Image Analysis for Rapid Assessment and Quality-Based Sorting of Corn Stover

Josephine N Gruber,Rachel D Colby,Kuan-Ting Lin,Rachel M Emerson,Ling Ding,Jordan L Klinger,William A Smith,Amber N Hoover,Brad J Thomas,Allison E Ray,Bryon S Donohoe

doi:10.3389/fenrg.2022.837698

Josephine N Gruber, Rachel D Colby + Show 9 more

Open Access

https://doi.org/10.3389/fenrg.2022.837698

Copy DOI

Abstract

Imaging in the visible spectrum is a low-cost tool that can be readily deployed for in-field or over-belt monitoring of biomass quality for bio-refining operations. Rapid image analysis coupled with innovative preprocessing may reduce the impacts of feedstock variability through identification of contaminants or other material attributes to guide selective sorting and quality management. Image analysis was employed to evaluate the quality of corn stover in red-green-blue (RGB) chromatic space. This study used controlled, bench-scale imaging as a proof-of-concept for rapid quality assessment of corn stover based on variations in material attributes, including chemical and physical attributes, that relate to biological degradation and soil contamination. Logistic regression-based classification algorithms were used to develop a method for biomass screening as a function of biological degradation or soil contamination. This study demonstrated the use of image analysis to extract features from RGB color space to investigate variations in critical material attributes from chemical composition of corn stover. Fourier transform infrared (FT-IR) suggested a correlation between red band intensity and biological degradation, while detailed surface texture analysis was found to distinguish among variations in ash. These insights offer promise for development of a rapid screening tool that could be deployed by farmers for in-field assessment of biomass quality or biorefinery operators for in-line sorting and process optimization.

Highlights

The 2016 Billioin Tons Report (BT16) estimates that by 2040, more than 1 billion tons of biomass will be available to achieve a vision of a sustainable bioeconomy (US Department of Energy (DOE), 2016)
Hierarchical cluster analysis based on chemical components was used to glean key insights about sources of variability that affect quality, with samples grouping into four distinct clusters observed through principal component analysis (PCA) (Figure 3)
The Fourier transform infrared (FT-IR) results demonstrated a potential correlation between the hydrolyzed carbohydrates and the condensed and oxidized lignin in biologically degraded corn stover

Summary

Introduction

The 2016 Billioin Tons Report (BT16) estimates that by 2040, more than 1 billion tons of biomass will be available to achieve a vision of a sustainable bioeconomy (US DOE, 2016). Variations in lignocellulosic biomass material and quality attributes are often overlooked when assessing feedstock value and pathways for conversion to fuels, chemicals, and products (Ray et al, 2020). Hoover et al (2019) developed several multiple regression models where five chemical characteristics could be used to estimate biochemical conversion performance Using these models, an approach for a grading system was demonstrated that could be used to inform markets on the impacts of biomass variability. Hartley et al (2020) used discrete event simulation that investigated feedstock quality on plant uptime and overall impact to biofuel cost through feedstock delivery cost These works show how fundamental biomass information might enable real-time decision making on plant profitability and operability. In addition to capital and operational costs, it is imperative to consider the impact of storage method and format on chemical properties of biomass and overall process efficiency

Methods

Results

Conclusion