Predicting the biochemical methane potential of wide range of organic substrates by near infrared spectroscopy.

The use of near infrared spectroscopy (NIRS) as an alternative method to predict the biochemical methane potential (BMP) of a broad range of organic substrates was investigated. A total of 296 samples including most of the substrates treated by anaerobic co-digestion were used for NIRS calibration and validation. The NIRS predictions of the BMP values were satisfactory (Root Mean Square Error = 40 ml CH(4) g(-1) VS(fed); r(2) = 0.85). The integration of the entire substrate diversity in the model remained nevertheless difficult due to the specific organic matter properties of stabilised substrates and the high level of uncertainty of the BMP values. The elaboration of a model restricted to "fresh" substrates allows the practical use of the NIR technique to design and operate anaerobic co-digestion plants. The addition of more samples in the dataset in order to perform local calibrations would probably make the elaboration of a global NIR-model possible.