Search

ispartof: JOURNAL OF PROCESS CONTROL vol:117 pages:169-180 status: published

Model predictive control strategies refer to a set of methods relying on a process model to determine an optimal control signal by minimising a cost function. This paper reports on the application of predictive control strategies to a wheeled mobile robot. As a first step, friction forces originating from the motor gearboxes and wheels were estimated and a feedforward compensation was applied. Step response tests were then carried out to identify a linear model to design several simple control strategies, such as the Proportional-Integral-Derivative (PID) controller. The PID response constitutes the reference to assess the efficiency of two predictive control strategies: the generalised predictive control (GPC) and the linear quadratic model predictive control (LQMPC) algorithms. These control strategies were tested in simulation with Matlab and EasyDyn (a C++ library for multibody system simulations) and in real life experiments. All three control strategies offer satisfactory reference tracking but MPC allows a reduction of the energy consumption of up to 70 % as a result of set-point anticipation. LQMPC is the best in terms of input activity reduction.

Following the recent success of hot-melt extrusion in the pharmaceutical field, monitoring and control are increasingly applied. In this study, a mathematical model consisting of mass and energy balance partial differential equations is developed and parameters related to material transportation are inferred from experimental data collected from a pilot plant. This data is relative to the output residence time distribution of active product. Parameter identification is complemented by a parametric sensitivity analysis and the computation of confidence intervals. Direct and cross-validation results demonstrate the good predictive capability of the model, which could probably be exploited for model-based control at a next stage.

The advanced control of microalgal cultures usually requires the knowledge of several component concentrations, which are however not always measurable on-line. In this context, state estimation plays an important role, and software sensors should be robust to model uncertainties and measurement noise. In this study, two software sensors are designed in the form of extended Luenberger observers, using Lyapunov arguments and linear matrix inequalities (LMI). These observers are based on Droop model and a few available on-line sensors. The first observer design estimates the intracellular quota and substrate concentrations considering a linear differential inclusion modeling technique and a constant observer gain. On the other hand, the second one estimates only the intracellular quota concentration assuming uncertainties in the model parameters and a quasi-Linear Parameter Varying (quasi-LPV) representation of the nonlinear system. The results are presented considering simulated and experimental data from Dunaliella tertiolecta culture.

In biotechnological processes such as fed-batch reactors the lack of reliable and robust on-line sensors and the limited number of actuators make the task of operating at optimal conditions very difficult. We present a feedback controller that aims at regulating the substrate concentration at an optimum value such that biomass production is enhanced while by-product formation is not favored. We use a virtual output that is estimated using a bank of weighted super-twisting observers to drive an output-feedback extremum-seeking controller. The only online measurements needed are the biomass concentration and the oxygen and carbon dioxide mass transfer rates. Simulations on a fed-batch bioreactor model show its applicability.

The objective of this paper is to address the problem of state estimation in an Anaerobic Digestion Reactor (ADR) with unknown inputs - typically some influent concentrations, and to compare two different state and input estimation schemes. The first one is the classical Extended Kalman Filter (EKF) based on an augmented system that considers a slowly varying input (approach that has been already applied to this system and reported in the literature), whereas the second one is a recently proposed Unknown Input Observer (UIO), formulated in the spirit of a Kalman Filter, for continuous estimation with discrete measurements. The two filters are evaluated in simulation, demonstrating the superiority of the UIO.

This study investigates a real-time optimization strategy for micro-algae continuous processes. The objective is to determine the optimal dilution rate maximizing the biomass productivity on the basis of the on-line measurement of biomass concentration and minimum prior process knowledge. Two different micro-algae strains (with different growth rates) are considered in this study: Dunaliela tertiolecta and Isochyris galbana. The extremum seeking control shows good performances in both cases. Practical tuning guidelines can be derived based on the strain growth dynamics.

Hot-melt extrusion is a popular process in the field of industrial product manufacturing. In this work, a simple distributed parameter model consisting in partial differential equations derived from mass balances, well-adapted to the design of a control strategy, is expressed. Parameter identification is applied to a specific device with a complex screw geometry and provides a good representation of the extruder behavior. Due to the existence of strong nonlinearities and complex inputoutput coupling, a nonlinear model predictive control (NMPC) strategy is proposed and designed to regulate the output active pharmaceutical ingredient (API) concentration and the output flow. The effectiveness of this strategy is demonstrated by numerical validations and robustness analysis with respect to measurement noise and model uncertainties.

