Final Report Summary - SAGT (Self-adjusting gravity table using imaging analysis for quality control)
Final publishable summary report
Background and objectives
Removal of contaminants from ‘food grade’ quality grains is of great importance in food and grain processing. Man has progressively taken steps towards improving the sorting of the grains by quality by developing several fundamental cleaning operations. Historically, ‘winnowing’ has been employed to remove bran and stones from the harvested crop and, over the years, this ancient practice has been refined and developed into modern machinery, which clean harvested crops in very large volumes. The cleaning takes place in several steps from pre-cleaning, over grading/sieving, to fine-cleaning which are achieved using an assortment of machinery including gravity tables. This project is focused on the design and performance optimizations of such gravity tables.
These tables need to be monitored and adjusted regularly in order to perform optimally and this requires supervision and adjustment of skilled operators with years of experience. Sustaining high quality throughput from these machines is therefore labor intensive and the need for human intervention would mean that the overall grain quality of a particular table would be subjective. For these reasons, there is a wish to develop a system that automatically and objectively measures the performance of the machine and thereby enables a fully Self-Adjusting Gravity Table (SAGT).
Primary results
There is a need for the continuous improvement in both design and process optimization of technologies such as gravity tables, aiming at obtaining more controllable and robust systems. The knowledge gap in the description of the complete physics of gravity separation hampers such a concerted effort. Hence, the logical first steps during such a project would be to evaluate the entire operation of the equipment from ground-up, establishing relevant literature and results that can benefit the overall industry. Correspondingly, research initiatives were undertaken using a two-pronged approach to simultaneously improve the general literature in both multi-spectral imaging in grain applications and flow physics of grain separation.
The first research efforts showed that multispectral imaging devices are capable of measuring significant (and relevant) differences in the grain product profiles, and furthermore these differences actually correlate with grain properties such as density, shape etc. The subsequent studies conducted showed that the weight of grains (in this case wheat) could be accurately estimated from multispectral images captured using the VideometerLabTM instrument. The use of 5 wavelengths was adequate to ensure a sufficiently precise model for estimating the weight of the sampled grains. From these individual grain weights, the statistics of the weight distribution of grains across the heavy and light outlets was used to calculate a single measure of how well the gravity table was performing. This statistical measure, similar to the t-statistic (used in the Students t-test), describes by a single measure the separation of the weight distributions from the heavy and light outlets. Optimizing the performance of the gravity table is equivalent to maximizing the value of this measure.
Parallelly, efforts were directed towards better understanding the physics of gravity separation in order to establish a complete operating profile of gravity separators. These general results were then extrapolated to provide a holistic overview of gravity table operation. To this end, a coupled CFD-DEM numerical framework that facilitates the complete resolution of the relevant granular flow physics, was developed during the course of the project. This framework was instrumental in developing a fundamental understanding on the movement of grains on a gravity table, i.e. identification of segregation patterns and preferred zones of particle accumulation. A number of extrinsic operational factors were examined in order to prescribe the best operating conditions for the table. Additionally, the accuracy of this framework was assessed using a set of carefully curated experimental studies on a lab-scale gravity separator. At a quantitative level, experiments and simulations co-related well, and the developed framework was able to capture the main features of granular flow along with the accompanying segregation phenomena to a good degree of detail. It should be noted that this general framework to study granular flow can be extended to describe any phenomena involving grains/particles and is not restricted only to gravity separation.
The wealth of comprehensive findings that were generated during the course of the two-pronged research initiative culminated into the design of a system, which can be attached to an existing PLC controlled gravity table and automatically adjust the primary settings of the table, so that its output is always near the optimum. This system replaces the need for a skilled human operator by a computerized system that tirelessly monitors the output, adjusts the table, and ensures a consistent and objective grain quality. The entire system consists of the following components:
• Gravity table (Westrup KA-1200) with PLC interface for changing its settings.
• A mechanical system for sampling heavy and light grains from each end of the table.
• A multispectral imaging device (VideometerLabTM) for imaging the sampled grains at a selection of wavelengths.
• A mechanical system for presenting the grains to the imaging device.
• A PC with software for activating sampling, image analysis, measuring the performance of the table, and adjusting the settings of the table for optimal performance.
The SAGT software plugin has a simple UI with the capability to either read stored images (for re-running scenarios) or directly reading material off the line. The user-friendly interface provides a summary of the real-time measurements in an output table, and the corresponding console depicts the control scheme effected towards the system. A full scale proto-type of the SAGT machine was developed using these constitutive elements and a performance study was conducted during the last part of the project for evaluating if the proposed measure (the t-statistic) was indeed useful for optimizing the output of the table. The three primary settings of the table, side tilt, air velocity and eccentric speed, were varied one by one over a relevant interval, and the value of the t-statistic calculated. The study showed that the quality measure always reached a distinct maximum value at a specific setting for each of the three parameters, thus verifying that a simple search for the settings with the best performance would indeed be feasible.
Socio-economic impact
When the technology developed in the SAGT project is implemented as an add-on to gravity tables on a larger scale, the effect would be an improvement of the quality of the output of these sorting machines. This would result in a more precise separation of good and inferior grains, improving both the performance and profitability of the entire grain and seed cleaning line. A better and more efficient use of resources would lead to a more sustainable society. With the ever-increasing demands for more efficient use of the worlds resources, optimizing the quality sorting of both food and seed grain production could have significant positive societal implications particularly in view of the growing hunger crisis around the world. The food and grain industry need to be at their efficient best in order to tackle this growing problem and the SAGT project could serve as the stepping stone towards a more smarter and efficient tomorrow.
Results are being disseminated as publications (journals and conferences papers/posters) and as short reports on the SAGT-project website http://sagt-iapp.com/.
