Keywords: Process Optimization, Control, Control Applications
Abstract: This survey explores recent results in the development of optimization algorithms and formulations for moving horizon estimation (MHE), nonlinear model predictive control (NMPC) and dynamic real-time optimization (D-RTO), with extrapolative nonlinear dynamic (e.g., first principle) models. We consider Nonlinear Programming (NLP) and NLP sensitivity as natural tools for formulation and efficient solution of optimization problems for these three tasks. For MHE, we develop a maximum likelihood formulation that directly incorporates nonlinear models, and is seamlessly adapted to updating the arrival costs. This approach is also extended easily to M-estimators, which essentially eliminate biased estimates due to gross errors. For NMPC, we develop formulations that incorporate well- understood stability and robustness properties, even for large, complex plant models. Finally, recent work has extended these capabilities from setpoint tracking objectives to more general stage costs that are economically based, thus leading to a robustly stable strategy for D-RTO. In concert with these problem formulations, the realization of MHE, NMPC and D-RTO requires the application of a fast NLP solver for time-critical, on-line optimization, as well as efficient NLP sensitivity tools that eliminate computational delay, and guarantee stability and robustness. Algorithms that meet these demands are explored and outlined for these tasks. Finally, a number of challenging distillation case studies are presented that demonstrate the effectiveness of these optimization-based strategies.

Keywords: Batch Process Modelling, Optimization and Control, Modelling and Identification, Process Optimization, Control
Abstract: In this paper a novel method for batch-to-batch modelling and optimization, Iterative Learning Partial Least Squares Optimization (IL-PLSO) is proposed. This method uses a recursive technique to update a multi-way PLS model so that it is able to track the varying dynamics from one batch to the next. Based on the model obtained at the end of one batch, a Quadratic Programme (QP) is used to identify the required trajectory for the primary manipulated variable in the subsequent batch to ensure that the target end-point quality is met. This target quality can be gradually increased to optimise the productivity, or yield of the process. The capabilities of the proposed IL-PLSO method are illustrated through its application to optimise the end-point product quality of a benchmark simulation of a fermentation process. In this application, the proposed algorithm is able to identify an optimal trajectory for the manipulated variable after approximately 10 batches. The results are shown to compare very favourably with alternative approaches.

Keywords: Batch Process Modelling, Optimization and Control, Process and Performance Monitoring, Fault Detection, Supervision and Safety
Abstract: In this paper, a new statistical process analysis and quality prediction method is proposed for multiphase batch processes. A two-level phase division algorithm is designed to capture and trace quality-related inner-phase evolution which in general goes through three statuses sequentially, i.e., transition, steady-phase and transition. Partial least squares (PLS), canonical correlation analysis (CCA) and qualitative trend analysis (QTA) are effectively combined to distinguish different inner-phase process statuses. Their different characteristics are then analyzed respectively for regression modeling and quality analysis. Meanwhile, the uneven-length problem of batch processes is handled in different inner-phase parts so that online quality prediction can be performed at each time. The application to the injection molding process illustrates the feasibility and performance of the proposed algorithm.

Keywords: Batch Process Modelling, Optimization and Control, Control Applications, Inferential sensing, State Estimation and Sensor development
Abstract: The Czochralski process is the only method used commercially for production of monocrystalline silicon for semiconductor and solar cell applications. This paper explores the use of mathematical modeling as an aid in estimation of system state variables in the standard Czochralski process. A state-space model of the process is presented, describing the dynamics of the crystal radius and meniscus height with crystal radius as measured output. For the purpose of estimating the actual crystal radius during growth, three types of state estimators are developed based on the state-space model; the Kalman lter, the extended Kalman lter and the unscented Kalman lter. It is found that the latter two provide highly accurate state estimates with excellent noise suppression.

Keywords: Batch Process Modelling, Optimization and Control, Control Applications, Inferential sensing, State Estimation and Sensor development
Abstract: This paper suggests a method to control the particle size distribution in semi-batch emulsion homopolymerizations under structural plant-model mismatch. In this approach, firstly a nominal monomer feed input trajectory is applied to the plant up to a predefined time instant after the start of the batch (mid-course). By means of a calorimetric observer, all states of the system except the particle size distribution are estimated using the available measurements. The estimated states and the measured PSD at the mid-course of the process are used as the initial condition for an optimization which is done to compute the trajectory of the monomer feed from the mid-course up to the end of the batch. In this optimization, considering a structural plant-model mismatch, a hybrid model which comprises the nominal model of emulsion polymerization and an empirical component that corrects the predictions of this nominal model is used.