This paper describes the implementation and validation of a model predictive controller (MPC) for a photobioreactor system. The microalgae concentration, the solution pH and the illumination index, which is related to the lit PBR volume, are simultaneously controlled at imposed setpoint values by manipulating the dilution rate, the flow of CO 2 supplied to the culture and the incident light. The complex, time consuming nonlinear optimization is replaced by iterative quadratic programming optimizations. With a reduced number of tuning parameters and no demand for dynamics decoupling, the proposed controller improves considerably the speed of the control loop compared to the results available in literature.

in press; International audience; Submerged membrane bioreactors (sMBR's) are a promising technology for nitrogen removal in recirculating aquaculture systems (RAS's). However, there are still relatively few reports on the experimental application of this strategy. In this study, a laboratory-scale system, mimicking a RAS fitted with a sMBR, \rouge{was} designed and automated, and a simple dynamic sMBR model including biological and physical phenomena \rouge{was} validated. The system \blue{was} analyzed based on measurements collected by a data logging structure involving a programmable logic controller (PLC), an \rouge{industrial network protocol} and a LabView application software. This study confirms the \rouge{suitability} of sMBR systems \blue{within} aquaculture applications. The dynamic model has good predictive capabilities and could be used for the design of advanced control structures, such as model predictive control.

This paper presents a framework for the multivariable robust control of perfusion animal cell cultures. It consists of a cascade control structure and an estimation algorithm, which provides the unmeasurable variables needed in the design of the control law, and ensures the regulation of the cell and glucose concentrations at imposed levels by manipulating the bleed and the dilution rates. The cascade control structure uses a feedback linearizing controller in the inner loop and linear (PI) controllers in the outer loops, and requires the measurement of the cell concentration and the glucose concentration in the bioreactor. Two approaches are provided: the first one assumes the availability of an approximate model of the process kinetics and uses an extended Kalman filter (EKF) to estimate the system states; the second approach does not require the prior knowledge of the process kinetics. These are estimated from the available measurements using sliding mode observers (SMO). A receding horizon optimization algorithm is employed to (periodically) tune the gains of the outer loop controllers. The proposed framework is easy to implement and tune, and may be applied to a general class of perfusion cell culture systems. Its effectiveness and robustness are illustrated by means of simulation results.

The simulated moving bed (SMB) technology is increasingly applied in various fields, ranging from the food to the pharmaceutical sectors, for the chromatographic separation of fine (bio)chemicals. In this study, an adaptive controller acting on the fluid flow rates and commutation period is used to regulate the spatial location of the adsorption and desorption waves, and in turn the purity and productivity of the raffinate and extract effluents. This controller is based on a simple discrete-time model of the concentration fronts movement, derived from wave theory. A simple parameter adaptation scheme makes this controller robust to parameter uncertainties and drifts, and allows process start-up with minimum a priori knowledge of the separation parameters. In this study, the performance of the controller is demonstrated for two different applications: the separation of fructo-oligosaccharides (linear isotherms) and cyclopentanone–cycloheptanone (competitive Langmuir isotherms). Different plant/sensor configurations are also examined, indicating the potential of the control strategy even with reduced measurement information.

This study proposes a multiple-input/multiple-output robust approach to the control of bioprocesses based on a cascaded-loop strategy. The internal loop is a classical input-to-output feedback linearising controller which is obtained from the nominal dynamics of the bioprocess. Then, the outer loop is designed based on the internal model principle to obtain zero steady-state tracking error (and disturbance rejection) for constant signals while ensuring the robust stability of the overall closed-loop system. In addition, a robust Luenberger-like observer is proposed to estimate unmeasured state variables for the feedback linearising control law. The approach is applied to the simultaneous control of biomass and substrate concentrations in a perfusion/chemostat bioreactor, where the simulation results demonstrate the effectiveness of the proposed control strategy.

