Optimizing Wastewater Treatment Predictions through Markov Chain Monte Carlo Bayesian Inference in Activated Sludge Model 3

The Activated Sludge Model No. 3 (ASM3) captures key biological and chemical processes in wastewater treatment but is challenging to calibrate due to its high-dimensional parameter space. This study applies a Bayesian inference framework using Markov Chain Monte Carlo (MCMC) to estimate ASM3 parameters from simulated plant data under varying monitoring strategies. Sensitivity and identifiability analyses are used to isolate influential, identifiable parameters, streamlining the inference process. Three case studies explore the effects of monitoring frequency and prior knowledge on parameter recovery. Results show that the Bayesian approach can reliably quantify parameter uncertainty and, with adequate data, support more targeted monitoring strategies using robust model-based decision making.