Mechatronic Design and Control of a Waste Paper Sorting System for Efficient Recycling PI: Richard A. Venditti and M. K. (Ram) Ramasubramanian Industry Partners: Weyerhaeuser (Tom Friberg) MSS (Michael Grubbs)
Technology Description The primary challenge in the recycling of paper is to obtain raw material with the highest purity. Highly sorted paper stream will facilitate high quality end product, and save processing chemicals and energy. Current manual sorting techniques are not effective in reducing landfill waste. The project goal is to develop sensors for sorting grades of paper and board from a mixed stream automatically at high speed for more efficient recycling.
Project Strategy Key technical barriers. Development of stiffness measurement in real-time on free, non-oriented samples, and inferring the type of paper based on this information is a challenging problem. Combination of sensing technologies in real-time, namely, lignin, stiffness, color, and adhesives to come up with the sorting scheme. Describe your project s strategy for overcoming these barriers Investigating alternative stiffness measurement methods Developing improvements to the current stiffness measurements Evaluating other sensing techniques, gloss, color, IR temperature sensor for stickies identification Criteria for go/no-go decisions using Neural Networks. Sensor must be fast Sensor must provide info on paper type Sensor must be economical to implement Sensor must be rugged
Project Partners MSS, Inc., Michael Grubbs, General Manager Provides access to pilot facilities Provides feedback on progress reports Communicates needs of the industry Designs, manufactures and sells sorting equipment Weyerhaeuser, Tom Friberg, Researcher Provides feedback on progress reports Communicates needs of the industry Provides paper recycling perspective on research direction and progress
Commercialization Potential Market: Any recycling facility involved in the sorting and/or disposal of waste paper. Commercialization: As we develop prototype sensors we are testing them in a commercial environment with the industrial partner The review of commercial trials guide further work We have taken this approach successfully with the lignin sensor and have it commercialized. We have done similar trials with the stiffness sensors and have identified areas of improvement. Currently working on the stiffness sensor and sensor integration as two achievable milestones in the coming year.
Commercialization Status A lignin sensor has been commercialized by the industrial partner. MultiWave Sensor with Lignin Sensor operating at IMS Recycling in San Diego, CA for removal of OCC, Carrier-Board, plastics and trash from newspaper. The Lignin Sensor set-up for Carrier-Board identification in San Diego, CA is crucial for properly identifying the targeted materials. Unit shipping February 2006 to VISSER Waste Management in Udenhout, Holland. Two units that are being fabricated and will ship March 2006 - to Stora Enso in Cologne, Germany and to Cougle Recycling in Hamburg, PA. It is projected that the stiffness sensor will be commercialized in 2008.
Energy Savings Electricity: 10 million kwh for the US Industry. A 5% decrease in rejected recycled pulp may occur by recycling all sorted recovered paper rather than mixed. Up to 1% of the total amount of all paper and paperboard produced is rejected due to quality problems with recycled fibers. The rejected paper is typically re-pulped, blended at low level with fresh paper stock material and fed back to the paper machine. The use of higher quality pulp from recycled sorted recovered paper may decrease the 1% reject level.
Other Benefits to using sorted paper in recycling Makes recycling more cost effective and efficient promotes increased recycling rates reduce the need for virgin fibers reduce paper waste sent to landfills Utilization of sorted papers in recycling processes will decrease the amount of sludge and rejects generated in recycling Utilization of sorted papers in recycling processes will decrease the amount of water needed to produce recycled paper
Automated Paper Sorting System Lignin sensor Stiffness sensor Gloss sensor Color tracking sensor Stickies Sensor Decision making algorithm Actuating mechanism
Lignin Sensor The sensor measures lignin fluorescence when excited in the visible region. Newsprint samples, typically containing high lignin, produce high intensity. Ledger printing and writing grades with low lignin content produce low-fluorescence intensity. Gives normalized lignin content in paper Sensor output can be used as an input for the control algorithm
Lignin sensor-dynamic performance The sensor is able to identify papers moving at high speeds and is quite robust for sorting applications. The sensor can be successfully used as a part of a multi-sensor system to sort mixed office waste for more efficient recycling.
