Nowadays, the textile industry has a massive environmental impact, involving depletion of resources in the production of garments and a very short life cycle of clothes. Indeed, there are usually produced, used and later thrown away without a proper recovery, pilling up in landfills. One of the main obstacles for an effective recycling of garments is the sorting phase, where workers selectively divide dismissed clothes by materials only considering the information on the labels, which are often missing. Thus, more efficient ways of sorting textiles must be explored to foster the transition of all the textile supply chain towards more sustainable models. The experience developed in recent years for the recognition of plastics can be translate to the textile sector. Specifically, the technology based on the use of Near Infrared (NIR) spectroscopy can recognize the material from its characteristic spectrum, which is a sort of fingerprint of its chemical structure, acquired in the near infrared region of light. Thus, the main objective of this work is to explore the capabilities and limitations of NIR optical systems in classifying textiles by materials.
An experimental study has been performed testing four NIR instruments in the department of Physics of Politecnico di Milano. Data analysis tools and recognition algorithms have been developed using samples of cotton and polyester, the most widely produced materials, but also on other pure and blended fibers. The main insights we could gain from the experimental work are that surface roughness does not alter the classification performance significantly and that NIR technology should operate in the wavelength range of 1100-1750 nm to prevent different colors of garments from masking the material information. However, some black pigments cannot allow a clear classification since they mask the NIR signal totally. Moreover, some types of fabric are not distinguishable: linen cannot be discriminated from cotton, whereas viscose can be confused with wool and nylon. Even if different compositions of polyester-cotton and polyester-elastane provide different spectra, polyester content below 20% and elastane content below 5% cannot be detected. The tested instruments have limited industrial applicability due to the long acquisition time (several minutes) and the manual introduction of samples inside the instruments.
To address possible improvements in the textile supply chain, industrial partners were interviewed thanks to the collaboration with the external partner Circularity. The interviews highlighted the difficulties in implementing a circular business model based on recycled fibers due to economic and technical issues, where quality, color and purity of the fibers are the most important parameters to take into consideration. Further on, a survey was proposed to more than 420 potential customers to understand the key factors influencing their choices when shopping clothes. According to their opinion, tactile feel of the garment, price and information on sustainability are high-impact characteristics to convince them to switch to more sustainable clothes. Thus, the final product should have recycled fibers pleasant to the touch and price comparable to virgin materials, as well as clear information regarding the decrease in environmental impact should be pointed out. Finally, 80% of respondents expressed the will to buy clothes made of recycled fibers, showing general interest in the circular model enabled by the sorting of textile waste using NIR technology.
Principal Academic Tutors
Gianluca Valentini, Physiscs department, Politecnico di Milano
External Institutions
Circularity
Team members
Alessandro Dalla Porta, Materials Engineering and Nanotech, PoliMi
Davide Mammana, Design and Engineering, PoliMi
Elisa Vasta, Biomedical Eng, PoliMi
Giacomo Graziano, Nanotechnology for ICT, PoliTo
Randeep Singh, Mathematical Engineering, PoliMi
Leonardo Gambarelli, Mechatronics Engineering, PoliTo