, and peaking plants make a ton of money by running only a few days per year. It's almost the grid scale equivalent of having a gas generator so you can run extra window AC units because you keep tripping the breaker. Much cheaper to just turn the thermostat up a little bit and accept it being warmer, or turn off your cryptominers when the cost is that high.
Our solution in Toronto was to just run them half of the time all the time.
Haaaaaaard paywall: https://www.thestar.com/news/canada/ontario-gas-plants-were-...
Oddly for a peaker plant, they built these as combined cycle, which wouldn’t make financial sense if it was to only run for short periods.
A “true” peaker plant is inefficient but with lots of capacity but we basically build a “proper” power plant that has good efficiency, so it’s not too bad. But makes you wonder what went so wrong in the planning (even though the mistake turned out advantageous).
> "DOE uses a broad definition of VPPs (virtual power plants) that includes a variety of mechanisms for aggregating and orchestrating DERs (distributed energy resources)"
> "Just as different types of traditional power plants contribute to the grid in different ways (e.g., nuclear plants provide baseload generation, and wind farms provide variable generation), so too do different configurations of VPPs."
Historically grid structure in the USA was broken down into baseload, load-following and peaker plants, a choice made based on the technological limitations of the power sources (coal and nuclear power could not be rapidly ramped up and down. so they were assigned baseload status, while hydro and natural gas could).
That all fails when you put large amounts of solar and wind on the grid, e.g. a large solar PV system might generate 200% of demand at noon, (so you divert half or more to storage for use at night). Fundamentally this means scrapping the whole concept of baseload generation, and instead focusing on storage and distribution of energy from highly variable renewable sources - which is why nuclear is difficult to integrate smoothly with renewables, it's not very flexible. However, DOE is a very pro-nuclear agency, hence their reluctance to abandon the baseload concept.
I suppose VPPs are a useful abstraction for managing a wide variety of resources in one geographical area, but the reality is that replacing fossil fuels with electricity across the board will require truly massive grid infrastructure investments. This will require either steep increases in energy bills for consumers, or large tax-financed government subsidies of grid infrastructure, and possibly even nationalization of the grid.
- Counter-factuals; you're often paying someone to not do something, and it is really, really confusing and hard sometimes to measure whether something didn't happen. The car didn't charge at 4pm.. would it have charged if we hadn't told the charger to not charge at 4pm? What if the owner planned to leave at 4.10pm anyway?
- Double counting & profit sharing: There is a surprising amount of money on the table here and kind of a sudden frenzy - what happens when Ford auto-enrolls your new car into their VPP for their own purposes, and you sell it yourself to an aggregator like us, and your utility sells it too? And how do the proceeds get split?
I could never wrap my mind around how this makes sense. Nuclear has a fixed output, it doesn’t scale up and down the way fossil fuel plants do.
If your nuclear capacity can cover the worst case where solar and wind aren’t producing, it can cover all cases, and the solar and wind become irrelevant.
Seems like the linked concept is similarly making something more complicated in the hopes that somehow makes it better.
Like bundling subprime mortgages.
A fixed thermal output or a fixed electrical output?
It turns out when you run surveys, it’s the people that like to think about answers to questions that respond to them, which isn’t representative.
45pg pdf government report: https://www.auditor.on.ca/en/content/annualreports/arreports... which starts getting good at page 5.
Understated costs and exaggerated benefits. Peak reduction targets ended up with increases in peak demand.
As hard as it is for the HNer to believe, the vast majority of people will just do whatever they want whenever they want.
I see this everywhere: “why would anyone go to this expensive supermarket that sells the same crap as the cheap one I go to? Why would anyone go to that gas station, their price is always higher than the others in the surrounding area. Why would anyone buy that car, it has a terrible TCO, and on and on”
And this is in new construction.
Overall, it didn’t (and may never) make financial sense to put light users on smart/time of day metering.
You'd get more participation if appliances could be smarter without increasing cost. The holy grail of delay freezer defrost cycles until off-peak, and relaxing temperature thresholds a bit to pre-heat or pre-cool towards the end of off-peak and let things go a bit farther off the set point while on-peak aren't realistic for say water heaters, because the increased consumer costs to do that (more expensive equipment, more time to set settings, most likely some signalling or networking to maintain) don't justify the savings.
I'm guessing there's reasonable uptake on smart thermostats that can participate in demand control, because smart thermostats are desirable anyway, network anyway, and can control large enough loads to matter. Not everybody will want it, but I'd be ok with starting peak a degree cooler or warmer than my set point, and letting the temperature float a degree farther during peak, because HVAC uses a lot of kWh and why pay 2x for those... but then I'm in a low energy cost area with flat rates, so maybe I would change my mind after experiencing it. :)
Free LED light bulb handouts are more effective than complex expensive-to-implement programs requiring user intervention. Mandating low gpm showerheads works better than asking people to take short showers. Better windows per building code works better than asking people to choose the exotic greenbug window, etc.
