Hack On Cloud

https://youtu.be/0AUdwZISUdo

https://sunshine.hackoncloud.com/

I.About the Project

In addition to providing an interface to display the “available sunshine” for the past year, this project aims to provide decision makers (home owners, businesses, planners) with information and recommendations on solar panel installation using their expected electricity usage, “available sunshine” from NASA Power Web Services, our own developed data models and publicly available information from solar energy providers.

Given an user-specified location and the expected monthly electricity consumptions, the application provides a graphical display depicting the time series of the weekly average "available sunshine" for the past year, along with recommendations for installation of solar panels.

Our team believe that this is important to provide not only the public but also decision makers with such recommendations so that they can consider and make better investments for the future. Applications and main user groups could be businesses that are planning to build solar-powered factories, architects who want to design environment friendly buildings / houses, home owners who want to move to reduce their utilities bills or city officials who want to plan for future green electricity infrastructure.

Key Outcomes in the 2 days of the NASA Space Apps Challenge:

1. Visualisation of Sunshine information (weekly & monthly) given the user selected location
2. Development of a set of APIs that crawl information regarding solar panel specifications from local solar panel providers
3. Integration with NASA API to retrieve sunshine information, optimal angle information, solar irradiance given a location
4. A Data model that forecasts the solar irradiance in the future and recommends the best solar panel setup for the users, leveraging on state-of-the-art statistical methods & data modelling tools. This was also benchmarked against the publicly available industrial information
5. A Progressive Web App (PWA) that delivers all of the above and is accessible via anywhere by any device. In addition to this, rapid development an Android App that can be installed on any Android device (.apk file provided)

II.How We Addressed This Challenge?

We tackled this challenge by addressing two key problems for users: 1) Visualisation of the historical “available sunshine” and 2) Recommendation on solar panel installation.

1. Visualisation of the historical “available sunshine”

Based on the user-specified location (latitude & longitude), the application will call the APIs from NASA Power Web Services and retrieve the “available sunshine” information - All Sky Surface Shortwave Downward Irradiance. This information is provided on a daily basis and is aggregated by our data presentation layer, allowing users to view this data using “Monthly” or “Weekly” view.

Note: All Sky Surface Shortwave Downward Irradiance: The total solar irradiance incident (direct plus diffuse) on a horizontal plane at the surface of the earth under all sky conditions. An alternative term for the total solar irradiance is the "Global Horizontal Irradiance" or GHI.

2. Recommendation on solar panel installation

We found that visualisation of sunshine information does not bring much benefits to the public. Although there are many solar energy providers and information available on the market, it is still challenging for the public to consider and plan for transitioning to solar power, especially within the South East Asia region.

Our team decided to take one step further by combining the NASA sunshine information with our own data models to provide users with practical recommendations, including recommended panel size based on users’s needs, recommended panel angle for installation, along with calculation of breakeven period for planning, as well as savings in monetary value.

III.How Does the App Work?

1.User enters information including User Location, Average monthly electricity bills, Expected Solar Panel’s Duration.

2. Based on the above information, system provides users with information of historical sunshine information of the location and Recommended Panel Size, Panel Angle for installation and calculation of savings base on their submitted info.

IV.System Architecture

Coding Languages: .NET Core, Python, React.Js, Next.Js, JavaScripts, Tableau Coding Languages

We made use of the following tools for this project

• AWS API Gateway, AWS ECR, AWS Lambda Function, AWS S3, AWS Amplify
• Progressive Web App Technology, Tableau Visualisation, Google Map API, NASA API

The following diagram describes our solution:

V.Our Data Model

In order to create our data models, we created APIs to crawl information form various Solar Panel Providers and NASA. Information crawled includes:

• Solar Panel Capacity (C) (Unit: kWh per m2)
• Size of one Solar Panel in Square Meters (SIZE) (Unit: m2)
• Total Installation Cost per m2 (TC) (Unit: USD / m2)
• Maintenance Cost per m2 (MC) (Unit: USD / m2)
• Recommended Maintenance Duration (MD) (Unit: years)
• Electricity Price (P) (Unit: USD / kWh)
• NASA Sunlight Input based on User Location - NASA Surface Optimal [SI_EF_TILTED_SURFACE_OPTIMAL] (NASA) (Unit: kw-hr/m^2/day). NASA is a collection of 12 data points, each data point corresponds to each month within a year. The value for each month is averaged across the selected start year (set as 2019 by default) and selected end year (set as 2020 by default).
• Epsilon: We ran our model against benchmark data (publicly available sources from 3 solar panel providers) on expected output per panel.

