Abstract
The development of anaerobic digestion involves both co-digestion of solid wastes and optimization of the feeding recipe. Within this context, substrate characterisation is an essential issue. Although it is widely used, the biochemical methane potential is not sufficient to optimize the operation of anaerobic digestion plants. Indeed the biochemical composition in carbohydrates, lipids, proteins and the chemical oxygen demand of the inputs are key parameters for the optimisation of process performances. Here we used near infrared spectroscopy as a robust and less-time consuming tool to predict the solid waste content in carbohydrates, lipids and nitrogen, and the chemical oxygen demand. We built a Partial Least Square regression model with 295 samples and validated it with an independent set of 46 samples across a wide range of solid wastes found in anaerobic digestion units. The standard errors of cross-validation were 90 mgO2⋅gTS−1 carbohydrates, 2.5 ∗ 10−2 g⋅gTS−1 lipids, 7.2 ∗ 10−3 g⋅gTS−1 nitrogen and 99 mgO2⋅gTS−1 chemical oxygen demand. The standard errors of prediction were 53 mgO2⋅gTS−1 carbohydrates, 3.2 ∗ 10−2 g⋅gTS−1 lipids, 8.6 ∗ 10−3 g⋅gTS−1 nitrogen and 83 mgO2⋅gTS−1 chemical oxygen demand. These results show that near infrared spectroscopy is a new fast and cost-efficient way to characterize solid wastes content and improve their anaerobic digestion monitoring.