Shriti Dham

Utilizing Machine Learning to Find a More Sustainable Alternate to Polyester in Textiles

Started Mar. 28, 2024

Abstract or project description

Polyester is a type of thermoplastic polymer that can be found in a variety of different commodities, but it is popularly applied to the production of clothing. Each year, more than 50 million tonnes of polyester are used in textiles alone. The mass production of textiles has created several concerns for the environment and its future, as the production of polyester fabrics increases greenhouse gases in the atmosphere, and the waste of polyester products leads to microplastics, which are found in wildlife and the oceans. Polyester is a form of plastic, meaning it does not degrade easily, and this issue coupled with its other environmental impacts makes its effect long lasting and difficult to resolve. The use of polyester in varying products, but especially in clothing, is extremely common due to its versatility. This specific polymer is known to be durable, elastic, hydrophobic, a thermoplastic, and especially cheap to produce and buy. The aim of the project is to find an alternative polymer that has the same, or similar, performance and characteristics as polyester, but with a considerably lower impact on the environment and surroundings. Machine learning is advantageous in this project due to its ability to consider vast amounts of data, relatively quickly. Through the process of machine learning, several properties of new polymers can be predicted. First, graph neural networks are trained to predict polymer properties using existing databases. Then, these models are used to predict the property metrics of new, more sustainable polymers. This would result in the creation of novel polymers or natural materials to answer if polyester can be successfully replaced in textiles.