[71] - SolarEnergyViz.tech

You can make money while being sustainable

Comment

Overview

SolarEnergyViz.tech

Engie challenge with sprinkes of GitHubOps, DevOps, and WebDev.

Here is a list of things we've done:

  • Data Cleanup, Restructuring, Reformatting
  • Calculated Conversion Efficiency for each hour for years 2011 - 2023 (90,000 rows of data)
  • Predicted missing data from 2021 to 2023 (using Facebook Prophet time-series forecasting framework)
  • Energy comparison between EV and CAMBUS arrays
  • Deployed the results to SolarEnergyViz.tech

What it does

Calculates how much money you can make by being sustainable

Calculations

Trains a time-series prediction model to predict future solar irradiance (fill the gap for 2021 - 2023)

GitHub

Compares conversion efficiency of two different solar arrays

Visualizes 12 years worth of Conversion Efficiency

Viz

Cleans the dataset

GitHub

How we built it

We used technologies like GitHub workflows to automate the cleanup and deployment

We used Netlify and CI/CD pipeline to continuously build dataset and deploy the website solarenergyviz.tech.

Challenges we ran into

Some of the challenges we faced while working on the Engie challenge:

Missing Values

Some of the data wasn't clean.

We found missing variables, such as Calc Failed and [-11059] No Good Data For Calculation.

For Calc Failed, we calculated the value by adding daily hours into totals.

For [-11059] No Good Data For Calculation, we used that day's total, subtracted all of the other hours, and split the difference among the hours that had that value.

Extremely High Values

Some of the data provided contained very high values, such as 2.38572E+30. We made sure to filter that data out as to not skew calculations.

All of that data seem to happen during night-time when the kWH records would have been very low.

NSRDB not accounting for tilt, azimuth

The NSRDB data does not account for tilt and azimouth of the array as well as max generation capacity. To account for the difference in installation, we relied on the PVWatts calculator to find the ratio of change.

Month EV Solar Radiation (kWh / m^2 / day) CAMBUS Solar Radiaton (kWh / m^2 / day) Difference
January 4.11 2.53 1.58
February 4.79 3.23 1.56
March 5.71 4.25 1.46
April 6.22 5.02 1.2
May 6.71 5.83 0.88
June 7.26 6.43 0.83
July 7.74 6.60 1.14
August 7.31 5.98 1.33
September 6.74 4.99 1.75
October 5.39 3.62 1.77
November 3.87 2.47 1.4
December 3.17 1.99 1.18

Accomplishments that we're proud of

Three out of four team members have never attended a hackathon before.

What we learned

Learned Git, WebDev, DevOps, and Data Processing.

Built With

Share this project:

Updates