Keywords: Batch Process Modelling, Optimization and Control, Process Optimization, Control, Process Scheduling and Decision support, Integration between Scheduling and Control
Abstract: With the shift to high-value, low volume production, the problem of short term production scheduling for multipurpose/multiproduct batch processes has been realized as an important problem in industrial plant operations. Also, any industrial batch process invariably involves the production of harmful wastes along with the useful products. Due to stricter environmental regulations, these wastes cannot be disposed off without prior treatment. However, most scheduling problems considered in literature deal with objectives such as maximization of profit due to product sales, minimization of time span, minimize product tardiness, etc. with no consideration to minimizing downstream waste production cost. So in this study, the problem of short term batch scheduling is approached with the dual objective of maximizing profit and at the same time minimizing the downstream waste treatment cost. Since the problem of short term batch scheduling with waste management is inherently complex, a number of different decomposition approaches to solve complex multi-level optimization problems are presented. The model co-ordination approach is applied to a case example and the results elucidate the fact that the optimal solution is achieved at significantly lesser computational complexity, and agrees with the solution obtained when the optimization problem is solved without decomposition. The case example also illustrates the effectiveness and efficiency of model coordination approach in terms of computational effort.

Keywords: Batch Process Modelling, Optimization and Control, Process Optimization, Control
Abstract: A standard run-to-run model-based optimization approach consists of updating the model after every run and, then re-optimizing for the next run with the updated model. However, in the presence of model structure error, the convergence of this two-step approach to the true plant optimum cannot be guaranteed. This paper presents an alternative approach where a correction term is added to the model outputs such that the new updated model parameters simultaneously satisfy the identification and optimization objectives. The effect of parametric uncertainty on optimization results is considered explicitly by minimizing the expected value of the cost function. For computational efficiency, the uncertainty propagation step is performed using Polynomial Chaos expansions. The methodology is illustrated using a fed-batch process for penicillin production. When compared to the standard two-step approach, the proposed methodology provides significant increase in the amount of penicillin.

Keywords: Modelling and Identification
Abstract: A set of methods for identification of continuous-time transfer function models for integrating processes with time delay is proposed. The step, piecewise constant and piecewise linear inputs are considered which indeed cover most of the input signals commonly used in industries. For all of the three types of input signals, estimation equations to simultaneously obtain model parameters and the time delay are derived. The final parameter estimation equations are in a form suitable for the least-squares solution. Mathematical formulation of the methods is presented using the example of an integrating process with a first order lag dynamics and a zero which can be extended for other structures. An instrumental variable method to deal with the bias issue in least-squares solutions is used. Simulation results are presented to demonstrate the efficacy of the algorithms and their relative performance.

Keywords: Modelling and Identification, Process Optimization, Control
Abstract: Optimal experiment design for system identification involves determining an optimal input that is used to perturb the system so that the resulting input-output data is maximally informative. Plant friendly identification requires that constraints on input move sizes, output sizes or variance and experiment time be respected. The solution to the optimum input design problem depends on the unknown parameters to be estimated which is often approximated by an initial estimate. Use of the estimate is likely to result in loss in performance or violation of the constraints. An alternative is to formulate a robust optimization problem with uncertain parameters. The contribution of this work is to use the uncertainty sets originating from a prior identification exercise to solve a robust plant friendly input design problem. The methodology is derived for a general class of systems illustrated using numerical simulations. Simulations validate the expectation that the constraints are probabilistically more likely to be satisfied using the robust design than a nominal design based on uncertain parameters.

Keywords: Modelling and Identification, Process Optimization, Control, Control Applications
Abstract: Multi-stage spray drying is an important and widely used unit operation in the production of food powders. In this paper we develop and present a dynamic model of the complete drying process in a multi-stage spray dryer. The dryer is divided into three stages: The spray stage and two fluid bed stages. Each stage is assumed ideally mixed and described by mass- and energy balances. The model is able to predict the temperature, the residual moisture and the particle size in each stage. Process constraints are also proposed to predict deposits due to stickiness of the powder. The model predictions are compared to datasets gathered at GEA Process Engineering's test facility. The identified grey-box model parameters are identified from data and the resulting model fits the data well. The aim of the model is to facilitate development of efficient control and real-time optimization algorithms for multi-stage spray dryers. The complexity of the model has been selected such that it is suitable for this purpose in an economic optimizing MPC framework.

