Fuzzy approach to predict methane production in full-scale bioreactor landfills

Abstract

Bioreactor landfills (BRLs) aim to increase moisture content of municipal solid waste to enhance the biodegradationkinetics of the organic fraction and biogas production. Prediction of biogas production is a key tool to design anappropriate energy recovery system from BRLs. In this paper, a fuzzy-based model to predict methane generation infull scale BRLs is proposed. Eleven deterministic inputs (pH, RedOx potential, chemical oxygen demand, volatile fattyacids, ammonium content, age of the waste, temperature, moisture content, organic fraction concentration, particlesize and recirculation flow rate) were identified as antecedent variables. Two outputs, or consequents, were chosen:methane production rate and methane fraction in the biogas. Antecedents and consequents were transported in thefuzzy domain by a fuzzyfication procedure and then linked by 84 IF-THEN rules, which stated the effects of the inputparameters in a linguistic form. The fuzzy model was built and tested on seven lab-scale studies, representingdifferent operational conditions and waste qualities. The fuzzy model showed good performances in the prediction ofmethane generation, although lab-scale studies depicted ideal conditions that can be hardly reached in real BRLs. Inorder to deal with higher heterogeneities and lower data availability typical of full-scale landfills, new antecedentsand rules were added to the proposed model. With few adjustments based on the available information, the fuzzymodel could be applied to a retrofit BRLs located in Northern Italy. The results confirmed that fuzzy macro-approachcan be a powerful and flexible tool able to model the complex processes taking place in BRLs.