A novel approach to evaluate state estimation approaches for anaerobic digester units under modeling uncertainties: Application to an industrial dairy unit

Monitoring the health of the AD during operation is an important task to ensure near optimal operation. Typical measurements from ADs such as pH, alkalinity and chemical oxygen demand (COD) however, do not contain adequate information regarding the internal state of the complex biochemical processes occurring in the AD. State estimation is therefore adopted to estimate internal variables from a nonlinear anaerobic digestion model and available measurements. However, they all rely on the availability of an adequate and valid model of the process. In this work, we develop a new index to assess the performance of state estimation techniques. The proposed metric, Hurst exponent, is computed through the method of detrended fluctuation analysis (DFA). Hurst exponent obtained from the residuals, difference between the measurements and filtered output from estimation techniques, is used for quantifying the performance three popular estimation techniques in the presence of modeling uncertainties. The proposed method is generic and can be used for analysis of state estimation approaches for anaerobic digesters in various industrial plants. The utility of the proposed method is demonstrated on a model developed for the largest dairy unit in India. Results reveal that different estimators exhibit varying sensitivity to modeling uncertainties and tradeoffs arise between efficiency and robustness.