NIR/PLS models to predict pretreated corn stover slurry solids composition, isolated and in-situ

Determining the composition of pretreated biomass slurry is necessary to understand the performance of the pretreatment step, but can be time consuming and cumbersome by conventional laboratory methods. We present Near Infrared (NIR) spectroscopy and Projection to Latent Structures (PLS) multivariate analysis models to predict the constituents of the solid phase of pretreated corn stover slurry, using both solids isolated from the slurry and solids still in-situ. The isolated solids model is precise and has correlations and uncertainties that suggest equivalency to traditional methods. The in-situ model utilizes whole slurry scans to predict isolated solids constituents. This model saves significant time and effort. We present model statistics and compare the two models, including independent validation samples.