Keywords: Modelling and Identification
Abstract: The estimation of time-varying parameters in continuous-time nonlinear systems is considered under the framework of the modulating functions method. The parameter is approximated as a finite Fourier series, which is reconstructed from the estimated Fourier spectral coefficients. Unlike the popular polynomial approximation, this approach is general enough for piecewise smooth parameter changes. The locations of abrupt jumps are acurately identified by the presence of Gibbs phenomenon. The global Fourier spectral coefficients are then used to extract local finite Gegenbauer polynomial series to recover smooth parameter variations between the jumps. This method of resolution of the Gibbs phenomenon avoids the necessity of estimating a large number of Fourier coefficients for series convergence. A van der Pol oscillator simulation example is included to demonstrate the performance of the approach.

Keywords: Modelling and Identification, Process Optimization, Control, Control Applications
Abstract: This paper develops an on-line model parameter identification approach for multivariable systems, which are nonlinear in terms of state representation and/or in terms of parameters. Combining the observation theory and the model based predictive control theory, an optimal closed loop experiment design for on-line identification of model parameters is given. During only one experiment, an optimal time-varying input is computed to optimize a criterion, while the unknown model parameters are estimated at the same time. The criterion is based on the sensitivities of the model outputs with respect to the unknown parameters that are estimated. The approach does not require to measure all the process state. Moreover output constraints allow to maintain the behaviour into a prescribed region and/or stabilize the process in closed loop. This approach is illustrated through an unstable rolling delta wing with one input, two measured states and five unknown constant model parameters.

Keywords: Modelling and Identification
Abstract: The quality of a model determines the closed loop performance of model predictive controllers. However, identification of high quality multivariable models is a time and energy intensive exercise. The industrial model predictive controllers are designed using large dimensional multivariable models and they are often identified using ad-hoc single input bump tests. A novel multivariable input design approach is developed using a modified model predictive control objective function. It is shown that the proposed input design approach is trace optimal with respect to the covariance of model parameters. The approach is shown to work well in closed loop on both well and ill-conditioned processes even under model-plant mismatch while meeting input and output constraints.

Keywords: Process and Performance Monitoring, Fault Detection, Supervision and Safety
Abstract: Independent component analysis (ICA) is a newly emerging feature extraction method for non-Gaussian process monitoring. However, the extracted feature by ICA may not represent the original process data well, which can result in the degraded monitoring performance. In this paper, an improved ICA method based on the minimum mean squared prediction error criterion is proposed for process monitoring. A new criterion which can make the extracted non-Gaussian feature be efficient representation for the original process data is constructed as the objective function of the improved ICA by integrating the maximum negentropy criterion of the conventional ICA with the minimum mean squared prediction error criterion. Then the gradient ascent algorithm is applied to optimize the constructed objective function for seeking the feature extraction directions. Finally, a monitoring statistic is built based on the extracted feature to detect process faults. The simulation studies on the Tennessee Eastman benchmark process demonstrate that the improve ICA is more effective than the conventional ICA for improving the monitoring performance in terms of the fault detection rate.

Keywords: Process and Performance Monitoring, Fault Detection, Supervision and Safety, Control Applications
Abstract: The paper presents main features of a performance monitoring system, which, in addition to variables normally registered in industrial plants (controller output OP, controlled variable PV and set-point SP), makes use of additional variables made available by intelligent instrumentations and field bus communication systems. Experimental runs on a pilot plant scale have been carried out in order to introduce different types of valve malfunctions and to define suitable indexes (KPI) able to diagnose them. Subsequently, threshold values for the indexes have been calibrated and a logic has been developed to assign different performance grades. It is shown how the Travel Deviation allows specific evaluation of valve status and to detect different causes of malfunctioning. The same logic is implemented in an advanced release of an existing performance monitoring system and advantages in the accuracy of diagnosis are shown. Finally the system has been successfully validated by online implementation for control loops assessment of an industrial power plant.

Keywords: Process and Performance Monitoring, Fault Detection, Supervision and Safety, Modelling and Identification, Process Optimization, Control
Abstract: Complex technical systems, e.g. chemical plants, are prone to equipment failures. To ensure the safe operation of such systems, the occurrence of a fault has to be reliably detected. Set-based validation and identification methods are well suited for this problem as they are flexible with respect to modeling uncertainties and as they can provide guaranteed results. One of the main challenges of set-based approaches is, however, the complexity of underlying computations. Simplifying the problem formulation via a suitable approximation of the model is one way to reduce the computational effort. However, to retain the ability to diagnose faults, the underlying structure of the model has to be taken into account. We present a method to reduce the problem formulation based on causal reasoning and lifting technique that orders the system states according to the effects of occurring faults. We present an approach to derive such a reduction and illustrate its application considering two 5-tank configurations.