We have also came up with the following assumptions & variables

• YEAR: User’s expected duration of the solar panel - Entered via user inputs
• Material Efficiency (ME): (Unit: percentage) - Based on industry standards of 15 - 25%, We take the default value as 20%
• Expected Depreciation (D) per year (Unit: percentage) - We take default value as 0.5%
• Expected growth rate (G) of NASA Sunlight Input based on User Location: We ran regression on the historical of data from 1990 to 2020 to identified if there is any trend in Sunlight Input. Based on analysis, our conclusion is there is no change in terms of irradiance over the years (0 % / year)
• Total Solar Irradiance Forecast by Year

Historical Data: 1990 to 2020

• Data: The total solar irradiance incident (direct plus diffuse) on a horizontal plane at the surface of the earth under all sky conditions. An alternative term for the total solar irradiance is the "Global Horizontal Irradiance" or GHI.
• Parameter: All Sky Surface Shortwave Downward Irradiance (ALLSKY_SFC_SW_DWN)
• Total Solar Irradiance Forecast by Year

Historical Data: 2006 to 2020

• Data: The total solar irradiance incident (direct plus diffuse) on a horizontal plane at the surface of the earth under all sky conditions. An alternative term for the total solar irradiance is the "Global Horizontal Irradiance" or GHI.
• Parameter: All Sky Surface Shortwave Downward Irradiance (ALLSKY_SFC_SW_DWN)

VI.Our Recommendations Model

1. Recommended Panel Size

• We converted the average monthly electricity bill entered from users to expected electricity (EE) usage in kWh per month
• We calculated the output per panel per day on the lowest month (PO)
• Optimal Energy month XX in the first year = Value of month XX from variable NASA (where XX is any of the 12 months of Jan, Feb, Mar,…, Nov, Dec)
• Optimal Energy month XX in the second year = Optimal Energy month XX in the first year * (1-D)
• Optimal Energy month XX in any year after the second year till the end of variable YEAR = Optimal Energy month XX in the previous year * (1-D)
• Optimal Energy = Sum of Optimal Energy of each month within variable YEAR/Number of months per variable YEAR
• PO = Optimal Energy * ME * SIZE * Epsilon * (1+G)
• Recommended Number of Panels = EE / (PO * 30)
• Recommended Panel Size = Number of Panels * SIZE

2. Recommended Panel Angle & Orientation

• We retrieved from NASA POWER API the list of recommended angles and orientations in the last 12 months and display the information to users
• In addition to that, we also come up with a recommendation of a fixed angle and orientation for optimal performance if changing angle and orientation is not an option for user.

3. Potential Savings

• To calculate the net savings throughout the entire duration of the panel, we use the following formula:
• Net Savings = Savings Per Month * Duration in months of Panels - Total cost

4.Costs

• Total Installation cost = Number of Panel * Size * Cost per m2
• Maintenance cost / Year = Number of Panel * Size * Maintenance cost per m2
• Total cost = Total Installation cost + Maintenance cost / Year * Duration in years of Panels

VII.The Benefits

1.In addition to the visualisation of the sunshine data from NASA, public users now will be able to have better understanding of how we can utilise the sun for solar energy.

2.Decisions makers, especially in South East Asian countries, urban and rural, now have a better tool to make informed investment decisions on solar energy. Potential users of the tools would be:

• Home owners who plan for their next solar panel setup
• Architects / Property Developer who want to design environment friendly and energy-efficient buildings / houses
• City officials who plan for future electricity infrastructure
• These will hopefully encourage the adoption of green energy or solar energy, moving away from fossil fuels.

VIII.What could have been done given more time?

• Onboard new Solar Panel Providers for different countries
• Purchase professional version server for Tableau so that real-time API and integration can happen directly from Tableau to NASA API
• To include more technical features to the model that calculates the recommended panel size
• To include more features from NASA for visualisation . At the moment we selected only what we think is useful for the users.

We used the following space agency data from NASA POWER

1.All Sky Surface Shortwave Downward Irradiance

• This is the downward thermal infrared irradiance under all sky conditions reaching a horizontal plane the surface of the earth. It’s also known as Horizontal Infrared Radiation Intensity from Sky.
• We use this data to derive total potential solar irradiance at given location by daily, monthly

2.Optimal Solar Energy Generation: we use following data to provide users with an ideas of potential solar energy at their location and what are optimal setup of solar panels to generate highest solar energy at a given month:

SI_EF_TILTED_SURFACE_HORIZONTAL:

• This data gives users an idea of potential sonar energy at their location if panels are setup to facing surface tilted at a 0 degree angle (horizontal plane).

SI_EF_TILTED_SURFACE_LAT_MINUS15

• This data gives users an idea of potential sonar energy at their location if panels are setup to facing surface tilted at the latitude minus 15.

SI_EF_TILTED_SURFACE_LATITUDE

• This data gives users an idea of potential sonar energy at their location if panels are setup to facing surface tilted at the latitude angle.

