Model Predictive Control for Demand- Driven Biogas Production in Full Scale Biogas plants have the potential to provide demand-oriented electricity to com- pensate the occurring divergence between energy demand and supply by uncon- trollable sources like wind and solar power. The general flexibility of the biological process is proofed in particular under full-scale conditions for a biogas production according to the grid demand. A model predictive control was developed to calcu- late feeding strategies in order to fulfill a demand-oriented gas utilization timeta- ble. Full-scale experiments showed a high intraday flexibility in a wide range of the average gas production and high process stability in reaction to pulse feeding. The gas storage demand could be reduced significantly compared to the common constant feeding operation. Keywords: ADM1, Biogas, Feeding management, Model predictive control Received: July 10, 2015; revised: January 15, 2016; accepted: January 20, 2016 DOI: 10.1002/ceat.201500412 1 Introduction The financial incentives set by the Renewable Energy Sources Act (EEG) led to an increasing number of biogas plants in Germany. Until 2012, the plants have been designed and con- structed to produce a stable and constant electric energy out- put, i.e., base load energy. The changing conditions within the energy sector in Germany increase the need of controllable electric energy supply and force biogas plants to be able to compensate the divergence between electric energy demand and electric energy supply by other uncontrolled sources like wind and solar power [1]. Szarka et al. [2], Hochloff and Braun [3], and Hahn et al. [4] stated that bioenergy concepts seem to be a promising option to fulfill most of the requirements in the transition of the energy system from fossil to renewable sour- ces. The technical development of biogas plants nowadays fo- cuses on expansion on gas storage capacity and different multi- phase digester concepts and implies high investments [4–7]. An alternative approach especially at existing biogas plants could be the adjustment of the feeding management [8]. Experiments in lab-scale continuous stirred-tank reactors (CSTRs) to analyze different flexible feeding strategies with sugar beet silage, maize silage, and cattle manure showed a long-time stable process, even under highly dynamic operating conditions [8]. Grim et al. [9] investigated the economic effects of different flexibility scenarios for demand-oriented combined heat and power unit (CHP) operation under Swedish condi- tions and showed that feeding management reduced the stor- age requirement. For full-scale application of flexible feeding strategies, a process control system is of need to avoid process failures and thereby economic losses. Several authors discussed control scheme design and control methods in relation to bio- gas process control. Methods range from classical control methods such as feedback and feedforward control to advanced model-based and multivariable control systems [10–16]. Thereby, the respective objectives are quite different and reach from stabilizing gas production to the prevention of disturban- ces by ammonia and acidification. Especially model predictive control (MPC) is designated for a targeting operation to future set points. MPCs use an internal dynamic model of the process, a history of past control moves, and an optimization objective function over the receding pre- diction horizon to calculate the optimal control moves [17]. In literature, different complexities, tasks, and applications as well as used model types for MPC can be found. Bernard et al. [18] www.cet-journal.com ª 2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eng. Technol. 2016, 39, No. 4, 652–664 Eric Mauky 1,2 So ¨ ren Weinrich 1,2 Hans-Joachim Na ¨ gele 3 H. Fabian Jacobi 4 Jan Liebetrau 1 Michael Nelles 1,2 1 Department of Biochemical Conversion, DBFZ – Deutsches Biomasseforschungszentrum, Leipzig, Germany. 2 Faculty of Agricultural and Environmental Sciences, Chair of Waste Management, University of Rostock, Rostock, Germany. 3 State Institute of Agricultural Engineering and Bioenergy, University of Hohenheim, Stuttgart, Germany. 4 Hessian State Laboratory, Fachgebiet IV.5.4 Renewable Energies, Gießen, Germany. Supporting Information available online – Correspondence: Eric Mauky (eric.mauky@dbfz.de), Department of Biochemical Conversion, DBFZ – Deutsches Biomasseforschungszen- trum, Torgauer Straße 116, 04347 Leipzig, Germany. 652 Research Article