Keywords: Process and Performance Monitoring, Fault Detection, Supervision and Safety, Control Applications, Modelling and Identification
Abstract: We address in this paper the application of a recently proposed MPC performance monitoring method to a rigorously simulated industrial process. The methodology aims at detecting possible sources of suboptimal performance of linear offset-free MPC algorithms by analysis of the prediction error sequence, discriminating between the presence of plant/model mismatch and incorrect disturbance/state estimation, and proposing for each scenario an appropriate corrective action. We focus on the applicability of the method to large-scale industrial systems, which typically comprise a block structure, devising efficient and scalable diagnosis and correction procedures. We also discuss and support the application of this method when the controlled plant shows a mild nonlinear behavior mainly associated with operating point changes. A high-fidelity dynamic simulation model of a crude distillation unit was developed in UniSim Design and used as representative test bench. Results show the efficacy of the method and indicate possible research directions for further improvements.

Keywords: Process and Performance Monitoring, Fault Detection, Supervision and Safety
Abstract: A model based fault detection and diagnosis scheme is proposed in this paper and a lumped parameter MIMO model has been considered. The characteristics of single as well as multiple faults using residuals are investigated in air-conditioning (AC) system provided in passenger coach of an Indian Railways. The residuals of the fault are generated by using real life simulator and model. The faults are diagnosed by classifying residual patterns. The classifications of residual patterns have been done by using three approaches: hexadecimal decision table, least square support vector machine (LS-SVM) and a novel approach named as hybrid classification approach.

Keywords: Batch Process Modelling, Optimization and Control, Process and Performance Monitoring, Fault Detection, Supervision and Safety
Abstract: One most significant challenge in batch process monitoring, compared to continuous process monitoring, is handling three-dimensional historical data for batch processes. Conventional batch process monitoring approaches involve unfolding of such historical data into two dimensions. However, this simple unfolding technique is not applicable if batch duration is not constant. For monitoring of uneven length batch processes, function space analysis based principal component analysis (FSPCA) had been proposed earlier. However, this approach has some limitations because PCA is not effective when it comes to fault diagnosis of highly nonlinear systems. In this paper, a new technique called function space correspondence analysis (FSCA) is proposed for monitoring of uneven length batch processes. The proposed FSCA technique is intended to overcome such limitations and to achieve improved diagnostic performance. The improved performance is due to better discriminatory ability of the CA algorithm. Improved diagnostic capabilities of the proposed FSCA technique is demonstrated using fed batch penicillin cultivation process as a case study. The monitoring results demonstrate improved diagnostic performance with FSCA.

Keywords: Process Optimization, Control, Modelling and Identification, Control Applications
Abstract: Oil price significantly affects the economic performance of an oil field. In this study, Model Predictive Control (MPC) is used as a reference tracking method to provide corrective control action by meeting an objective, which balances short term and long term effects, based on an updated oil price. The long-term optimized production profile is based on a predicted oil price while the short-term objective is optimized with respect to the predicted short-term oil price by applying MPC. The fluid flow equations must be solved many times to find an optimal input for a short-term horizon; therefore, instead of a large-scale nonlinear reservoir model, local linear models are used to decrease the computation time for short-term optimization. The developed method is tested with several oil price scenarios on a synthetic case. The efficiency of the developed method is confirmed by comparing its economic performance to that using open loop long-term optimization.

Keywords: Process and Performance Monitoring, Fault Detection, Supervision and Safety, Inferential sensing, State Estimation and Sensor development
Abstract: Data from a medium-scale horizontal flow loop test facility is used to test fault diagnosis in managed pressure drilling. The faults are downhole incidents such as formation influx, fluid loss, drillstring washout, pack-off, and plugging of the drill bit, which are important to detect and handle in order to avoid downtime and possibly dangerous situations. In this paper a fault diagnosis scheme based on an adaptive observer and the generalized likelihood ratio test is applied on the experimental data. The different types of faults are detected and their location are isolated using friction parameter estimates. Results indicate that the method can in most cases identify the type of fault, whereas the location is sometimes more uncertain.

