Influence of Input Biomass Conditions and Operational Parameters on Comminution of Short-Rotation Forestry Poplar and Corn Stover Using Neural Networks

Two kinds of biomass, short-rotation forestry (SRF) poplar and corn stover, were milled with a hammer mill at any combination of several levels of moisture content at mill inlet, wH₂O = 7–30%, feeding chip size, 20–50 mm, and under varied operational milling conditions, such as the opening sizes of the internal screen, dₜₐᵣgₑₜ = 2–5 mm, and the angular velocity of the hammers, 2000–3000 rpm, to determine their influence on the milling energy consumption and the physical properties of the milled product (moisture content, particle size, bulk density, and angle of repose). The use of neural networks allows one to obtain three-dimensional predictive surfaces for each output variable, which describe tendencies and behavior as a function of the milling conditions. The moisture content at the mill inlet and opening sizes of the screen were found to be the key variables in the process. Specific energy requirements per oven dry tonne of about 28 kWh for poplar and 22 kWh for corn stover were obtained with almost dry biomass (wH₂O = 7%) and dₜₐᵣgₑₜ = 5 mm, but this energy consumption increases 10-fold under opposite conditions, in which screen blocking was promoted by the high moisture content and the small opening sizes. Under these conditions (wH₂O ≈ 30% and dₜₐᵣgₑₜ = 2 mm), particle drying can remove half of the initial moisture because of the strong increase in residence time of the particle in the mill chamber but also barely 1% under the opposite milling conditions, with low moisture and bigger opening sizes. Additionally, a bulk density increase between 11 and 98% for poplar and between 89 and 360% for corn stover was registered in relation to the previous chipped form and as a function of the milling conditions. With regard to the angle of repose, a higher moisture content and particle size made the handling behavior worse for both biomasses.