Optimizing control of Escherichia coli cultures depends on the availability of appropriate on-line sensors for the main culture components. An efficient and practical way to maintain E. coli cultures in the neighborhood of the optimal operating conditions is to regulate the acetate concentration at a constant low level. Unfortunately, reliable acetate probes are currently not available on the market. A way to overcome this lack of information is to design software sensors. In this work, observability conditions are first examined, and Extended Kalman Filters (EKF) are developed for various hardware sensor configurations, taking account of their reliability and cost, and including a possible kinetic parameter estimation. The filters are validated using experimental data collected on a lab-scale bioreactor.

This work addresses the application of control systems to the optimization of a monoclonal antibodies (MAb) production chain. The attention is focused on the maximization of hybridoma fed-batch culture productivity. The proposed model presents kinetics showing strong nonlinearities through min-max functions expressing overflow metabolism. A nonlinear model predictive control (NMPC) algorithm, choosing the best trajectory over a moving finite horizon among different sequences of inputs, is suggested in order to optimize productivity. Sensitivities of selected objective functions are considered in a minimax robust version of the NMPC in order to choose the best configuration with respect to practical operating conditions.

The cultivation of animal cells in perfusion allows the production of various biopharmaceutical products. In this work, the observability properties of a nonlinear dynamic model of these animal cell cultures is assessed using a method based on a natural dynamical interpretation of the observability/detectability concepts, leading to the description of the indistinguishable dynamics of the system. Following this analysis, a Kalman filter is designed to reconstruct on-line variables which are difficult or expensive to measure directly with a hardware sensor.

In this study, a multivariable control structure is developed to simultaneously control the concentrations of cells and of one of the nutrients in an animal cell cultivation system operated in perfusion. A cascade control structure is considered consisting of (i) an inner loop with a partially linearizing feedback controller, tuned so as to ensure robustness with respect to parameter uncertainties and non-canceled nonlinearities; and (ii) an outer loop involving two linear predictive controllers. The resulting control strategy shows robustness and performance properties similar to more computationally demanding strategies (such as a multivariable nonlinear MPC strategy), while requiring less measurements and involving an easier implementation.

This paper presents the bifurcation analysis of an anaerobic digestion model, which includes hydrogen production along with methane production. Analytical expressions of the equilibria are given and the conditions for their occurrence are provided. The analytical developments are validated by means of bifurcation diagrams built with specialized continuation software. Special emphasis is put on revealing the relationship between hydrogen and methane outflow rate. The complex system behavior is illustrated by means of phase portraits.

In the field of industrial product manufacturing processes, hot-melt extrusion becomes more and more popular. In this context, a distributed parameter model consisting in partial differential equations derived from mass and energy balances is applied to a specific device with a complex screw geometry. Several numerical simulations bring out strong nonlinearities and input-output coupling. Following these observations, a nonlinear model predictive control (NMPC) strategy is proposed and designed to regulate the output API concentration and the output flow. Numerical validations and robustness analysis with respect to measurement and model noises are achieved to show the effectiveness of the strategy.

International audience; In this study, a simple dynamic model of a submerged Membrane BioReactor (sMBR) is proposed, which would be suitable for process control. The system dynamics is first analyzed showing the existence of three different time scales. The existence of slow-fast dynamics is central to the development of a dedicated parameter estimation procedure. The proposed model structure is validated using realistic simulation data from a detailed simulator built in a well-established environment, namely GPS-X. Finally, a nonlinear model predictive control is designed to illustrate the potential of the developed model within a model-based control structure. The problem of water treatment in a recirculating aquaculture system is considered as an application example.

Several mathematical models have been developed in anaerobic digestion systems and a variety of methods have been used for parameter estimation and model validation. However, structural and parametric identifiability questions are relatively seldom addressed in the reported AD modeling studies. This paper presents a 3-step procedure for the reliable estimation of a set of kinetic and stoichiometric parameters in a simplified model of the anaerobic digestion process. This procedure includes the application of global sensitivity analysis, which allows to evaluate the interaction among the identified parameters, multi-start strategy that gives a picture of the possible local minima and the selection of optimization criteria or cost functions. This procedure is applied to the experimental data collected from a lab-scale sequencing batch reactor. Two kinetic parameters and two stoichiometric coefficients are estimated and their accuracy was also determined. The classical least-squares cost function appears to be the best choice in this case study.