Keywords: Inferential sensing, State Estimation and Sensor development, Modelling and Identification
Abstract: In the last two decades, several methods based on sequential Monte-Carlo (SMC) and Markov chain Monte-Carlo (MCMC) have been proposed for Bayesian identification of stochastic non-linear state space models (SSMs). It is well known that the performance of these simulation based identification methods depends on the numerical approximations used in their design. We propose the use of posterior Cramer-Rao lower bound (PCRLB) as a mean square error (MSE) bound. Using PCRLB, a systematic procedure is developed to analyse the estimates delivered by Bayesian identification methods in terms of bias, MSE, and efficiency. The efficacy and utility of the proposed approach is illustrated through a numerical example.

Keywords: Inferential sensing, State Estimation and Sensor development, Modelling and Identification, Batch Process Modelling, Optimization and Control
Abstract: Physiologically based pharmacokinetics(PBPK) model can predict absorption, degradation, execration and other metabolism in drug delivery system. Thus it can be useful for regulating dose and estimating drug concentration at a particular time during the clinical demonstration. PBPK model is expressed as a set of differential equation with various parameters. Bio-chip experimental data are often noisy and sparse. This makes it difficult to estimate parameters with conventional least squares approaches. The resulting parameters often have a large confidence region. This work presents a Bayesian inference algorithm with an objective function suitable for PBPK model. A Markove Chain Monte Carlo(MCMC) method is employed to estimate the posterior distribution of the parameters. We illustrate the approach with a Tegafur delivery system.

Keywords: Process Optimization, Control, Interaction Between Design and Control
Abstract: This paper presents a simple analytical method for the design of full matrix PI controller based on the direct synthesis approach. By proposing the practically desired closed-loop diagonal transfer function to reduce interactions between individual loops, analytical expressions for PI controller are derived for several common types of process models, including first order plus time delay models and second order plus time delay models. Compared with the existing direct synthesis approaches, the proposed controller design method requires no approximation of the inverse of process model or Maclaurin's series expansion. Furthermore, it is applicable to high dimensional multivariable systems with satisfactory performance and robustness. Several examples are introduced to demonstrate the effectiveness and simplicity of the design method.

Keywords: Process and Performance Monitoring, Fault Detection, Supervision and Safety, Inferential sensing, State Estimation and Sensor development
Abstract: Multivariate data analysis techniques are widely used in getting better insight into the processes in the fields like Chemometrics, speech processing, plant wide oscillation detection and astronomy. In the present study, the problem of extracting the spectra of a pure component from Near Infrared (NIR) mixture spectra containing a known diluent is tackled. Different multivariate data analysis methods such as Ordinary Least Square (OLS), Principal Component Regression (PCR) and Non Negative Matrix Factorization (NMF) are modified to solve the problem. It is shown that including partial knowledge such as the spectra of the known diluent in the data analysis techniques results in better estimation of the pure component spectra.

Keywords: Control Applications
Abstract: The control of a Rotary Inverted Pendulum (RIP) is a well-known and a challenging problem that serves as a popular benchmark in modern control system studies. The task is to design controllers which drives the pendulum from its hanging-down position to the upright position and then hold it there. The swing up is achieved using an energy based controller. In energy based control the pendulum is controlled in such a way that its energy is driven towards a value equal to the steady-state upright position. Then a mode controller switches between the swing-up controller and stabilizing controller near the upright position. For stabilization control, two control techniques are analyzed. Firstly, a sliding mode controller (SMC) is designed to stabilize the pendulum. Secondly, a state feedback controller is designed that would maintain the pendulum upright and handle disturbances up to a certain point. The state feedback controller is designed using the linear quadratic regulator (LQR). The responses of the LQR controller and SMC controller are compared in simulation.

Keywords: Control Applications, Process Optimization, Control, Inferential sensing, State Estimation and Sensor development
Abstract: In this work, the authors have proposed a constrained nonlinear model predictive control (NMPC) scheme for a boiler-turbine system. The proposed control scheme relies on the non-linear state space model proposed by Bell and Astrom, 1987 for prediction. A derivative-free Kalman filter has been designed to estimate the state variables of the boiler-turbine unit. These estimated values have been used as an initial condition for predicting the future states and outputs in the NMPC formulation. In order to account for model-plant mismatch and to achieve offset-free control, innovation based correction of predicted states and outputs as suggested by Ricker, 1990 has been incorporated in the proposed NMPC formulation. The extensive simulation studies show that the proposed control scheme effectively handles the input constraints of the boiler-turbine unit and meets the required electrical demand without requiring careful selection of operating points.

