(This is an adaptation of a recent Twitter thread)
I woke up this morning with a product idea possibly of interest to readers of this Substack. In summary: I want there to be an energy demand-shifting advisor app for residential power customers, and I think it would be a good thing for decarbonization. I don’t know of any such app, either released or in development. If there is one, I want to help it get better and get it to more people faster. If there isn’t one, I might end up starting a github project to make there be one. Please let me know if you know of one, or if you want to help!
The motivation was the current California heat wave which is straining our power grid. I’m trying to do my part to be a good grid citizen and help avoid blackouts, and I’d like that to be easier. I also think making good grid citizenship easier would help accelerate decarbonization: the ability to time-shift demand and thus reduce peak loads can make it much cheaper and less risky to increase the percentage of power generated by renewables. This is partly informed by my reading books like Saul Griffith’s Electrify and partly by my experience at my most recent employer.
Yesterday, during the first big demand peak event of the heat wave, I was proud to be able to say that my house was not only not drawing from the overstrained grid, but actually sending 3-4 KW back to the grid to help reduce the strain. In large part that was due to the house’s Powerwall discharging at max rate, as part of the Tesla Virtual Power Plant pilot program. And that was totally automated, yay.
But the other part of making the house an optimal citizen was minimizing our power consumption during that time. That was a very manual process. It started the night before the peak, when I opened windows to get as much cool night air in as possible. Then in the morning I closed the windows, ran the dishwasher and laundry so they’d be done pre-peak, and turned off the hot tub and the radiant floor heat in the main bathroom. A few hours before peak I briefly ran the heat pump in A/C mode to precool the house. Then during the peak I turned off the heat pump, restarting it when the peak was over to cool the house back down so we could sleep comfortably. And of course I had to remember to not charge the car or the bikes, too.
This was a lot, but I probably could have done more: e.g. maybe I could have had our fridge and freezer run less during the peak by temporarily adjusting them up a few degrees. Others who have more, or different, electric appliances than I do will have more, or different, things to optimize: e.g. preheating water in an electric tank water heater before the peak. And even though I’m a 99%ile early adopter, none of it was done for me, unlike the Powerwall discharge schedule. It’s not just old equipment like the 2004 hot tub that requires manual effort: for example, my 2020 heat pump doesn’t have a smart thermostat.
So if I’m in a highly manual state in 2022, most of the ideal load-shifting population will also be there for the next 10-20 years. Automation will spread steadily but unevenly and incrementally: people will buy new load-shifting gadgets one at a time, not all at once. Even in a future with everything from motorized windows to dynamically-adjusting refrigerators, we’d still need some manual decision making, like deciding whether to run the dishwasher in the morning or wait till late night.
Thus we should look for cheap ways to help people do better at manual demand-shifting. Software engineers know that checklists, playbooks etc are an indispensable complement to automation, and that timely information that helps a person do the right thing can be powerful. Let’s apply that to my case. Say I didn’t know, or was too distracted by life to remember, to do all that stuff: how might we help with that?
Imagine an app that aimed to advise and nudge people to demand-shift optimally. You might set it up by filling out a questionnaire about your main electrical demand loads: do you have a hot tub? Do you have A/C? Do you drive an EV? Is your thermostat smart already? etc. That way the app would know what it needed to nudge you to do manually ahead of a high demand event.
To know when to nudge you and how, it’d need at least two data sources: the weather forecast and the local power company’s day-ahead forecasts of unusually high demand times (like the “Flex Alerts” we get in California). Then, starting the day ahead, it could give you timely notifications with actionable, personalized instructions for each load-shifting step— including letting you know when the peak was over and the coast was clear to do power-intensive things again.
The core customer base would be a small set of greentech geeks. But with the right publicity, you could imagine it appealing to anyone on a time-of-use power rate plan who wanted to save a bit of money as well as helping out the grid. It’s unlikely that would ever generate enough revenue to make it worthwhile for any company to devote a full time development team to it; but this is a simple enough mechanism that it shouldn’t need such a team. Moreover, a free, open-source, volunteer-maintained, standard app to which interested parties could contribute might be a nice way of getting something with broad usability that improved steadily over time.
That’s the idea. Please let me know anything relevant to it in comments!