SI_EF_TILTED_SURFACE_LAT_PLUS15

• This data gives users an idea of potential sonar energy at their location if panels are setup to facing surface tilted at the latitude plus 15.

SI_EF_TILTED_SURFACE_VERTICAL

• This data gives users an idea of potential sonar energy at their location if panels are setup to facing surface tilted at a 90 degree angle (vertical surface).

SI_EF_TILTED_SURFACE_OPTIMAL

• We use this data gives users an idea of optimal sonar energy at their location if panels are setup in an optimal way based on our suggestions.

EF_TILTED_SURFACE_OPTIMAL_ANG

• Given that users want to achieve optimal solar energy generation, we use this data to recommend users which angle is the best each month to yield the most solar energy.

SI_EF_TILTED_SURFACE_OPTIMAL_ANG_ORT

• Given that users want to achieve optimal solar energy generation, we use this data to recommend users which direction (north/south) the panels should be setup to point to.

Problems:

1) We use Tableau as the main tool for data analysis and visualisation, and the community version of Tableau we are using does not allow us to consume backend APIs with parameters. That means we were unable to pass in Latitude, Longitude and respective sunshine information to our Tableau data visualisation model.

To overcome this, we decided to

• Use iframe with parameters to pass Latitude and Longitude to Tableau
• Instead of calling the NASA API real time to get sunshine information for user’s input location, we used the NASA API to download regional data (South East Asia) to our own database. We retrieved all data for a grid of [5x5] region to cover South East Asia. We then further divided each grid into smaller regions and compared with user’s Latitude and Longitude

• Information will then be retrieved based on the respective grid user is in, and the data will be visualised via Tableau.

2) Another problem we have was that the NASA API was not available for a period of time when we were testing the APIs, which may block the development of the whole team.

To deal with this, we are split into 3 teams: Data Team, Front-end(FE) Development and Back-end (BE) Development. FE and BE were using mock data to come up with the web version, as well as the Progressive Web App version. The Data team used the method in (1) to limit the dependencies on NASA API should it is not available again.

3) Another problem we tried to overcome is whether in the next 10-25 years, the selected location will receive the same irradiance in the last year.

For this, our team leveraged on Tableau Analysis and Prediction model to go through 14 years worth of Sunshine data in Singapore (2006-2020) and forecast total irradiance by both years and months of the year, and came to the conclusion that while there is no definite trend, these values may vary 5% depending on the years. This was also included in our data model.

Achievements

1. The biggest achievement for us would be the knowledge gained over the course of 2 days regarding solar energy. From how irradiance is converted to solar energy, into understanding of technical concepts such as Material Efficiency, Solar Panel Capacity, Expected Energy Depreciation, and even concepts available from NASA (covered in the Space Agency Data section).
2. We are also very proud of ourselves coming up with a solution that combines our strengths and bring value to the end users of the app:

• Visualisation of Sunshine information (weekly & monthly) given the user selected location
• Development of a set of APIs that crawl information regarding solar panel specifications from local solar panel providers
• Integration with NASA API to retrieve sunshine information, optimal angle information, solar irradiance given a location
• A Data model that forecasts the solar irradiance in the future and recommends the best solar panel setup for the users, leveraging on state-of-the-art statistical methods & data modelling tools. This was also benchmarked against the publicly available industrial information
• A Progressive Web App (PWA) that delivers all of the above and is accessible via any where by any device. In addition to this, rapid development an Android App that can be installed on any Android device (.apk file provided)

Resources

1. Project Sunroof: Project Sunroof
2. Solar Panel Installation and Maintenance Cost in Singapore [Updated 2021] - Solar AI Technologies
3. NSR | National Solar Repository of Singapore
4. Describe Forecast Dialog Box
5. Powerwall | Tesla
6. https://www.bca.gov.sg/GreenMark/others/handbook_for_solar_pv_systems_edited_copy.pdf
7. Understanding How Solar Power Is Measured - Solar Bay
8. Solar Panel Maintenance Costs | Solar PV Maintenance Cost
9. Insulation, Heating & Solar Panel Experts | YES Energy Solutions
10. Solar Panel Home Installation Cost
11. Impact of Degradation Rates on Solar PV Financing for Projects Located in the United States

Tools

• AWS API Gateway, AWS ECR, AWS Lambda Function, AWS S3, AWS Amplify
• PWA, Tableau Visualisation, Google Map API, NASA API
• Data Connectivity
• Web Data Connector Home
• How Forecasting Works in Tableau
• GITHUB: https://github.com/HackOnCloud/NasaAppChallenge2021

Appendix

#apps, #sunshine, #solar-panel, #practical, #green-energy

This project has been submitted for consideration during the Judging process.