Keywords: Modelling and Identification, Control Applications
Abstract: Abstract: In this paper, we consider a closed loop subspace identification problem. Here the open loop processes are unstable. By using the subspace identification algorithm, the closed loop system is first identified. The plant dynamics are extracted from the identified closed loop system. Three unstable processes are simulated and identified by the MON4SID algorithm.

Keywords: Process and Performance Monitoring, Fault Detection, Supervision and Safety
Abstract: A novel risk-based fault detection method has been developed. The proposed method provides a dynamic process risk indication based on the probability of happening a fault and its consequence. In this method instead of generating an alarm based on residuals or signals an alarm is activated only when the calculated risk of operation exceeds the acceptable threshold. This is an important concept as it can funnel the attention and effort of operators to the faults which poses the most operational or safety risk. Application of this new risk-based approach provides early warning of the fault as well as the associated risk with the fault. Methodologies were developed to apply the concept with model based fault detection algorithm as well as multivariate history based fault detection techniques. In this paper we show the model based approach by combining Kalman filter with the risk based approach. The history-based approach was demonstrated using principal component analysis (PCA). This method has more power in discerning between operational changes and abnormal conditions which have potential to cause accidents.

Keywords: Process and Performance Monitoring, Fault Detection, Supervision and Safety, Inferential sensing, State Estimation and Sensor development, Interaction Between Design and Control
Abstract: In this article, a novel dynamic Bayesian network based networked process monitoring approach is proposed for fault detection, propagation pathway identification and root cause diagnosis. First, process network structure is designed according to the prior process knowledge including process flow sheets and used to characterize the causal relationships among different measurement variables. Then, the dynamic Bayesian network model parameters including the conditional probability density functions of different nodes are learned from historical process data to quantify the causality among those variables. Further, the new monitoring index is derived from the likelihoods of the entire process network for detecting abnormal operating events. With the captured process abnormality, the novel probabilistic contribution indices within Bayesian network are proposed to identify the major fault effect variables. Subsequently, the fault propagation pathways from the downstream backwards to upstream process are isolated through the variable contribution indices and hence the ending nodes of the identified pathways are determined as the root-cause variables of the abnormal events. The proposed approach is applied to the Tennessee Eastman Chemical process and the results show that the presented method can accurately detect abnormal events, identify fault propagation pathways, and diagnose the root-cause variables.

Keywords: Process Optimization, Control, Modelling and Identification, Control Applications
Abstract: Anaerobic digestion with biogas production has both economic and environmental benefits. 25% of all bioenergy in the future could potentially be sourced from biogas. Although anaerobic digesters have seen wide applicability, they typically perform below their optimum performance as a consequence of the complexity of the underlying process. There is thus a requirement for a configurable monitoring and optimization system with associated sensors to optimize the production of biogas, combined with a degree of flexibility to account for the quality and content of the digestate. This work involves the development of a generic advanced process control system to optimize the performance of anaerobic digesters.

Keywords: Process Optimization, Control, Inferential sensing, State Estimation and Sensor development, Process and Performance Monitoring, Fault Detection, Supervision and Safety
Abstract: The sensor network design procedure developed by Nabil and Narasimhan (2012), which relates process economics and data reconciliation, involves the formulation of a mixed integer cone program. The solution to this problem yields the globally optimal sensor network. A branch and bound procedure can be used to nd the global optimum, however, for systems with large numbers of variables, this approach may require a large amount of computational time to nd the solution. In this paper, a specialized branch and bound algorithm is proposed for solving the sensor network design problem, which uses certain heuristics to obtain a solution faster. One involves a low rank factorization to reduce the size of the relaxed problem. The other involves an approximation of the global lower bound for the branch and bound solution.The utility of this algorithm is demonstrated on a simple ow network, a small but realistic evaporator system, and a medium sized steam metering network.

Keywords: Process Optimization, Control
Abstract: An online adaptive dynamic programming based iterative algorithm is proposed for a two-player zero sum linear differential game problem arising in the control of process systems affected by disturbance. The objective in such a scenario is to obtain an optimal control policy that minimizes the specified performance index or cost function in presence of worst case disturbances. Conventional algorithms for designing the worst case optimal control require full knowledge of system dynamics. The algorithm proposed in this paper is partially model-free and solves two-player zero sum linear differential game problem without knowledge of state and control input matrices.

Keywords: Process and Performance Monitoring, Fault Detection, Supervision and Safety, Modelling and Identification, Process Optimization, Control
Abstract: National critical infrastructures like power plants, grids, water distribution system etc employ a hierarchy of controllers exchanging data over a network. They employ sophisticated control algorithm implemented in software. Various researches have examined the attack scenarios in such embedded control systems from control theoretic perspectives. In this paper we revisit these theoretical attacks and postulate that such attacks could be detected by statistical techniques and hence may be used to design security monitors. The postulate is confirmed by simulation results.