The simultaneous estimation of unmeasured state variables and unknown inputs is particularly relevant in environmental applications, such as biological wastewater treatment or the culture of microalgae in open environments. In this study, an extension of the unknown input observer (UIO) recently proposed in (7) to nonlinear continuous-time models and discrete-time measurements is presented. This UIO takes account of the process and measurement noises, and uses a linearization around the estimated trajectory, in a way similar to the extended Kalman filter. This extended continuous-discrete UIO is applied to state and input reconstruction in continuous cultures of phytoplankton. In particular, the problem of simultaneously reconstructing the extra- and intra-cellular nitrogen concentrations together with the light intensity or the dilution rate from measurements of biomass in cultures of the marine microalgae Dunaliella tertiolecta or the cryptophyceae Rhodomonas salina are considered. Simulation results and processing of real data demonstrate the performance of the method.

This paper presents a simple switching control policy for driving an anaerobic digestion system from an arbitrary initial state to an optimal setpoint, characterized by the maximum biogas production. A high wastewater treatment efficiency is achieved and a fast convergence to the optimal setpoint is obtained. The main features of this control strategy are its simplicity and ease of implementation, as it is based on the measurement of the biogas outflow rate, which is commonly available in anaerobic digestion systems.

A simple procedure is presented to estinlate a) the number of macroscopic reactions needed to represent the time-evolution of the key components of a bioprocess, b) its stoichiometric matrix, and c) the structure and parameter value of its kinetics. The first two stages use the mathematical technique of Maximum Likelihood Principal Component Analysis to determine the stoichiometric matrix of the process, and the third stage uses Likelihood Ratio Testing to navigate through the branches of a decision tree nlade of various kinetic structures. The perfornlance of our identification procedure is illustrated all along the paper, with a simulated example of anaerobic digestion.

Biological and environmental systems are often influenced by unknown inputs or disturbances, which makes monitoring or state estimation more delicate. In this study, the simultaneous estimation of unmeasured state variables and partly unknown inputs is considered. Only qualitative prior information on these inputs is used in the design procedure, leading to the concept of quasi-unknown input observers (QUIO). These software sensors are applied to the estimation of concentrations, flow rates and light intensity in phytoplanktonic cultures in the chemostat. Implementation and numerical tests are discussed, based on simulation and experimental data.

Optimizing control of E. coli cultures depends on the availability of appropriate on-line sensors for the main culture components. A simple and efficient approach to maintain E. coli cultures in the neighborhood of the optimal operating conditions is to regulate the byproduct (i.e., acetate) concentration at a constant low level. Unfortunately, reliable acetate probes are currently not available on the market, and it is necessary to design a software sensor. In this work, observability conditions are first examined, and Extended Kalman Filters (EKF) are developed for various hardware sensor configurations, taking account of their reliability and cost. The filters are validated using experimental data collected on a lab-scale bioreactor.

This paper presents a strategy for the optimization of biogas outflow rate in an anaerobic digestion process described by a two-population model. The methodology relies on the solution of two optimization problems: steady state optimization for determining the optimal operating point and transient optimization. The latter is solved using the maximum principle of Pontryagin. The proposed control law, which drives the process from an initial state to the optimal steady state while maximizing the biogas outflow rate, consists of switching the manipulated variable (dilution rate) from the minimum to the maximum value and then to the optimal value at well defined instants. This control law substantially increases the stability region of the optimal equilibrium point, enlarging it in some cases to almost the entire state space. Aside its efficiency, the strategy is also characterized by simplicity, being thus appropriate for implementation in real-life systems. Another important advantage is its generality: this technique may be applied to any anaerobic digestion process, for which the acidogenesis and methanogenesis are respectively characterized by Monod and Haldane kinetics.

In this paper, experimental investigations of an adaptive RST control scheme for the regulation of the ethanol concentration in fed-batch cultures of S. cerevisiae is presented. Our main objective is to prove efficiency and robustness of this controller in experimental applications ranging from laboratory to industrial scales. The controller only requires one on-line measurement signal, making it easily implementable in an industrial environment. Disturbance rejection is ensured thanks to an on-line parameter adaptation procedure, which delivers as a side product an estimate of the growth rate that can be used for process monitoring purposes. The robustification of the controller is achieved in a simple way, using the observer polynomial.

