Modelling and control of anaerobic wastewater treatment processes

Abstract

The application of anaerobic digestion process for the wastewater treatment of industrial wastewaters has been growing continuously. Nevertheless, monitoring, diagnosis and control of the process are not consolidated topics, and they subjects require of more oriented research to structure simple tools for the industrial application. This thesis contains concerned with ADM1 modelling and the development of control systems for anaerobic wastewater treatment plants. The developed controllers were based on fuzzy logic, heuristic functions, and applying different control strategies as cascade configuration in anaerobic wastewater treatment process. The thesis it is organized in 6 chapters that can be bracketed into three groups covering: Introduction (Chapter 1), Materials and Methods (Chapter 2); Modelling (Chapter 3); and Supervision/Control (Chapters 4, 5 and 6): Chapter 1 provides a general overview of anaerobic digestion, including process aspects, mathematical modelling and control in anaerobic wastewater treatment plants, as a background knowledge for the interpretation of the results into the next chapters. Different aspects of anaerobic digestion (AD) are introduced, including the main reactions and treatment technologies. A general description about the Anaerobic Digestion Model Nº 1 (ADM1) is given. Anaerobic control aspects as control objectives, disturbances, manipulated variables, controlled variables and control structures are analysed and discussed. Moreover, a review of the state of the art of control in anaerobic wastewater treatment process is presented. In Chapter 2, the materials and methods used to carry out this thesis are shown. The reactor used (a hybrid anaerobic Upflow Bed Filter USBF) is described jointly with the in-line and on-line measurement devices and actuators which were applied for monitoring and control purposes. Different analytical techniques are explained and grouped according to the associated phase which is monitored. Furthermore, characteristics and composition of the three types of wastewaters are presented at the end of this chapter. A methodology based on principal component analysis (PCA) for determining the minimum number of reactions that have to be taken into account for modelling anaerobic wastewater treatment (AWT) units for agro-food industries is presented in Chapter 3. PCA technique was extended and applied to simplify the ADM1 model, which is very complex and it needs a large number of parameters and states. This hinders its calibration and may render it inadequate for development of state observers and/or controllers. Therefore, a reduced model was derived using the appropriate number of reactions. In this sense, a smaller stoichiometric matrix was identified and new kinetic functions were proposed on the base of well known biological kinetics (Haldane or Monod). Then, kinetic parameters were estimated and the model was validated. The reduced developed maintains the appropriate qualitative behaviour of the ADM1 In Chapter 4, a variable gain controller (VGC) previously developed was applied to automatic restart-up of AWT plants. The validation of the system was carried out by simulation with ADM1 and by closed-loop control of a pilot scale USBF reactor. It was found that the time required for restart-up of AWT plant after short programmed stop periods (one month) can be highly reduced in comparison to manual procedure by using this control system (only two days are needed). Additionally, in order to test the controller for handling of organic feedings, a COD increase of a 50% was applied. A major advantage of this controller is the low cost of the measurement devices required and its suitability for keeping a high effluent quality through indirect control of effluent COD concentration by measuring H2 concentration in biogas. A fuzzy logic based system for diagnosis and control of an AWT plant was developed in Chapter 5. The diagnosis and control system was developed using expert's knowledge related with normal or overload process states for establishing a set of rules considering the selected variables. The diagnosis and control system was validated at three different levels: simulation with ADM1; open-loop methodology with experimental data from an USBF reactor; and finally in closed-loop control treating wastewater containing carbohydrates. A new multi-objective control strategy, based on VFA and methane flow rate as measured variables and the feed flow rate as control variable in order to adjust the organic load applied, is described in Chapter 6. The approach considers two operational objectives, the control on the effluent quality and the control on the maximum production rate of methane; and takes advantage of the difference between the dynamics of the liquid and gas phase. The control strategy consists of a cascade controller with a reference signal for methane flowrate in the internal loop. The performance of the controller is illustrated via numerical simulations with the ADM1 respect of influent perturbations. Moreover, the controller has been validated in closed-loop control of a USBF reactor treating wastewater containing ethanol under different operational scenarios. The controller supplies an adequate control action, and it demonstrated a high reliability, to achieve the desired set-point in each episode for the studied disturbances.