Keywords: Modelling and Identification, Process Optimization, Control, Control Applications
Abstract: Self-optimising control is a useful concept to select optimal controlled variables from a set of candidate measurements in a systematic manner. In this study, use self-optimizing control tools and apply them to the specific features of sewer systems, e.g. the continuously transient dynamics, the availability of a large number of measurements, the stochastic and unforeseeable character of the disturbances (rainfall). Using a subcatchment area in the Copenhagen sewer system as a case study we demonstrate, step by step, the formulation of the self-optimising control problem. The final result is an improved control structure aimed at optimizing the losses for a given control objective, here the minimization of the combined sewer overflows despite rainfall variations.

Keywords: Process and Performance Monitoring, Fault Detection, Supervision and Safety, Modelling and Identification, Control Applications
Abstract: Model Predictive Control (MPC) is a powerful tool in the control of large scale chemical processes and has become the standard method for constrained multivariable control problems. Hence, the number of MPC applications is increasing steadily and it is being used in application domains other than petrochemical industries. A common observation by the industrial practitioners is that success of any MPC application requires not only efficient initial deployment but also maintenance of initial effectiveness. To this end, we propose a novel high level automated support strategy for MPC systems. Such a strategy consists of components such as performance monitoring, performance diagnosis, least costly closed loop experiment design, re-identification and autotuning. This work presents the novel technological developments in each component and demonstrates them on a distillation column case study. We show that automated support strategy restores nominal performance after a performance drop is detected and takes the right course of action depending on its cause.

Keywords: Modelling and Identification, Batch Process Modelling, Optimization and Control
Abstract: We present an algorithm for projective integration that is computationally efficient for integrating systems of differential equations with multiple time-scales. Adaptive projective integration is a technique that uses a few inner integration steps to generate data to fit to a local reduced-order model. This reduced-order model is then used to extrapolate forward in time to estimate the states at some future time. This inner-outer integration is iterated until the desired integration is complete. The method uses an adaptive projective horizon to control for error generation during the integration. By examining an example Brusselator system, consisting of three non-linear differential equations, we show two orders of magnitude savings in computational time using adaptive projective integration over explicit Euler's method.

Keywords: Process Optimization, Control, Control Applications
Abstract: In this contribution we study the application of non-linear model predictive control to a continuous polymerization of acrylic acid in tubular reactors with multiple side injections of monomer. The background of this work is to transfer the polymerization from semi-batch to continuous operation.Model Predictive Control (MPC) is the obvious candidate to control such a multi-input system. The reactor configuration and polymerization reaction make the application of MPC very challenging. The controller employs a discretized dynamic pde model of the process and optimizes the productivity of the plant online while keeping the product quality parameters within the predefined constraints. The spatial domain of the model is discretized by applying the Weighted Essentially Non Oscillatory (WENO) scheme. Besides testing the controller for a nominal case in which the control model is identical to the existing plant, the controller has been simulated for two model-plant mismatch cases, caused by fouling and feed impurities. For the both cases, a moving window estimation scheme is applied to estimate the unknown parameters and to update the model used by NMPC. The results show that the controller can increase the product throughput considerably and has a robust performance in the presence of the modelplant mismatch. Moreover, the effect of formulating the quality constraints as soft constraints is studied.

Keywords: continuous polymerization, tubular reactors, optimizing control, pde models, parameter estimation.

Keywords: Modelling and Identification, Control Applications, Inferential sensing, State Estimation and Sensor development
Abstract: A method for a pump-integrated estimation and control of the fluid level in open storage tanks is proposed. Automatic control with a variable speed pump can reduce the energy consumption significantly, compared to fixed speed solutions. The proposed method uses only sensors that are by default integrated into the pumping unit. The modeling and tuning effort is higher than for fixed speed solutions, however. We outline automatic procedures for process identification and controller synthesis that reduce the user workload to a minimum and therefore can be expected to increase the level of acceptance of an energy efficient automatic control solution despite the added complexity.