This work addresses the control of a lab-scale fed-batch culture of Escherichia coli with a nonlinear model predictive controller (NMPC) to determine the optimal feed flow rate of substrate. The objective function is formulated in terms of the kinetics of the main metabolic pathways, and aims at maximizing glucose oxidation, while minimizing glucose fermentation. As bioprocess models are usually uncertain, a robust formulation of the NMPC scheme is proposed using monotonicity arguments and a min-max optimization problem. The potentials of this approach are demonstrated in simulation using a Monte-Carlo analysis.

The purpose of this paper is to report on the development of a procedure for inferring black-box, yet biologically interpretable, dynamic models of bioprocesses based on sets of measurements of a few external components (biomass, substrates, and products of interest). The procedure has three main steps: (a) the determination of the number of macroscopic biological reactions linking the measured components; (b) the estimation of a first reaction scheme, which has interesting mathematical properties, but might lack a biological interpretation; and (c) the "projection" (or transformation) of this reaction scheme onto a biologically-consistent scheme. The advantage of the method is that it allows the fast prototyping of models for the culture of microorganisms that are not well documented. The good performance of the third step of the method is demonstrated by application to an example of microalgal culture.

In this work a late lumping approach is used in order to design a state feedback linearizing controller for influent disturbance attenuation and regulating the nitrogen concentration at the output of a denitrification biofilter. This controller is associated to a distributed parameter observer to estimate all the states needed to compute the controlled input. It results in an output feedback nonlinear controller with stable closed-loop dynamics.

Mathematical modeling of sedimentation has attracted considerable attention in the past decades, and nowadays, one of the most popular models of secondary settler in activated sludge processes is the one proposed by Takacs et al. [I. Takacs, G.G. Patry, D. Nolasco, A dynamic model of the clarification-thickening process, Water Res. 25 (10) (1991) 1263–1271]. This model is based on a discretization in finite volumes (or layers) of the spatial domain, and in a rather inconsistent way, the number of layers is usually considered as a model parameter chosen so as to fit experimental data. In this study, a simple convection–diffusion partial differential equation (PDE) model is first formulated and solved using a Method of Lines strategy allowing the use of various spatial discretization methods with largely improved accuracy and efficiency. Model parameters are estimated using experimental data collected in batch settling experiments by De Clercq [J. De Clercq, Batch and continuous settling of activated sludge: in-depth monitoring and 1D compression modeling, Ph.D. Thesis, Universiteit Gent, Faculteit Ingenieurswetenschappen, Belgium, 2006], showing the good model predictive capability. Finally, the PDE settler model is coupled with a standard ASM1 representation of the activated sludge process, and implemented in a MATLAB dynamic simulator.

The systematic derivation of macroscopic biological reaction schemes has received an increasing attention in the last decade. This paper builds upon recent results, to suggest a new methodology for reaction network derivation from noisy measurements of external substrates and cell products. Moreover, based on the concept of model equivalence, it develops a two-step algorithm that aims at improving the robustness of the whole identification procedure.

Abstract This work proposes a robust control strategy for the optimizing control of fed-batch cultures of S. cerevisae. The process dynamics is characterized by a nonlinear kinetic model based on the bottleneck assumption and ethanol inhibition for a possible excess of substrate feeding. The control strategy is based on the feedback linearization technique, where the resulting free linear dynamics is designed so as to ensure a certain robustness to plant parameter variations. A feedforward loop achieves the correct critical substrate value, which is a function of the ethanol and oxygen in the culture medium. In addition, a robust Luenberger-like observer is designed taking plant parameter variations into account. Numerical experiments demonstrate the potential of the proposed approach as a tool for control design of fed-batch cultures.

Model-based control of bioprocesses is a difficult task due to the challenges associated with biological system modeling and the lack of on-line measurements. In this study, two robust controllers using minimal a priori process knowledge and minimal measurement information are designed to maximize biomass productivity in aerobic cultures of Saccharomyces cerevisiae . This latter objective can be achieved through the regulation of the ethanol concentration at a low constant value. The linearization of Sonnleitner’s model allows simple transfer function models to be derived, which describe the relation between the ethanol concentration, the substrate feed and an exponential disturbance – image of the substrate demand for cell growth – in the different operating (respirative and respiro-fermentative) regimes. The two controllers are based on these linear models and use a RST structure, but differ in the way the exponential growth disturbance is handled. In the first controller, the disturbance is represented by a linear model, whereas in the second controller, the disturbance is measured on-line via the oxygen transfer rate signal and a feedforward control action is used to cancel the disturbance effect on the ethanol concentration. Particular attention is paid to the robustification of the controllers to measurement noise, neglected high frequency dynamics and uncertain stoichiometry coefficients using the observer polynomial. Tests in simulation show the controller performance.

