Indirect Monitoring of Anaerobic Digestion for Cheese Whey Treatment

Abstract

Efficient monitoring is an open problem in the operation of anaerobic digestion processes, due to the lack of accurate, low-cost, and proper sensors for the on-line monitoring of key process variables. This paper presents two approaches for the indirect monitoring of the anaerobic digestion of cheese whey wastewater. First, the observability property is addressed using conventional and nonconventional techniques, including an observability index. Then, two model-based observer techniques, an extended Luenberger observer, a sliding mode observer, and a data-driven technique based on fractal analysis are formulated and discussed. The performance and capabilities of the proposed methodologies are illustrated on a validated model with experimental data of the anaerobic digestion of cheese whey. Experimental pH measurements are used for the data-driven approach based on fractal analysis. The experimental data sets correspond to experimental conditions (pH > 7.5 and temperature (T) = 40 °C) favoring volatile fatty acid (VFA) production (30 g/L) with simultaneously acceptable biogas production (3420 mL). Results also show that the proposed observers were able to predict satisfactory key process variables. On the other hand, the fractal analysis provides reliable qualitative trends of VFA production and chemical oxygen demand (COD) consumption.

Diagnosis and Monitoring of Volatile Fatty Acids Production from Raw Cheese Whey by Multiscale Time-Series Analysis

“…Model-based estimation methods are based on the reconstruction of unmeasured variables from the measurements available through the mathematical model of the process [11,17]. Model-based estimation approaches applied to AD processes include state-observers [18][19][20][21][22], Bayesian methods [23], and filtering techniques [24,25]. The most significant limitation of model-based approaches is that they depend on the accuracy of the process model, i.e., the lower the precision of the model, the greater the estimation error.…”