- 4:18 PM

Plantwide Optimizing Control of Bioprocesses: Bio-Ethanol Process Case Study

Silvia Ochoa, Jens-Uwe Repke, and Günter Wozny. Chair of Process Dynamics and Operation, Berlin Institute of Technology, Straße des 17. Juni 135, Berlin, 10623, Germany

The main purpose of any industrial production process is to achieve all the time, despite the presence of disturbances, the highest profitability for the whole plant. For this reason, the purpose of process control should be to attain optimal operation instead of maintaining a predefined set point [1]. Furthermore, taking into account that in real operation, the different operating units are usually highly interconnected (i.e energy and mass recycle loops), the control of the process should be addressed from a plantwide control perspective because optimal operation of the individual units does not necessarily lead to optimal operation for the whole plant. Plantwide control is a large-scale problem which deals with up to hundreds of measurements and manipulated variables [2]. Approaches for plantwide control are usually classified into optimization-based [3-4] or heuristic-based [5-6]. Usually the optimization-based approaches apply nonlinear optimization methods to select the control structure, whereas the heuristic-based approaches use engineering knowledge coupled with simulation analysis to define a suitable control structure for the process.

In this paper the use of a plantwide optimization-based approach, designated as plantwide optimizing control, is proposed for the bio-ethanol process. In this approach, plantwide optimization and control are integrated into a single dynamic large-scale real-time optimization problem, where the objective function is defined by the plant profitability and the manipulated variables are used as decision variables. In order to solve the problem in real time, different strategies are required to increase the efficiency of the method. Therefore, the optimization routine is only re-called on-demand by the activation of an optimization trigger vector depending on the presence of disturbances and on the plant performance. The values of the trigger vector indicate whether each manipulated variable is considered as decision variable during the optimization or not, and determine the relative size of the search region for each decision variable.

The plantwide optimizing approach has been applied to the continuous bio-ethanol production process from starch, taking into account saccharification, fermentation, cells recycle and distillation stages. Although the bio-ethanol process has ten available manipulated variables, only seven of them are taken as decision variables during the optimization problem, because the remaining are used in an internal control loop proposed for biomass control [7]. This internal loop comprises a split-range controller for keeping the biomass concentration in the fermentor at its optimal value and a ratio controller for achieving a suitable viscosity in the biomass recycle slurry. The biomass internal loop is connected in cascade with the optimization layer involved in the plantwide optimizing control, receiving the optimal set points that should be tracked in order to maximize the profitability objective function defined for the whole plant.

The motivation for selecting the bio-ethanol process as case study relays on the fact that, in spite of recent advances in purification technologies and in the modification of microbial strains for making them more resistant to the different stress and inhibition factors, the economical feasibility of the bio-ethanol industry is still questioned. For this reason, the optimization and control of this process remain as challenging problems [8]. The results obtained using the plantwide optimizing approach for the continuous bio-ethanol process will be presented and compared to a two-layer nonlinear model predictive controller (NMPC) with set point optimization, showing that plantwide optimizing control is a very promising alternative for achieving optimal operation in bioprocesses even in the presence of disturbances.


[1] S. Engell. Feedback control for optimal process operation. Journal of Process Control, 17, 203-219, 2007.

[2] A. C. B. de Araújo. Studies on Plantwide Control. Ph. D. Thesis. Norwegian University of Science and Technology, Trondheim, 2007.

[3] S. Skogestad. Plantwide control: the search for the self-optimizing control structure. Journal of Process Control, 10: 487-507, 2000.

[4] G. Stephanopoulos and C. Ng. Perspectives on the synthesis of plant-wide control structures. Journal of Process Control 10: 97–111, 2000.

[5] W. Luyben, L. B. D. Tyréus and M. L. Luyben. Plantwide process control. McGraw-Hill. USA, 1998.

[6] H.R. Lausch, G.Wozny, M.Wutkewicz and H. Wendeler. Plant-wide control of an industrial process. Trans IChemE, 76: 185-192, 1998.

[7] S. Ochoa, J-U. Repke and G. Wozny. Integrating Real-Time Optimization and Control for Optimal Operation. Proceedings of the CHEMPOR 2008 conference, Braga, Portugal, 2008, 151.

[8] S. Ochoa, A. Yoo, J-U. Repke, G. Wozny and D.R. Yang., Modeling and Parameter Identification of the Simultaneous Saccharification-Fermentation Process for Ethanol Production. Biotechnol. Prog., 23: 1454-1462, 2007