This paper proposes a nonlinear model predictive control scheme with particular application to the regulation of acetate concentration in order to maximize the biomass growth during fed-batch cultures of E. Coli. A reference feed rate which enables to maintain the acetate concentration at a specified level is first determined by means of optimal conditions. This feed rate is further used as the reference control trajectory within the NMPC algorithm. Finally, to avoid discretization problems during NMPC application, the on-line optimization is moved into a nonlinear programming strategy using the control vector parameterization approach (CVP). Some simulation results obtained on a fed-batch E. Coli bioreactor validate the efficiency of the proposed control strategy.

The objective of this study is to design a robust receding-horizon observer for systems described by nonlinear models with uncertain parameters. Robustification in the presence of model uncertainties naturally leads to the formulation of a nonlinear min-max optimization problem, which can either be solved numerically or which can be converted to a simpler minimization problem using linearization along a nominal trajectory and recent results in linear robust receding-horizon estimation. This method is first evaluated in simulation and then with real-life experimental data collected from continuous cultures of phytoplankton.

The optimization, monitoring and control of simulated moving bed (SMB) units require the use of a process model and the estimation of the model parameters. A systematic numerical procedure for determining parameters of SMB models from batch experiments is presented and evaluated. The unknown parameters are estimated by minimizing a cost function measuring the difference between experimental and simulated concentration profiles. In contrast with previous studies, parameter identifiability is studied and errors on the estimated parameters are calculated. A sensitivity analysis is used to design the experiments and to compare the identifiability of different chromatographic models. Then, the influence of local minima is evaluated by applying the numerical procedure on fictitious measurements generated from a model with known parameters.

In this work, attention is focused on two on-line optimization strategies for a batch MMA polymerization process, based on a physical process model and temperature measurements. Several model uncertainties are considered, including the variation of the heat transfer coefficient between the cooling jacket and the reactor medium as well as unknown parameters in the description of the gel effect. The first strategy is based on a worst-case analysis (formulated as a min–max problem), leading to an open-loop control scheme. The possibility to combine this scheme with a feedback controller is also evaluated. The second strategy is based on a scheduled optimization approach, i.e., the model parameters are periodically re-estimated on-line, and optimal trajectories are re-evaluated.

In this contribution, the open-loop optimal control (dynamic optimization) of a Simulated Moving Bed (SMB) chromatographic separation process is considered. The objective is to compute the optimal feed concentration and/or feed flow rate, over each switching period, with maximum flexibility. This problem is solved numerically using the combination of the control vector parameterization scheme with suitable state-of-the-art global nonlinear programming problem solvers. The advantages of the proposed approach are illustrated through the solution of two case studies, achieving significant improvements in the process productivity when compared to traditional feeding profiles.

The Simulated Moving Bed (SMB) technology is a continuous chromatographic process which is important in various fields, from sugar to enantiomer separation. In this paper, a systematic identification procedure for determining parameters of SMB models from batch experiments is validated with experimental SMB data. Parameters are first estimated from elution peaks. Then a cross-validation with SMB experiments is performed so as to assess whether the parameters identified from batch experiments may be used in a SMB model. This part of the work requires a careful modelling of the dead volumes within the SMB process.

State estimation methods allow the vehicle position and velocity to be reconstructed by combining information from sensors and vehicle model. From a security point of view, position and velocity have to be known with a high level of confidence in order, for example, to avoid vehicle collision. In this paper, a confidence interval observer is developed to enclose positioning variables with some confidence degree (or integrity level). Based on a vehicle model and intervals bounding, with some associated probability, the uncertain initial conditions, inputs, model parameters and measurements, a predictor provides intervals for the state estimate trajectory between two measurement times. At each measurement time, confidence intervals from the sensors and the predictor are combined with union and intersection operations so as to satisfy the specified integrity level. Finally, the shortest non-empty intervals are chosen among the safe intervals. This method is illustrated with simulation tests based on an autonomous underwater vehicle described by a nonlinear model.