Stiffness Sensor Can be used for differentiating different grades of paper based on their relative bending stiffness values Can work together with Lignin and Gloss detection sensors for better sorting Is more useful for sorting cardboard from mixed paper feed when compared to other sensors
Stiffness Sensor Design Constraints Should be non-contact in nature Short response time Should be compatible with the existing conveyor systems
Current Techniques for Stiffness Measurement Contact Methods: Contact transducers generate ultrasonic waves on the surface of the paper Excessive noise due to mechanical vibrations is a problem Finer grades and paper boards are difficult to identify using this method
Current Techniques for Stiffness Measurement Non Contact Methods: Air coupled piezoelectric transducers, air coupled capacitive transducers Poor coupling of energy between the transducer and the paper surface Hard to implement online Laser ultrasonic measurement technique is an exception
Why need different sensor design? All the previously mentioned techniques are for testing paper webs of almost constant thickness These methods are aimed at calculating the exact elastic constants Equipment is complex For sorting there is no need to find the elastic constants Unlike paper webs, the thickness of paper on a sorting conveyor varies widely from one sample to another.
Stiffness sensor setup
Distance sensor Non-contact in nature High resolution Output is linearly proportional to the distance Output is not affected by target s optical properties
Distance sensor performance
Microcontroller Controls the solenoid valve timing A/D conversion of distance sensor output Varies the load by varying the load timing Runs the control algorithm Identifies the samples based on the output of the algorithm
Parameters which influence the deflection Orientation of the sample with respect to the conveyor belt (machine direction, cross machine direction) Thickness of the sample Basis weight Modulus of elasticity Distance between the supports Intensity of the loading Conveyor speed Coefficient of friction of the conveyor belt
Static stiffness sensor Paper samples sitting on fixed supports are loaded pneumatically Samples with various elastic properties are studied Deflection values are obtained for these samples at a given load Variation of the deflection with respect to various parameters is studied
Nozzle pressure profile Pressure profile of the nozzle that was used for static testing of paper samples
Static testing results
Static testing results
Pilot plant trials of stiffness sensor To better understand the problems involved during the operation of the sensor, the stiffness sensor was tested on a high speed moving conveyor The dynamic response of the stiffness sensor was evaluated on a moving conveyor at the MSS Inc, Nashville, TN research/manufacturing site Load on top of the sample was applied by the air jet from flat fan nozzle
Pilot plant trials of stiffness sensor Static Test Dynamic Test Flat fan nozzle Distance Sensor Conveyor speed = 280 ft/min
Dynamic test results 30 25 Def l ection, mm 20 15 10 Nozzle height=1" Nozzle height =7" 5 0 Copy Paper Yellow Ruled Paper Filter Paper Medium Card Stock Heavy Card Stock Speciality Card Stock Card board Nozzle inlet pressure = 10 psi, Samples were loaded in MD
Stiffness sensor characterization Step 1: Identifying different grades of paper which are commonly found in the recovered paper Testing the selected grades of paper to find the mechanical properties Step 2: Building an FEA ( Finite Element Analysis) model of the system Simulating the original loading and boundary conditions of the system Using the simulation output for decision making
Paper samples material data Four samples of different grades are picked and their mechanical properties are investigated in order to build the FEA model Material test data for 105µm thick sample
Paper samples test data Paper grade Copy paper Medium card stock Heavy card Stock Specialty card stock Thickness ( µm) 105 206 229 234 Grammage (g/m 2 ) 75 145 200 175 Modulus of elasticity in machine di rection (GPa) 3.98 1.6898 1.7935 1.6103 Modulus of elasticity in cross machine direction (Gpa) 1.27 1.1143 1.090 1.1123
FEA model An FEA model of the system is constructed Paper samples are modeled as orthotropic shell elements Material test data is used to create the material model Large displacement formulation is used for the elements The conveyor supports are modeled as rigid bodies The material properties of the actual samples are used in the model Actual Loading and boundary conditions are simulated
