Using Artificial Intelligence to Maximize the Efficiency of Cloudy Day Solar Reflectors
Amy Li
David Thompson Secondary
Floor Location : S 108 N

When someone says “Vancouver,” what is the first thing that comes to mind? Rain? Probably. Unfortunately, for solar power enthusiasts, rain means that solar panels produce only about 10-25% of their regular power, in part due to inefficiencies that become significant in low light environments.

The purpose of this project was to design a solar reflector that works on cloudy days. First, a pinhole camera-like apparatus was used to measure the extent of light scattering on cloudy days. This data was then implemented in a python program that simulated the process of natural selection to generate an efficient reflector shape. Finally, this design was 3D printed and tested under real overcast conditions.

The pre-design results indicated that most of the scattered light on cloudy days fall within 40 degrees of the incident light direction from the sun. This data was used by the program to generate a reflector shape that theoretically multiplies the light intensity by two with three times the area. When this design was tested outside, the short circuit current produced with the reflector was an average of 2.65 times the current produced without the reflector.

Taking shunt resistance into consideration, the power increase that results from concentrating sunlight onto a solar panel can be greater than the increase in illuminance, especially on cloudy days. A solar plant that uses this design could potentially reduce cost and increase efficiency.

There is room for much more improvement in this design, and room for much more improvement in the solar power industry as well. By harnessing new ideas, we can turn a dream of a solar future into a reality… even for places like Vancouver.