O PAPEL VOL 77 - PP 78-81 - JUL 2016

ABSTRACTA real-time in-line stickies, dirt and contaminant detection sensor suitable for mill environment is presented in this work. Stickies, in this context, are defined as contaminants in a pulp sample with at least one image attribute (such as translucency, hue, saturation, etc.) different from the overall sample. Measurement trends from the in-line sensor were validated against existing stickies measurements in multiple paper mills before operating decisions were made entirely based on this real-time measurement. Mills where in-line stickies measurements have been implemented have, on average, been able to make process decisions 8-10 hours faster than laboratory measurements. ANDRITZ PulpVision is an in-line sensor, built on a machine learning algorithm to detect and classify contaminants in a pulp stream. Another mode of PulpVision is to measure fiber morphology and shives in pulp. This application is essential for monitoring pulp quality and performance of unit operations such as refiners. Detection accuracy is not affected by the presence of bubbles, flocculation, and consistency variation in pulp samples. 

A trial using PulpVision was conducted throughout an old corrugated container (OCC) plant to generate profiles of stickies and dirt. It was found that in this specific recycling line, Primary fine screens (PFS) were tested to be inefficient for stickies removal, while cleaners did not play a positive role in dirt removal. This stickies sensor is tested to be powerful to provide real-time feedback of the equipment performance in paper recycling plants.

*Authors references:1. ANDRITZ Automation, Decatur, GA, USA2. ANDRITZ Brasil Ltda., Belo Horizonte, Minas Gerais, Brazil