AI data centers, EV's and the Grid

[Brad]

My local utility is trying convince state regulators to let them establish a subsidiary that would supply power and connection infrastructure to AI data centers and probably regular data centers as well.  The radio news says they expect AI data centers to be using 12% of all the power in a few years.

Now combine that with charging all the expected personal and commercial EV's everyone is talking about.  We've already discussed what happens to the local grid in a subdivision if all homes get fast charging stations and most plug in at night.  Plus we've discussed truck stops offering charging for lots and lots of trucks being comparable to a large town.

There is no way the present grid can handle those loads, and nobody seems to be planning for it on a national level.  Heck some places have brownouts now when everyone runs their air conditioners, throw in EV's and AI and you'll have meltdown.

[ergophobe]

We desperately need grid upgrades, that's for sure.

>> brownouts now when everyone runs their air conditioners, throw in EV's

In an ideal situation, EVs would all have bidirectional chargers.

What should happen is dynamic pricing that changes based on demand. You plug in your EV and you just say that if the sell price goes above X, you'll let the car be drawn down to Y%. If the buy price drops to Z, then buy power.

The Kia EV 9 has a 95kWh battery and supports V2G (vehicle to grid) bidirectional charging. That's equivalent to seven Tesla Powerwall 3s.

Let's say the owner only charges to 80% (standard unless planning a long trip) and is only willing to let the utility drawn down to 50% SOC leaving the owner with 90 miles of range (assuming 20% reserve; 150 miles of range if drawing the battery to zero).

That's still like having three Tesla Powerwalls available to the utility. If you have four hours of grid stress during the hottest part of the day, that should support at least 5 houses running AC that the utility no longer needs to provision.

Currently most cars do not have V2G (V2L - vehicle to load - is becoming more commong) and few households are set up to actually use V2G because that requires an interconnect agreement with the utility and some device that prevents backfeeding the grid.

This does NOT require any sophisticated grid upgrades, just a meter that can run backwards and a data connection that can tell the inverter which way to run. The actual electricity once it uses your house is just power on the grid, which means that for the most part it's going to your neighbors' houses. It only becomes a problem when so many EVs are participating in the system in, let's be honest, a rich neighborhood, that it has to get exported long distances. But we are so far from that it isn't worth worrying about and, before it becomes a problem, you just kill any incentive programs.

This is a really cheap and efficient way to roll out battery capacity. Right now, PG&E will give you up to $7500 in incentives if you install a battery in your house and you let them draw from it during peak times and during grid emergencies.

But that means that we have a current system where huge EV batteries sit in driveways and parking lots doing nothing useful 97% of the time and large household batteries sit in a utility room doing nothing useful 90% of the time.

The ROI on interconnect equipment for the massive battery in the driveway that you already own would be far better than on a big stationary battery + inverter that you buy for no reason other than backup power.

So in your scenario, having a lot of EVs with bidirectional charging becomes a plus during a heat wave instead of a pure liability as it is now.

Without offering substantial incentives, I don't think you can get people to make the relatively small additional investment in their home and car, so I don't expect to see this soon.