Background and objectives
Removal of contaminants from ‘food grade’ quality grains is of great importance in food and grain processing. Man has progressively taken steps towards improving the sorting of the grains by quality by developing several fundamental cleaning operations. Historically, ‘winnowing’ has been employed to remove bran and stones from the harvested crop and, over the years, this ancient practice has been refined and developed into modern machinery, which clean harvested crops in very large volumes. The cleaning takes place in several steps from pre-cleaning, over grading/sieving, to fine-cleaning which are achieved using an assortment of machinery including gravity tables. This project is focused on the design and performance optimizations of such gravity tables.
These tables need to be monitored and adjusted regularly in order to perform optimally and this requires supervision and adjustment of skilled operators with years of experience. Sustaining high quality throughput from these machines is therefore labor intensive and the need for human intervention would mean that the overall grain quality of a particular table would be subjective. For these reasons, there is a wish to develop a system that automatically and objectively measures the performance of the machine and thereby enables a fully Self-Adjusting Gravity Table (SAGT).
Primary results
There is a need for the continuous improvement in both design and process optimization of technologies such as gravity tables, aiming at obtaining more controllable and robust systems. The knowledge gap in the description of the complete physics of gravity separation hampers such a concerted effort. Hence, the logical first steps during such a project would be to evaluate the entire operation of the equipment from ground-up, establishing relevant literature and results that can benefit the overall industry. Correspondingly, research initiatives were undertaken using a two-pronged approach to simultaneously improve the general literature in both multi-spectral imaging in grain applications and flow physics of grain separation.
The first research efforts showed that multispectral imaging devices are capable of measuring significant (and relevant) differences in the grain product profiles, and furthermore these differences actually correlate with grain properties such as density, shape etc. The subsequent studies conducted showed that the weight of grains (in this case wheat) could be accurately estimated from multispectral images captured using the VideometerLabTM instrument. The use of 5 wavelengths was adequate to ensure a sufficiently precise model for estimating the weight of the sampled grains. From these individual grain weights, the statistics of the weight distribution of grains across the heavy and light outlets was used to calculate a single measure of how well the gravity table was performing. This statistical measure, similar to the t-statistic (used in the Students t-test), describes by a single measure the separation of the weight distributions from the heavy and light outlets. Optimizing the performance of the gravity table is equivalent to maximizing the value of this measure.
Parallelly, efforts were directed towards better understanding the physics of gravity separation in order to establish a complete operating profile of gravity separators. These general results were then extrapolated to provide a holistic overview of gravity table operation. To this end, a coupled CFD-DEM numerical framework that facilitates the complete resolution of the relevant granular flow physics, was developed during the course of the project. This framework was instrumental in developing a fundamental understanding on the movement of grains on a gravity table, i.e. identification of segregation patterns and preferred zones of particle accumulation. A number of extrinsic operational factors were examined in order to prescribe the best operating conditions for the table. Additionally, the accuracy of this framework was assessed using a set of carefully curated experimental studies on a lab-scale gravity separator. At a quantitative level, experiments and simulations co-related well, and the developed framework was able to capture the main features of granular flow along with the accompanying segregation phenomena to a good degree of detail. It should be noted that this general framework to study granular flow can be extended to describe any phenomena involving grains/particles and is not restricted only to gravity separation.
The wealth of comprehensive findings that were generated during the course of the two-pronged research initiative culminated into the design of a system, which can be attached to an existing PLC controlled gravity table and automatically adjust the primary settings of the table, so that its output is always near the optimum. This system replaces the need for a skilled human operator by a computerized system that tirelessly monitors the output, adjusts the table, and ensures a consistent and objective grain quality. The entire system consists of the following components:
• Gravity table (Westrup KA-1200) with PLC interface for changing its settings.
• A mechanical system for sampling heavy and light grains from each end of the table.
• A multispectral imaging device (VideometerLabTM) for imaging the sampled grains at a selection of wavelengths.
• A mechanical system for presenting the grains to the imaging device.
• A PC with software for activating sampling, image analysis, measuring the performance of the table, and adjusting the settings of the table for optimal performance.
The SAGT software plugin has a simple UI with the capability to either read stored images (for re-running scenarios) or directly reading material off the line. The user-friendly interface provides a summary of the real-time measurements in an output table, and the corresponding console depicts the control scheme effected towards the system. A full scale proto-type of the SAGT machine was developed using these constitutive elements and a performance study was conducted during the last part of the project for evaluating if the proposed measure (the t-statistic) was indeed useful for optimizing the output of the table. The three primary settings of the table, side tilt, air velocity and eccentric speed, were varied one by one over a relevant interval, and the value of the t-statistic calculated. The study showed that the quality measure always reached a distinct maximum value at a specific setting for each of the three parameters, thus verifying that a simple search for the settings with the best performance would indeed be feasible.
Socio-economic impact
When the technology developed in the SAGT project is implemented as an add-on to gravity tables on a larger scale, the effect would be an improvement of the quality of the output of these sorting machines. This would result in a more precise separation of good and inferior grains, improving both the performance and profitability of the entire grain and seed cleaning line. A better and more efficient use of resources would lead to a more sustainable society. With the ever-increasing demands for more efficient use of the worlds resources, optimizing the quality sorting of both food and seed grain production could have significant positive societal implications particularly in view of the growing hunger crisis around the world. The food and grain industry need to be at their efficient best in order to tackle this growing problem and the SAGT project could serve as the stepping stone towards a more smarter and efficient tomorrow.
Results are being disseminated as publications (journals and conferences papers/posters) and as short reports on the SAGT-project website http://sagt-iapp.com/.