MASS FLOW SENSOR DEVELOPMENT FOR AN AIR SEEDING CART

The air seeding cart is an important piece of farming equipment used in the seeding process. Three factors which are necessary to monitor during the seeding process are the seeding rate (material mass flow rate), air flow rate, and blockages. In current practice, there are systems that monitor and report air flow and blockages but not the actual seeding rate. Presently, the seeding rate is based on the metering calibration before the seeding process starts, which requires a lot of time and energy from the operator. If that goes wrong, it not only takes longer, but also costs more money and increases the already significant stress and fatigue which farmers and operators have during the seeding period. Therefore, the development of reliable, and easily calibrated, on-line sensors for flow monitoring would be beneficial. Further, such sensors would facilitate closed-loop control of the flow rate itself. In order to develop a laboratory prototype for mass flow measurement, a model for mass flow estimation was established. This was accomplished by using pressure transducers to determine the pressure drop across an elevation in the primary air cart run (between the air seeding cart and the air hoe drill). An air seeding test station was designed and developed for the study. Three different types of seeds and a granular fertilizer were chosen and tested. These tested materials were canola, wheat, chickpea and urea fertilizer (46-0-0). The general form of the model was developed using data from the canola tests. The input parameters for this mass flow estimation model were pressure drop and air flow information. The average percent error of the material mass flow rate’s full range was under 10%, except for the highest rate which tested up to 20%. Overall, more than 75% of the estimations had percent errors being less than 5%. The form of the model was also applicable to other individual tested materials with the percent error of their full ranges up to 20%. However, their average of their median error was around 5% of their full ranges. The general model was also applied to the combined data from all tested materials. The results were not as accurate as when the model was applied to the individual tested material. The median of the percent error (of material mass flow rate full range) varied from as low as 1% to as high as 30%, depending on the tested materials. Nevertheless, it demonstrated that there were consistencies between the behaviour of the four tested materials