.png)
16 hours ago
2
Network protocols for distributed energy resources - OpenADR, Matter, Energy Protocol, EEBus etc - want to send prices to devices. That’s not good enough.
Big battery sitting in our driveways
There are about a quarter billion cars in Europe, and about the same in the US. They will all be EVs eventually.
In the US you’ll have 250,000,000 EVs sitting at their level 2 chargers at home and in parking garages. What’s the power of that fleet, ballpark? Say the level 2 chargers will be around 22kW on average - hand waving here. That makes about 5.5TW of power.
Is 5.5TW a lot? You know, it’s all relative. The big white fusion reactor at the center of the solar system with wireless energy transfer via electromagnetic radiation is a lot bigger, just the radiation that reaches earth carries 173,000TW continuously - but 5.5TW isn’t nothing. You’d need 55 US nuclear fleets to put out 5.5TW, for instance.
Many of us are very excited about this very large battery and how to control it!
Making the battery turn on and then off again
There are two ways of controlling assets made up of millions of small devices like these. Either you tell them what to do directly - “Turn on, turn off, discharge 10kW” - or you make them participate in some form of market, using prices to decide if they should stop or start on their own.
The major protocols for controlling these fleets all support the first option - direct control. But they don’t support markets. Instead they’ve invented a third option: Publishing an infinite volume price list. I want to argue this is a misguided approach.
Infinite Volume Price Lists
Recently released Energy Protocol for DC grids from EETF looks super dope.. but like its older competitors, it specifies infinite-volume price lists:
3.1.2. Price Forecast (ID 0x30) This structure is typically used in the context of energy pricing. Pos Name Value type Description 0 timestamp Timestamp Time from when it is valid 1 amount Amount Price amount 2 currency Currency Currency code (ISO 4217)OpenADR has the PRICE event which does the same. Unlike Energy Protocol, OpenADR at least supports separate import/export prices, but again no volumes.
Matter has some analogue, OCPP, EEBus etc etc - all the same story.
Volumes
This post can be summarized very simply as: What do you think will happen when someone sends a price list to the 5.5TW battery that says “the price is really good at 2.15pm today”? You’ll move 5.5TW of power to that one hour, a stampede!
In practice you’d send price lists to regions or groups of devices - but you’ll find you need to subdivide such small groups that you lose the benefit of the price signals. The whole point was supposed to be that you didn’t need to centrally schedule individual devices.
Changing prices
Lets say you’ve sent some prices out for tomorrow and that you’ve said at 3pm prices will be low. As a result, a lot of flexible load has scheduled to 3pm.
Now lets say one of your nukes needs to do an unplanned stop around that time: you need to shed a GW or so at 3pm. What happens now?
Do you send out a new price list raising the price for 3pm? What about the homes that don’t see your new price due to network issues, will they be forced to pay the new high price, or the first price? What about homes that do see the price but no longer have flexibility - they chose to delay charging until 3pm because you said prices are low, but now you raised them.. why should they be penalized?
Even if you did: You don’t want to move all the load, you just want to move about a GW, a specific volume.
Infinite volume price lists create this problem where things get really confusing if you raise prices after you’ve sent them out. If you solve it by not updating prices you’ve sent out that makes them too slow to be useful, forcing you to lock in 24-hour spans. At the VPP I work on, we reschedule thousands of times per hour for hundreds of thousands of devices. That would be impossible to do with fixed infinite-depth price lists.
We need a control mechanism that is as fast as the phenomena - sudden stops of large assets, wind and solar changes - the big battery needs to respond to.
Markets solve this. Rather than change agreed-upon prices under people feet you’d send an offer to buy energy at a high price at 3pm. Homes that want to can take you up on it. If they don’t they keep the original price you agreed to.
Actual markets
Actual markets - energy, commodity, stock, super markets - don’t work like this. Actual markets have prices and volumes.
The stock market doesn’t offer you inifinite Caterpillar Corporation shares at whatever is the current market price. There’s some fixed volume offered at that price, and if you bought all of it the price would change.
If you walk into a Lucky’s Market and beeline to the rutabagas, they will have a price and a volume. Sure they will sell you rutabagas for $2.99/lb! They will sell you the ones they have in the box. If you try to fill your cart with 100 rutabagas because the price is great, guess what? Lucky’s will run out of rutabagas.
The price of some commodity depends on the volume of demand, you can’t know the price without knowing the volume. Wanna buy 10 rutabagas? Lucky’s will sell you that for $2.99/lb. Wanna buy a million rutabagas right this second? You’re gonna have to pay way more.
Energy Markets & spot prices
“Wait” you’ll say. “What about electricity prices? They have fixed price lists, all over the world?”
There’s two variants - long term fixed(ish) tariffs and variable rate tariffs.
Boone County Electric cooperative will sell you kWhs for $0.09/kWh right now. And over here in the EU, I spend all day hawking 15-minute spot priced kWhs, fresh off the central EU daily auction!
The fixed(ish) tariff can’t be used to control anything interesting. If you want to reduce charging of the giant battery for 10 minutes right now to cover some lull in wind production, that ain’t gonna happen by asking Boone County Electric Cooperative’s web designers to update the tariff.
The spot auctions - they are price lists with infinite volume, aren’t they?
No, they are literally the opposite: The price is settled by traders forecasting energy volume demand and production cost and a system that finds the clearing price for those volumes. By the time residential consumers see those prices, it’s already said and done. If you want to buy an additional kWh now you need to go to a faster auction and negotiate a new price.
What actually happens for retail customers in Europe is:
- We try to forecast how much energy they will consume in a given time span, and we buy that volume
- This forecasted volume and similar forecasted production capacity sets the price
- We’ll let our residential customers consume the volume we bought at the price it cleared at
- But! Usually we’re slightly off. We bought a bit too much or a bit too little; to match customer demand we then buy or sell additional kWh at completely different prices than spot
- Those other volume/price lots show up as “markup” or grid fees to the customers
There are no prices without volumes.
This worked ok when consumption was inflexible and EV load was small enough to not affect overall prices. We’re way past that now. What we need now is to smear the fleet peak over multiple hours and to be able to replan the schedules as needed througout the day. Fixed infinite volume prices can’t do that.
What to do instead
If I were designing a protocol for distributed energy resources, I would look at how this kind of pricing is organized for larger assets in the grid - the wholesale energy markets.
Two popular ways:
- Continuous Auctions (eg. Europe’s ‘Intraday’ energy market: similar to a stock exchange with continuous bids and asks, clearing when prices match)
- Interval Auctions (eg. Europe’s daily Day-Ahead energy market: participants send in supply or demand curves and an algorithm finds an optimal price for all curves in one single run each day)
A new protocol like EP has a great opportunity to find a to do this better - something like nodal pricing continuous auctions for DC grids?
