This study investigates the influence of chemical composition on the biochemical methane potential (BMP) of twelve different batches of fruit and vegetable waste (FVW) with different compositions collected over one year. BMP ranged from 288 to 516LNCH4kgVS-1, with significant statistical differences between means, which was explained by variations in the chemical composition over time. BMP was most strongly correlated to lipid content and high calorific values. Multiple linear regression was performed to develop statistical models to more rapidly predict methane potential. Models were analysed that considered chemical compounds and that considered only high calorific value as a single parameter. The best BMP prediction was obtained using the statistical model that included lipid, protein, cellulose, lignin, and high calorific value (HCV), with R2 of 92.5%; lignin was negatively correlated to methane production. Because HCV and lipids are strongly correlated, and because HCV can be determined more rapidly than overall chemical composition, HCV may be useful for predicting BMP.
The anaerobic digestion (AD) of a high diversity blend of fruit and vegetable waste (FVW) generated in tropical conditions as a single substrate was performed. A continuously stirred tank reactor (CSTR) operated in semi-continuous regime was used for AD. The reactor performance was monitored with gradually increasing organic loading rates (OLRs) from 0.5 up to 5.0 gVS L–1 d–1. The biochemical methane potential (BMP) of FVW determined by batch bottles was 360 LN CH4 kgVS–1, with a biodegradability of 79%. A stable pH with an adequate level of buffering capacity was observed during the entire experiment. Methane yield indicated the best performance at an OLR of 3.0 gVS L–1 d–1, with 285 LN CH4 kgVS–1 added, reaching 79% of BMP. At an OLR over 3.0 gVS L–1 d–1 accumulation of volatile fatty acids (VFA) was detected; in particular, propionic acid was monitored, and a decreased methane yield was detected. Biogas production rate was 1.55 LN L–1 d–1 and showed linear increase according to increases in the OLR.