In a competent implementation, real-time metering can enable near-real-time pricing where consumers are incentivized to vary their consumption to help grid stability.
Your EV can modulate its charging very quickly, and "background" loads like electric heating, water heater or even A/C can also be modulated somewhat quickly (though not as quick as an EV's inverter).
The meter needs however to make sure you indeed complied with the demand in order to pay you fairly (otherwise if people can defect on their obligation and still get paid, it defeats the purpose of the scheme of ensuring grid stability).
Doing this at the substation is not granular enough because then you can no longer determine who contributed what and whom to pay, which then removes any incentive for people actually participate.
In theory, this should lead to significant savings and efficiency benefits, as everyone opted into the scheme can now be used as an on-demand load to dump excess power (during which power is not only free, but the consumer may even be paid to consume that power) to smooth out supply/demand fluctuations.
Of course, the UK's smart meter scheme is administered by Capita, so don't expect any of this to actually happen, work reliably or actually lead to any kind of significant benefits, but in theory, it would be a great thing as long as it's done by competent people without corruption/mismatched incentives.
Octopus is doing hourly blocks, and is limited to a single provider. It's an interesting idea and broadly heads in the right direction, but is far away from what I'm describing.
What I describe would be real-time with seconds-level granularity and operated by the grid operator as part of the distribution network and thus provider-agnostic. The idea is to (financially) incentivize people to shift their demand and production around grid fluctuations in real-time - this should allow everyone to get "more" out of the grid by better coordinating supply and production and respond to unexpected events.
My energy supplier (Octopus) gives me a tarrif with half-hourly pricing (smart meters in the UK only report with 30 minutes resolution).
This let's me know when electricity is in less demand or high supply and schedule my day around it. Making my electricity bill cheaper and likewise putting less strain on the supplier which reduces their costs too
The other week, it was not very windy but it was cold, so the energy price spiked pretty high (£1/kwh). Other times the energy price falls pretty low, and I regularly see it be negative for some hours of the night or rarely up until the afternoon.
Paired with a home battery it can be pretty effective, but I don't have one and instead I just work around it and use my electrical heating more when it's cheaper and rely on my insulation to last throughout when it gets more expensive during the day. I've also started cooking dinner later to get past the evening hump.
Just shifting some electricity usage around is a pretty solid tactic with dynamic pricing.
I'm currently paying wholesale + 0.49 c/kWh margin for my electricity and I'm averaging out to 6.67 c/kWh (incl. margin) in December. Wholesale average has been about 6.61 c/kWh. Add transfer and tax on top of that and it's around 0.13 €/kWh total.
I live in an apartment so the two big electricity-wasters are the dishwasher and the washing machine. Delaying doing the laundry by a couple of days or running the dishwasher half a day later keeps the average much lower. And if you have an EV, just charge it during the night when wholesale prices get closer to 0€.
Here's my prices for tomorrow for those who are interested. They publish an API for this too which I have loaded into a spreadsheet for my own entertainment
They also identify outages and let you do automatic efficiency stuff — you can enroll in a plan where your car charges during off-peak periods, for example.
20-30 year old mechanical meters tend to be pretty accurate, but when they aren't, they tend to read low. Especially if your usage is fairly low, the minimum usage to push the wheels forward may be higher than your standby loads.
If you've been on a meter that reads low for a long time, a new meter (mechanical or electronic) will be a big jump.
Mechanical meters either accurately measure use, run to slow or don't work at all. The worse thing that can happen is that other people indirectly have to pay for the missing killowatts. In contrast there are plenty of stories where people get bills 10-20 times what they are used to.
Any half decent programmer on HN knows what to do when something has almost no chance of happening.
Some of these things display only billable KWH while it isn't so expensive to display current consumption too.
You can also measure it twice so that one can (remotely) see two entirely different readings and replace the meter with one that works properly. It could also display the amount of money owed. (estimate) It would make it slightly more expensive for everyone rather than accidentally billing some people truly crazy amounts.
Just like a bank account can have a configurable limit for withdrawals the smart thing to do is to set a limit per day, week, month and have an alarm go off and/or shut everything down.
Say my monthly bill is 200, I don't want the previous 7 days to exceed 500.
Just a flashing led and a 5000 euro bill isn't smart. Imagine diagnosing a server by how fast the led flashes. Imagine replacing the speedometer in your car with a flashing light. You will know how fast you drove when the fines come in.
Apparently the UK government estimated that in 2016 and 2019 the average household saved £11 on their bill due to this feedback from smart meters. This page estimates the cost at £15 so 2/3rds is recovered just from that effect.
There's no way it only cost £15 in total to install a smart meter. The cost of the meter has to be at least that much and then you have paying a guy to actually install it (which in my case took 3 attempts and caused a gas leak, which took 4 or 5 guys several hours to fix - although hopefully that's unusual).
15 pounds cost per year, 11 pounds saving per year, for an average user.
Not sure how many years the cost is assumed to be spread over and what the ongoing running costs are but that's the per average customer cost apparently.
