Development of Data Processing Scheme for Plastic Identification Instruments

60608 Rapid and accurate identification of the type of plastic waste during sorting operations is essential for producing good quality recycled products at competitive prices. Although instruments are available for automated sorting of plastics, they have not been universally adopted. Even the most sophisticated use crude and inefficient data processing methods, which limit speed and accuracy. This project will develop data processing methods (identification algorithms) for use with these instruments, which make spectrally resolved optical reflectance measurements. The new algorithms will enhance the speed and accuracy of identification and provide the ability to sort between plastics with small variations in composition. Phase I obtained a spectral database of commonly used plastic containers and determined the optimal number and location of spectral bands necessary for the identification of each type of plastic. In a comparison with a commercial instrument, the new algorithm had an accuracy of 100% versus 83% for the commercial instrument. Phase II work will be conducted in association with a leading manufacturer of automated sorting equipment to implement the algorithm for the identification of the five major types of plastics. The result will be a final design for implementation of the algorithm in a commercial system. Commercial Applications and Other Benefits as described by the awardee: The data processing methods should increase the speed and accuracy of systems that identify waste plastics prior to recycling. Emphasis will be given to the identification of common household plastics processed in materials recovery facilities. Improvements in sorting should reduce the cost of recycling and raise the value of recycled plastics by increasing the purity of recovered product. Sorting of industrial plastics such as automobile parts also could benefit from the use of faster and more accurate identification.