Abstract

A control strategy for optimising the performance of anaerobic co-digestion in terms of methane productivity, digestate quality and process stability is presented. A linear programming approach is adopted to calculate the feeding of multiple substrates for maximum methane productivity, subjected to restrictions based on experimental and heuristic knowledge. Process stability is quantitatively assessed by an empirical diagnosis function comparing alkalinity ratio measurements against reference values (outputs between (-1,1]). A second empirical diagnosis function is defined to compare methane flow rate measurements against a reference value of maximum capacity (outputs between (0,11). A variable-gain control function (outputs between (-1,11), derived from the diagnosis functions, is defined to determine the quantitative change applied to the most active constraint of the substrate blend optimisation problem leading to a new set-point of feeding substrates blend. The control strategy works in a closed-loop architecture under which the process performance for each blend of substrates is continuously assessed. The system was successfully validated in a 1 m(3) hybrid Upflow Anaerobic Sludge Blanket - Anaerobic Filter (UASB-AF) reactor, treating blends of substrates (gelatine, glycerine and pig manure supernatant) at OLR values between 0.71 and 6.33 gCOD/L d over a period of 210 days at mesophilic conditions. (C) 2015 Elsevier Ltd. All rights reserved.