Study of the granular fertilizers and the centrifugal spreader using Discrete Element Method (DEM) simulations

摘要

Since the beginning of production of mineral fertilizers,the worldwide crop yield has been increased dramatically, supplying the foodfor the exponentially growing population. In the 90’s of the 20thcentury however, a major concern about the environmental implications offertilizer spreading has been developed. Over-doses can harm the environment seriously,as well as the crop itself. As a consequence, the spreadability of fertilizerand the uniformity of the spread pattern has become an important topic in thispart of agricultural engineering. In Europe, 90% of the fertilizer spreading applicationsare performed by centrifugal spreading. The popularity of this technique lies intheir low price, easy maintenance, and large working width. However, its designis highly sensitive to machine characteristics as well as fertilizer particleproperties and weather conditions (e.g. wind). As a consequence, research aboutthe mechanism of spreading has become inevitable. A lot of effort on theinfluence of machine characteristics on the spread pattern has been done byOlieslagers (1997). In this thesis, the emphasis will be put on thestudy of fertilizer particles. These particles interact with the machine partsas well as with each other. Hereto, the “Discrete Element Method“ computationaltechnique is introduced. Basically, DEM calculates all possible interaction betweenobjects using force models based on spring – dashpot systems, and describestheir motion using classical dynamics. By changing the physical properties ofthe individual particles, such a method allows a comprehensive study of theparticle flow on a rotating disc. The physical properties of the fertilizer serveas the input parameters of the model. However, due to nature of fertilizerparticles, they are hard to determine. Several techniques have been proposed inthis thesis to provide either an accurate measurement or an acceptableestimation of these properties. The accuracy of the model will largely dependon value of the input parameter as well as the sensitivity to this parameter. Before trying to simulate the whole spreadingprocess, the interaction of a single particle with the machine parts (e.g.disc, vanes,..) of the spreader is considered. This step is primordial since itallows a decomposition of the physical problem as well as an easy validation ofthe simulation model with experiments. Having obtained this insight, simulationsof a particle flow on the disc (multi particle simulations) can be carried out.Both the performance of these simulations and the experimental validations areintensive tasks, requiring a lot of time. However, they are mandatory in order toget full insight in the parameters that affect the spread pattern and thepredictive value of the model. Generally, it was found that the DEM simulationsperform qualitatively well. The cause of the deviations (below 25%) betweensimulations and experiments remains uncertain but is likely to be found in theinsufficient knowledge of the initial conditions of the particles (e.g. outflowof the bin) as well as the contact force mechanisms. Inthe light of the accuracy of the measured particle properties and themodel output, a sensitivity study is carried out in this research. Themain goal herby is to gain insight in the parameters thatinfluence the spread pattern, and those who don’t. Although done in aqualitative way, such a study can provide valuable information for themanufacturer in the development of new fertilizer kinds or spreader designs. Inthis research, it was found that friction plays a crucial role in the dynamics ofthe particles on the disc, and the friction coefficients should be treated asvelocity dependent parameters. An important part of the errors made by themodel could be explained by this.