Tabil, LopeRoberge, Martin2022-04-282021-112022-04-28November 2https://hdl.handle.net/10388/13926A proof-of-concept discrete element method simulation of a tillage operation in postharvest corn residue was desired within industry. Prerequisite to the fulfilment of this desire was the measurement of the material properties necessary to inform such a simulation, as well as code capable of producing virtual postharvest corn residue suitable for discrete element method simulation software. Corn leaves a large amount of residue in the field after harvest- tillage is used to manage this residue. The discrete element method is an increasingly promising way to simulate postharvest corn residue for the purpose of designing tillage equipment. Postharvest corn residue samples were collected in 2017 and 2018 from Illinois fields seeded with Beck’s 6165AM hybrid. The roots, stalks, stalk piths, stalk rinds, cobs, husks, leaf sheaths, leaf midribs, and leaf blades were evaluated using three-point bending tests, shear tests, and tension tests. Simulations of three-point bending were used to validate virtual roots, stalks, and husks. Code to produce randomly generated parametric postharvest corn residue compatible with the discrete element method was written. The values for particle density, moisture, ultimate load, second moment of area (at the cross-section subject to load), proportional limit stiffness, secant stiffness, proportional limit modulus, secant modulus, yield strength, and ultimate strength were calculated for the shear and three-point bending trials. Generalized additive models predicting ultimate load as a function of particle density and the second moment of area were developed for roots, stalks, husks, and leaves in both three-point bending and shear. Virtual postharvest corn residue (randomly generated, parametrically defined) was inserted into a virtual soil bin and a simplified tillage operation was simulated. The findings of this work indicate that it would be beneficial to develop an algorithmically driven parametric virtual field model that fills itself in an organic manner such that statistically modeled physical properties are automatically tied in a well-understood manner to the particles used for simulations. Such (machine learning) algorithms are simple to implement and improve once written, relative to manual data processing.application/pdfDEMDiscreteDistinctElementMethodTillageTillSimulationVirtualPostharvestCornMaizeResidueChaffGeneralizedAdditiveModelThreePointBendingShearTensionTensileCompressionFlexuralTestTestsTrialTrialsParametricHybridParticleStiffnessDensityMoistureYieldStrengthModulusRootStalkHuskLeafMidribBladeSoilStatisticalMachineLearningFieldGeometryPredictiveStrainGaugesPropertiesCharacterizationLoadMechanicalBondedPithTensorElasticityAxialTransverseThesis2022-04-28