Nuclear can and does regulate its output. And the fuel rods should last longer if you run at less than full output. But fuel cost is not a large component of nuclear energy cost, so it doesn't make a lot of economic sense to operate them far above or below their steady state maximum capacity.
I agree that nuclear doesn't make a good backfill for intermittant sources; but it's not because they can't, it's because the economics aren't good.
They do in France. You kind of have to, if you produce 70% of your power in nuke plants. But in the US, it's not done in any meaningful sense. All US plants run at 100% capacity all the time.
The reason for that is that it's incredibly complicated to ramp a nuke plant. If you take a reactor and reduce its heat output significantly, the currently active volume of fuel moves away from the decay product equilibrium - the reactor poisons its own fuel rods with neutron absorbing decay products. This makes a subsequent increase in output power extremely dangerous and thus impossible in practice. The only way to increase power safely is to wait for the radio poison to decay naturally, which can takes tens of days (and depends on by how much you reduced power).
The only reason the French can ramp the power of their nuclear plants up and down is that they have so many of them, and they are all controlled by a central authority. This means the reactors can take turns in reducing power, each night a different set of reactors poison their fuel and recover slowly over the next week or two.
There is also of course an economic side to this. In the US, nuclear power is so economically risky that planning a plant that is not trying as hard as it possibly can to recoup it's enormous investment would sound crazy to investors. It would make zero economic sense to not run at 100% all the time, and would probably make any newly build plant a billion dollar loss instantly.
This is plainly false; nation-wide energy consumption isn't fixed, it adapts to supply. Not all usage of energy is equally valuable, if the prices go up, you might choose to turn off your TV, but you'll probably keep your fridge running. With more energy, more activity can happen at lower prices.
This is a very, very old, and very, very solved problem.
Today we'd solve it with batteries: Discharge when demand outstrips supply, recharge when supply outstrips demand.
In the past, nuclear power was often paired with pumped storage: https://en.wikipedia.org/wiki/Pumped-storage_hydroelectricit...
Today a lot of pumped storage facilities have outlasted the nuclear power plants they were built to support. Instead of pumping at night, they pump when there's excess solar or wind.
Because battery storage is legitimately better in terms of reaction time and efficiency: pumped hydro is 2 min from 0 to 100%, with a total efficiency ratio between .65-.85 [1], gas plants are at 3-5 minutes (peakers)/up to 30 min (combined power+heat) [2] and .35-.50 efficiency ratio (more if thermal energy is captured, but that only makes sense for non-peaker plants that also provide district heating)... whereas a battery plant's reaction time can be measured in milliseconds.
That's orders of magnitude faster and is what makes decentralized battery storage worth it.
[1] https://um.baden-wuerttemberg.de/de/energie/versorgungssiche...
[2] https://www.ingenieur.de/fachmedien/bwk/erneuerbare-energien...
[3] https://www.deutschlandfunknova.de/nachrichten/australien-ri...
If you can convince every supermarket in the area to designate a quarter of their overhead lights as "optional," and then install a device that allows the "virtual power plant operator" to turn off all of those lights during an unexpected spike in electricity demand, you've just "generated" the additional 100kW the grid needed in that moment, just like a regular fossil-fuel power plant would have. (And you then pay the supermarket for the electricity they "generated" by turning off their lights.)
Do the same with smart thermostats, or tapping into people's power walls, etc etc etc, and you may be able to "generate" many megawatts of power without owning a plant.
(I also wish my fridge and other devices had a better ability to capture the ebb and flow of electricity prices. Even trying to do this with a “smart” thermostat, an ecobee was very difficult. Required lots of manual scheduling.)
In fact, I want it to increase its net power consumption by running hard when rates are low, and backing off when they’re expensive. So overall cost will decrease without any spoilage impacts while consumption goes up.
This approach also minimizes the amount of time the fridge is working against my air conditioner. And maximizes the time it’s working with my furnace on cool nights.
The most thermally/kwh efficient fridge keeps its temperature the same at all times, but that’s not the most cost efficient/climate efficient approach.
It’s like we’re not ready to have that conversation.
My consideration depends heavily on engineering of it. For example, applying this in NYC is almost impossible, and dangerous for the system. Again you can have DER aggregation at smaller levels, but hitting 20% of pick load levels seems extremely challenging.
VPPs are a great way for utilities and other companies to get customers hooked into long-term contracts and milk for more fees.