FEA model Conveyor-2 Conveyor-1 Paper sample Gap = 40mm
Paper Orientations on Conveyor
Finite Element Simulations Conveyor Speed Orientation Nozzle Pressure 10psi 20psi 300 ft/min MD MD-30Degrees MD-60Degrees CD 25psi 30psi 10psi 20psi 25psi 30psi 10psi 20psi 25psi 30psi 10psi 20psi 25psi 30psi
Conveyor Speed = 300 ft/min, MD Response of 105µm paper sample to applied load; conveyor speed =300 ft/min, load = 20psi
Conveyor Speed = 300 ft/min, MD Response of 229µm paper sample to applied load; conveyor speed =300 ft/min, load = 20psi
Conveyor Speed= 300 ft/min, MD
Conveyor Speed = 300 ft/min, MD 30 Response of 105µm paper sample to applied load; conveyor speed =300 ft/min, load = 20psi
Conveyor Speed = 300 ft/min, MD 30 Response of 229µm paper sample to applied load; conveyor speed =300 ft/min, load = 20psi
Conveyor Speed = 300 ft/min, MD 30
Conveyor Speed = 300 ft/min, CD Response of 105µm paper sample to applied load; conveyor speed =300 ft/min, load = 10psi
Conveyor Speed = 300 ft/min, CD Response of 229µm paper sample to applied load; conveyor speed =300 ft/min, load = 10psi
Conveyor Speed = 300 ft/min, CD
Time Response Curves, MD-300ft/min
Conveyor Speed = 1200 ft/min, MD Response of 105µm paper sample to applied load; conveyor speed =1200 ft/min, load = 10psi
Conveyor Speed = 1200 ft/min, MD Response of 229µm paper sample to applied load; conveyor speed =1200 ft/min, load = 10psi
Conveyor Speed = 1200 ft/min, MD
Damping caused by the surrounding air Response of the paper when there is no viscous pressure acting on top of it
Damping caused by the surrounding air Response of the paper when there is viscous pressure acting on top of it
Response of the sample to pneumatic load Applied pneumatic load is equal to the load applied by the cylindrical nozzle operating at 5 psi and held 1 above the conveyor surface, conveyor speed=1200ft/min
Time Response Curves, MD-1200ft/min
Comparison of Response Curves 20psi-MD-300 20psi-MD-1200
Future Work Stiffness sensor development completion Use of RF sensors for fast response and higher speed sorting Commercialization of stiffness sensor to identify carrier boards and other stiff materials IR imaging based sensors for stickies identification Neural network control algorithm implementation
Flutter of paper Flutter can be defined as the dynamic instability of an elastic body in an air stream The vibration modes of the samples subjected to lateral load depend on the elastic constants of the samples For a given value of tangential load, stiff samples vibrate at a much lower frequency whereas flexible thin samples vibrate with larger amplitudes and higher frequencies
Flutter based sorting setup Tangential load
Stiffness sensor performance enhancement The use of frequency domain analysis (web flutter in a fluid flow) to compliment the results obtained from the deflection data. Results from the classic Flag Flutter problem show that the flutter frequency is related to the bending stiffness as shown. This method also eliminates the requirement for the paper samples to be at a constant height from the sensor, thereby making it more robust
Stiffness Sensor - Performance Enhancements Potential Sensors for frequency analysis Laser Distance Sensor from LMI Technologies (USA), Inc Resolutions down to 0.001mm Standardized with optical filters to reduce the influence of ambient light High speed, Analog outputs (V), up to 100 khz Optional modulated version (-M) to exclude any influence from external light Fast laser intensity control for object color changes LK-G series from Keyence (USA), Inc Resolutions down to 0.01micrometer High speed, Analog outputs (V), up to 50 khz Resistant to ambient lighting conditions Flutter frequency measurements which can then be used to correlate to the stiffness measured from the displacement data.
Future Needs Development of the stiffness sensor can be completed by June 2007 and commercialization can be accomplished by January 2008. Exploration and development of the IR sensor is very useful and can be a new project for potential support. Funding runs out end of 2006. Additional support for one year can significantly influence the outcome of this research.
Acknowledgement This research was supported by the U.S. Department of Energy under the Industries for the Future Program, Forest Products Industry Agenda 2020; project number DE-FC07-00ID13880