In this paper, a macroscopic model of yeast culture is presented in the context of bioethanol production optimization. A discussion of the model formulation, based on the seminal work of Sonnleitner and Kappeli, is provided. A parameter identification of the resulting model is first performed and the dispersion of the parameter values is explained through a sensitivity analysis. Following the continuous process stability analysis, a closed-loop optimization of the bioethanol outflow rate using a model-free extremum-seeking algorithm is suggested. Numerical results show the efficiency of the method and its consistency with respect to a realistic time-scale performance.

hal-01154418; International audience; Submerged membrane bioreactors are increasingly applied for wastewater treamentbut requires a tight control of the membrane fouling so as to ensure safe and efficient operation.The objective of this paper is to design a nonlinear model predictive control to minimize theirreversible resistance while keeping the trans-membrane pressure, which is a good indicatorof membrane fouling, at an acceptable level. To this end, the manipulated variables are thepermeate flow and the air scouring flow, which allows the material layer formed on the membrane(in short the “cake”) to be detached. The NMPC structure is tested in simulation consideringa detailed simulator as the reference process, and a reduced-order model as the predictor. Theresults show that the process can be regulated until the irreversible resistance takes the mainrole in the fouling resistance. When this state is reached, a chemical cleaning is required, or alarger trans-membrane pressure has to be accommodated.

In this study, a simple dynamic model of a submerged membrane bioreactor (sMBR) is proposed, which would be suitable for process control. In order to validate the proposed model structure, informative data sets are generated using a detailed simulator built in a well-established environment, namely GPS-X. The model properties are studied, including equilibrium points, stability, and slow/fast dynamics (three different time scales). The existence of slow-fast dynamics is central to the development of a dedicated parameter estimation procedure. Finally, a nonlinear model predictive control is designed to illustrate the potential of the developed model within a model-based control structure. The problem of water treatment in a recirculating aquaculture system is considered as an application example.

Mathematical modeling and the development of predictive dynamic models are of paramount importance for the optimization, state estimation, and control of bioprocesses. This study is dedicated to the identification of a simple model of microalgae growth under substrate limitation, i.e., Droop model, and describes the design and instrumentation of a lab-scale flat-plate photobioreactor, the associated on-line and off-line instrumentation, the collection of experimental data, and the parameter identification procedure. In particular, a dedicated methodology for parameter identification is discussed, including the determination of an initial parameter set using an analytical procedure, the selection of a cost function, the evaluation of confidence intervals as well as direct and cross-validation tests. ispartof: Bioprocess and Biosystems Engineering vol:38 issue:9 pages:1783-1793 ispartof: location:Germany status: published

In this study, a low-cost RGB sensor is developed to measure online the microalgae concentration within a photo-bioreactor. Two commercially available devices, i.e., a spectrophotometer for offline measurements and an immersed probe for online measurements, are used for calibration and comparison purposes. Furthermore, the potential of such a sensor for estimating other variables is illustrated with the design of an extended Luenberger observer. ispartof: Sensors vol:15 issue:3 pages:4766-4780 ispartof: location:Switzerland status: published

This paper is concerned with a pilot-scale fixed-bed biofilter used for nitrogen removal from municipal wastewater. Process modelling yields a set of mass balance partial differential equations describing the evolution of the component concentrations along the biofilter. Based on sets of experimental data collected over several months, unknown model parameters are estimated by minimizing an output-error criterion. The resulting distributed parameter model and a few pointwise measurements of nitrate, nitrite, and ethanol concentrations are then used to design observers allowing the unmeasured biomass concentrations to be reconstructed on-line. First, it is demonstrated that asymptotic observers are not applicable to the given model structure. Then, a receding-horizon observer is designed and tested, showing a very satisfactory performance.

State estimation methods allow the vehicle position and velocity to be reconstructed by combining information from sensors and vehicle modelling. In a railway application, measurement signals from several sensors are available at asynchronous times, e.g., signals from odometers, radars and accelerometers. A Kalman filter can be easily designed based on a linear discrete-time model. However, in the train security package, only one accelerometer is available and moreover, this accelerometer is sensitive to rail track gradient. To circumvent the problem of estimation bias, a robust filter is developed, which takes uncertainties on the acceleration measurements and asynchronous data into account.