Utility Data Hackathon - Dec 2023
Energy Insights Takes 2nd Place!
At the start of the week, I came across a link for the Utility Data Hackathon, hosted by utility data company Bayou Energy and early-stage investment company Stepchange. With a full-time job and a family, I am constantly overcommitted, so this is the type of thing I shouldn’t entertain. But, the prompt was right up my alley, the kickoff and presentations fell in open slots on my calendar, and I convinced myself that I could do my part during the late night hours (who needs sleep anyways)? So, I signed up.
The hackathon description is what hooked me:
This is your chance to explore an API that answers questions like
How much gas has our house used in the last year?
How has my electricity usage fluctuated?
How much do I spend in different periods
Can I hook this up to visualize this in my house?
What can we understand about energy usage when mashed up with real estate data or weather data?
What are the opportunities for reducing my energy cost and carbon emissions?
We'll provide you with early access to an instant utility data API. The API gives access to household or business utility data (bill PDFs, 15 min interval data, and account info) upon authentication.
I’ve certainly done enough exploration of my own utility data, so I thought it would be nice to explore someone else’s for a change.
Bayou set up a Slack workspace and sent us a bunch of ideas and Bayou API details in advance, so I spent the night before testing the API connection with dummy data and brainstorming a few ideas.
The Hackathon
The last time I was involved in a hackathon, I was the one organizing it. It’s a thankless job, and no matter how much you prepare, there are always details you can’t test in advance. Which of the registered participants will actually show up? You don’t know, so you can’t finalize the teams in advance. Will your Zoom account support multiple simultaneous breakout rooms? (Usually not, but it’s not something you realize is a problem until the event is underway and dozens of participants are trying to get into the breakouts simultaneously.)
The hackathon organizers, including James Gordey and Ben Eidelson (Climate Papa), handled these challenges while also supporting the product (Bayou’s API) which was central to the event itself, not to mention mentoring hackathon teams. They recruited great team mentors and judges with experience across many utility companies and startups.
They also managed to land the perfect number of participants: enough where it’s competitive and a lot of cool ideas are generated, but not so many that teams don’t get enough time to present and interact with each other. By my count there were about 35 participants across nine teams.
I introduced myself in the Slack channel and noted that I was hoping to join up with one or more software engineers who would complement my data and product skillset nicely. The organizers paired me with two other people, but only one showed up to the event: Sam Kortchmar. I got lucky: Sam is a former SpaceX software engineer, and a good one.
The Problem
Sam and I had a few ideas around the Bayou API’s granular time series electricity and gas consumption data which we wanted to connect to the Shovels API featuring data on permits (which was also made available to event participants). After some input from our team mentors, including James, Ben, and Dhanur Grandhi, we opted to focus on a solution that addresses a problem faced by utility customers but also utilities/transmission operators, electricians/contractors, and electrical device manufacturers.
Some customers might be motivated to electrify for climate reasons, but most just care about saving money. Fortunately, electrification does that too, it’s just not widely factored into purchasing decisions (yet).
Despite the climate and economic reasons to electrify everything, in practice this proves quite hard. If you’re a utility, electrician, or manufacturer, you don’t know who has already electrified and who is still running on internal combustion engine cars, gas heaters, and gas stoves. Your business depends on generating more business: selling and installing EVs, heat pumps, electric or induction stoves and other electrical appliances. You know that there are plenty of customers out there, you just need to reach them.
Furthermore, the onboarding process of electrification companies like Elephant Energy and WattBot require several steps including question and answer sections about your current home appliances. Product managers recognize that extra steps in onboarding represent friction that impedes customer activation and slows the amount of time to the “aha” moment and value realization. Reducing this friction is important to growing the business.
We need low-friction solutions that bring together the supply and demand to electrify households, and quickly. Fortunately, utility data can help with that.
The Solution
From my prior dabbling (partially document on this site), we knew that certain appliances create conspicuous patterns in electricity data. For example, from my own PPL utility data, we can see distinct signatures of solar panels, a smart theromostat with heat pump, and level 2 EV charging.
We figured we could apply this to Bayou Energy’s interval data, which includes electricity and gas usage down to 15 minute intervals similar to my PPL data. We were limited to just one customer’s data in 15 minute intervals in the project, so we weren’t able to build any sophisticated models to do this at scale. However, it proved to be enough to identify a couple of patterns and build a demo app.
In our one example customer, we found the clear signature of a level 2 EV charger. We could also tell when the homeowner went on vacation and when they turned off their space heater as the weather warmed in April. All of these were confirmed by the customer themself. This provided proof that our core concept had merit.
We can flag these insights for the utility customer to better educate them on how their actions tie to their energy usage and, ultimately, their utility bill and their wallet.
With this knowledge, we can also make recommendations to help the customer save money. Maybe they didn’t notice that their EV charger’s performance has degraded. Do they need a recommendation for a good electrician and a new state-of-the-art charger? Perhaps their HVAC usage could be optimized with a smart thermostat, or their electricity usage could be optimized for their utility’s time of use rate?
This serves as kind of a marketplace for electrification services. It could fit into a new product or marketplace, or integrate into a utility’s existing customer web portal.
The Process
It’s tough for just two people to put together an entire project in such a short time frame, but Sam and I complemented each other on multiple levels. We each brought some ideas to the table; I had a headstart on working with customer utility data, but he was able to throw together a working demo incredibly quickly based on some initial work I did in R.
Furthermore, Sam’s schedule and mine made for convenient handoffs. After some initial brainstorming and knowledge sharing, Sam was able to start exploring the APIs and building a prototype. When I became available later in the evening, he already had a initial prototype up and running, and I was able to put together some analysis to guide our next steps. He came back online late at night (though three hours earlier in his Pacific time zone) and took my updates and filled out the core functionality. I was back awake by the time he ran out of steam around 4am PT, and by the time he was up I had a draft slide deck. He was able to review the deck and add some final pieces and cosmetic touches before our presentation slot on Friday.
Energy Insights
We’re quite please with how this turned out. Apparently the judges agreed, awarding us second place out of the nine participating teams. We will receive Bayou API credits and Electrification coloring books (the latter of which will be a big hit at my house).
Here’s our slide presentation and demo site, if you’re curious. Feedback is more than welcome! Whether this goes anywhere, I’m not sure, but I enjoyed the collaboration and networking with others working in the energy sector full time or as a hobby.