(452f) Enhanced Feedstock Characterization and Modeling to Facilitate Optimal Preprocessing and Deconstruction of Corn Stover

Chemical and physical heterogeneity in herbaceous biomass feedstocks such as corn stover are due to substantial differences in plant tissue types. These differences can contribute significant challenges to handling, preprocessing, and conversion in biorefining processes. A second, related challenge is the problem of quantifying this heterogeneity within feedstocks as it relates to processing in a biorefinery (e.g., handling in screw feeders, response to particle size reduction, response to pretreatment and enzymatic hydrolysis, solid-liquid separations). In this talk, we will describe results on novel biomass fractionation technologies and developing improved characterization tools for assessing biomass heterogeneity and the responses to fractionation. Different corn stover anatomical fractions (cob, rind, stem pith, leaves, etc.) are shown to exhibit significant differences in chemical composition, bulk physical properties and respond differently to pretreatment and enzymatic hydrolysis. Through combinations of comminution, sieving, and air classification, we demonstrate several preprocessing strategies for recovering corn stover fractions enriched or depleted in different anatomical components that have the potential to facilitate streamlined processing in a biorefinery. Next, we develop physics-based predictive models for predicting the partitioning of corn stover anatomical components during air classification by differences in properties and validate that these models can be used to optimize these separations to achieve target processing outcomes. Finally, by coupling near-infrared (NIR) spectroscopy to machine learning tools, we develop models that can be used to predict not only composition, but the relative abundance of anatomical components.