Watts Happening

Inspiration

Electric utilities plan major transmission projects years in advance, but projects owned by different companies may be developed near one another without those potential coordination opportunities being immediately visible.

The Sperry Tech Gridlock Challenge inspired us to explore this problem by identifying geographic and timeline overlap between planned transmission projects, with geographic proximity serving as the primary signal.

That led us to build Watts Happening: a tool that analyzes planned transmission projects across the United States to make it easier to discover where projects from different companies could potentially coordinate construction activities, resources, and scheduling.

Using nationwide project data spanning 163 utility companies, we expanded the idea beyond comparing a small number of utilities and created a system capable of finding project-to-project coordination opportunities across the country.

What It Does

Watts Happening is an interactive dashboard that identifies and ranks potential coordination opportunities between planned transmission projects owned by different utility companies.

We analyze projects primarily based on geographic proximity, with planned in-service timing and project cost providing additional context.

When projects contain multiple substations, we compare their substation locations and use the shortest substation-to-substation distance as the distance between the two projects.

Projects within approximately 40 km (25 miles) can then be surfaced as potential coordination opportunities.

Through the dashboard, users can:

• Explore planned transmission projects on an interactive map
• Discover nearby projects owned by different utility companies
• View ranked coordination opportunities
• Compare project distances and planned in-service timing
• View known or model-estimated project costs
• See potential areas for coordination
• Review rough estimates of potential cost savings

Depending on how close two projects are, Watts Happening can highlight opportunities such as sharing crews, equipment, deliveries, laydown yards, access roads, permitting efforts, and right-of-way coordination.

How We Built It

We built the frontend using React, JavaScript, Vite, Tailwind CSS, and Leaflet to create an interactive interface and geographic visualization of planned transmission projects.

Our backend uses Node.js and Express.js, with MongoDB and Mongoose for storing and working with our processed project and substation data.

Before loading the data into the application, we built a data-processing pipeline in Python using Pandas, NumPy, scikit-learn, and Jupyter Notebook.

Because many projects did not have reported costs, we trained a Random Forest regression model using projects with known costs. The trained model was then used to estimate costs for projects where reported cost information was unavailable. Our processed data maintains a distinction between known and model-estimated costs.

For each eligible pair of projects owned by different companies, we evaluate their substations to determine their minimum geographic distance. Nearby projects can then be evaluated using distance, planned in-service timing, and project cost to surface promising coordination opportunities.

We also developed a prototype cost-impact model that uses project proximity and cost to provide a rough estimate of potential savings from coordination.

Challenges We Faced

One of our biggest challenges was working with real-world infrastructure data.

Projects were not always represented by a single location. A single project could be associated with multiple substations, which meant we had to design our analysis around the relationships between companies, projects, and substations rather than treating every row as an independent project.

Another challenge was incomplete project cost data. Removing every project without a reported cost would have significantly limited our analysis, so we developed a machine-learning pipeline to estimate missing project costs while keeping those predictions distinguishable from known values.

We also had to determine how to translate geographic proximity into an understandable potential financial impact. Rather than presenting estimated savings as guaranteed results, we developed a transparent prototype heuristic that applies distance-based savings assumptions to the smaller project's cost.

Finally, working with a nationwide dataset meant considering a large number of possible project relationships while ensuring that projects owned by the same company were not treated as cross-utility coordination opportunities.

What We Learned

This project taught us that building a useful data product involves much more than creating an interface.

We had to combine data cleaning, machine learning, geographic analysis, database design, backend development, and frontend visualization into a single system.

We also learned how important data relationships are when working with real-world datasets. Understanding that one company can own multiple projects and one project can contain multiple substations significantly influenced how we structured both our database and our geographic analysis.

Building the cost-estimation model also gave us experience applying machine learning to incomplete real-world data and integrating those results into a larger application.

Most importantly, we learned the importance of communicating the limitations of our analysis. A predicted project cost is different from a reported cost, and two projects being geographically close does not automatically mean coordination is feasible.

Watts Happening is designed to surface opportunities worth investigating, giving utilities a starting point for discovering where coordination could potentially save time, resources, and money.

Built With

• css
• eslint
• fastapi
• html
• javascript
• jupyter-notebooks
• leaflet.js
• mongodb
• numpy
• openpyxl
• pandas
• python
• react
• scikit-learn
• vite