IFAC PapersOnLine 50-1 (2017) 12673–12678 ScienceDirect ScienceDirect Available online at www.sciencedirect.com 2405-8963 © 2017, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. Peer review under responsibility of International Federation of Automatic Control. 10.1016/j.ifacol.2017.08.2245 © 2017, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. Keywords: optimization, anaerobic digester, extremum seeking control, model-based algorithm, performance criterion. 1. INTRODUCTION Anaerobic digestion (AD) is one of the most widely spread bioprocesses of great importance in the wastewater processing technology, having - in addition - an intake in energy producing. In the last decades, the issue of the control of anaerobic digestion plants (ADP) has been a constant challenge, being stimulated by the evolution of specific instrumentation and control methods. Since 2006 (Steyer et al., 2006), a large range of control techniques dedicated to ADP, starting with model-free control (PI/PID) and continuing with non-linear model-based control, non- parametric adaptive control, robust control, neural networks and fuzzy logic-based control algorithms have been mentioned. Recently, in the extensive synthesis paper (Jimenes et al., 2015) new evolutions in AD modelling, instrumentation, software sensors and new control solutions such as expert systems, model-based and adaptive linearizing control, sliding mode control, machine learning methods for software sensor in ADP (Gaida et al., 2012) are being analysed. The paper (Petre et al., 2013), oriented towards adaptive and robust-adaptive control strategies for AD processes, as well as many other similar strategies, illustrates that the works mentioned in the two synthesis papers do not exhaust the variety of approaches proposed for ADP control. An important objective of AD control is the optimization in relation to a criterion which takes into consideration the gaseous methane production. This issue is treated in various approaches based on ADP dynamic model (Stamatelatou et al., 1997), (Gaida, 2014), (Haugen et al., 2014) and (Sbarciog et al., 2011). A distinct approach of AD optimization is extremum seeking control (ESC) which uses minimum information regarding the bioprocess dynamics. The exploration of the possibilities of this approach, in bioprocess optimization, generally and in ADP, particularly, is widely reported in literature (Wang et al., 1999), (Bastin et al., 2009), (Caraman et al., 2015), (Dochain et al., 2011), (Guay et al., 2004), (Liu et al., 2006), (Marcos et al., 2004), (Lara- Cisneros et al., 2015), (Vargas et al., 2015) and (Ifrim et al., 2016). The present paper deals with a comparative study of two distinct static digester optimization approaches based on a performance criterion which expresses the maximization of the methane production, penalizing the environment pollution with discharged water. The first approach uses ESC, two control versions, stochastic and deterministic, being analysed comparatively. The second approach starts from the characteristics of the optimal regimes obtained on the basis of a mathematical model. It gives the optimal setpoint of the dilution rate as a function of the substrate concentration from the inflow. This variable is non-measurable on-line, but is estimated using an Extended Kalman Filter (EKF). The paper is structured as follows: the next section presents AD model and the optimization criterion. Sections 3 and 4 present the performances obtained using the extremum seeking algorithm (ESA) and the model-based algorithm, respectively. Section 5 is dedicated to the comparison of the two approaches based on five evaluation criteria. A conclusion section ends the paper. 2. ANAEROBIC DIGESTER MODELING AND OPTIMIZATION CRITERION Two types of models were developed in modelling of anaerobic digestion processes: complex models, such as ADM1 (Batstone et al., 2002), which makes a detailed description of the phenomena that take place in the case of these processes, or simpler, of reduced order, such as the one proposed in (Bernard et al., 2001), which is useful in control Abstract: The paper deals with the optimization of an anaerobic digester with respect to a performance criterion which targets the maximization of the methane production. It is penalized by the environment pollution with the discharged water. The paper includes a comparative study regarding the performances of two optimization approaches: the extremum seeking control, in stochastic and deterministic versions, and the model-based control respectively. The latter uses a characteristic of the optimal regimes which supplies the optimal setpoint (the dilution rate) as a function of the substrate concentration in the inflow, estimated by an extended Kalman filter. *”Dunărea de Jos” University of Galati, Domnească Street, no. 47, 800008, Galati, ROMANIA (Tel: +40 236 470 905; e-mail: segiu.caraman@ ugal.ro). ** LUNAM, Université de Nantes, ONIRIS, GEPEA, UMR-CNRS 6144, University Bd. CRTT-BP 406, 44600 Saint-Nazaire, France (e-mail: mariana.titica@univ-nantes.fr) S. Caraman * , M. Barbu * , G. Ifrim * , M. Titica ** , E. Ceanga * Anaerobic Digester Optimization Using Extremum Seeking and Model-Based Algorithms. A Comparative Study