Sensor based optimisation of eddy current separation in bottom ash recycling

Abstract

Non-ferrous (NF) metals in municipal solid waste incineration bottom ash are mostly recovered by an eddy current separator (ECS) to retrieve the positive economic value and to strongly improve the environmental quality of the mineral ash contents before reusing them in road foundations or other applications. The splitter in an ECS is a metal plate that splits the falling product stream of the ECS in a mineral product and a NF metal product. Variations such as moisture content in the feed of the ECS make it necessary to continuously make adjustments to the splitter position in order to achieve the highest possible value recovery. The main research idea in this work is to adjust the splitter continuously by means of a sensor that can count both the metal particles as well as all the particles in the falling materials stream into the metals product. In this work the gain in recovery and retrieved value provided by the sensor was investigated and a preferred strategy is proposed to control the splitter by means of the information generated by the sensor. Particle trajectory tests were conducted to evaluate the changes in the trajectories due to variations in the bottom ash (BA) feed and to assess the possible consequences of these feed changes for metal recovery. The feed material used in the tests was the 1-6 mm size fraction, since ECS metal recovery from this fine fraction is the biggest challenge. The falling material stream from the ECS was intercepted by a 60 mm deep, flat container that was subdivided into multiple rectangular slots of 10 mm width. The material collected during 3 to 8 seconds of ECS processing was dried, weighted and classified and the NF metal particles in the different size fractions were analysed. Results from the ECS particle tests were used to simulate the system of ECS and sensor to assess how a sensor system would respond to variations in the BA. Optimal economic value recovery in the metal product is achieved at 29-32% metal product grade, which complies with 6% sensor grade. This sensor grade is measured by the sensor as it samples the falling materials stream through a 30 mm tube at the position of the splitter. A special result is that this optimum holds for both feed with 12% moisture and feed with 14% moisture. The count ratio of metal particles to all particles (z) was z=0.12 for both feed with 12% moisture and feed with 14% moisture content and the average particle mass ratio of non-metal to metal (k) was k=1.3 for feed with 12% moisture and k=1.7 for feed with 14% moisture. A reduction in moisture content of 2% will require an adjustment of 25 mm of the splitter away from the ECS to maintain the optimal economic value of the metal product. Vice versa, a 2% increase in moisture will require an adjustment of 25 mm of the splitter towards the ECS. The sensor grade, metal product grade and the count ratio all showed a monotonically increasing behaviour with increasing splitter distance, giving the ideal conditions for reliable automated control of the splitter by a the desired value-optimizing criteria. Simulations of ECS behaviour for different splitter positions showed that the sensor can add value to the metal product when compared to a human operator, who would adjust the splitter perhaps a few times per day, which is the common practice in industry. This conclusion holds for both feed with 12% and 14% moisture content, but the gain increases when the feed material becomes wetter. Moreover, control of the splitter as based on the sensor readings shows that the theoretical optimal value recovery in the metal product can be realized using the sensor.