Keywords: Batch Process Modelling, Optimization and Control
Abstract: A population balance equation (PBE) model that accounts for drop breakage and coalescence in high pressure homogenization was used for emulsion product design. Six adjustable parameters were estimated by nonlinear optimization from measured drop volume distributions at a specified operating condition. The values of two parameters were estimated at four different homogenization pressures and interpolated to allow improved prediction over a range of pressures. Using two alternative objective functions, the parameterized model was used to determine the pressure of each homogenization pass needed to achieve the target drop size distribution at the final pass. Homogenization experiments performed to validate the model predictions produced measured distributions in very good agreement with two target distributions.

Keywords: Batch Process Modelling, Optimization and Control, Modelling and Identification, Inferential sensing, State Estimation and Sensor development
Abstract: In order to realize stable production in the steel industry, it is important to control molten steel temperature in a continuous casting process. The present work aims to develop a gray-box model that predicts the molten steel temperature in the tundish (TD temp). In the proposed approach, two parameters in the first-principle model, i.e., overall heat transfer coefficients of ladle and tundish, are optimized for each past batch separately, then the relationship between the two parameters and measured process variables is modeled through random forests (RF). In this inner gray-box model, the statistical models update the physical parameters according to the operating condition. To enhance the accuracy of TD temp estimation, another RF model is developed which compensates errors of the inner gray-box. The proposed approach was validated through its application to real operation data at a steel work.

Keywords: Batch Process Modelling, Optimization and Control, Control Applications
Abstract: In this study, we propose a method to estimate the parameters of a control model for a batch process by using previous batch data. We focus on the case of multistep batch processing, where appropriate control input values often exhibit strong dependency on the prior processing history of the feed (called “feed characteristics” hereafter), e.g., the equipment or operating conditions used in the previous processing steps. In such cases, it is a common practice to use the data from those previous batches with identical feed characteristics as the new batch. As batch operations become more complicated, however, the variety of feed characteristics is increased and consequently the chance of finding recent batch data with identical feed characteristics is reduced. To combat the shortage of usable data in this context, it is important to enable the utilization of not only data from batches of identical feed characteristics but also those from batches of “similar” feed characteristics. This paper attempts to address this need in a practical manner. By using MANOVA (multivariate analysis of variance), a popular statistical inference method, statistical similarities among the estimated parameter values for different feed characteristics can be evaluated and substitutable sets of the feed characteristics can be identified. Results from the statistical analysis can increase the amount and/or recency of the data used in the batch control input calculation. We suggest some specific rules for selecting among available previous batch data by considering both the feed characteristic similarity and time-immediacy. The proposed method has been tested on real manufacturing industrial data and the results showed practical viability and significant potentials of the method.

Keywords: Batch Process Modelling, Optimization and Control, Process Optimization, Control
Abstract: Model Predictive Control (MPC) and its first order approximation, the Neighboring Extremals (NE) have been used for real-time optimal control in the presence of model uncertainties for several decades. Traditionally, both MPC and NE would only correct for deviations in states considering the underlying model to be nominal - a procedure that is valid for additive disturbances. However, in the presence of model uncertainties, a simple illustrative example in this paper shows that such a MPC scheme or a NE controller could cause corrections in the wrong direction, thereby deteriorating performance. The paper, thus, addresses reformulating NE feedback considering sensitivities with respect to the model parameters. The feedback then has two components - one based on state deviations and the other based on parameter deviations. Note that this formulation also requires some primitive form of parameter estimation. The illustrative example shows the efficacy of this approach and the importance of incorporating the knowledge of parameter variations in real-time optimal control.

Keywords: Process Optimization, Control, Batch Process Modelling, Optimization and Control, Control Applications
Abstract: The use of membrane microfiltration in the production of beer is becoming an attractive alternative. Due to fouling there is the need of performing frequent backflushes and the more expensive membrane-damaging chemical cleanings. An optimal operation of the installation would minimize the costs by reducing the number of chemical cleanings and the consumption of energy while complying with the task of processing, on time, the assigned amount of beer. This paper discusses the opportunities for a more efficient, optimal operation of such a plant by using a model based real time optimization scheme that uses the framework of predictive control but with an economic motivated target.

Keywords: Process Optimization, Control
Abstract: The paper presents an adaptive strategy to reject periodic disturbances with unknown period based on a combination of model predictive control and repetitive control. A novel period estimator is presented. For the integer period case, the estimator is designed based on integer programming. For the non-integer period case, it is designed based on a two-step optimization, namely integer programming followed by a constrained least square method. With the estimated period, feedforward compensation is made to improve the tracking performance asymptotically. Simulation results are given to show the effectiveness of the algorithm.

Keywords: Inferential sensing, State Estimation and Sensor development, Modelling and Identification
Abstract: We propose an algorithm for designing optimal inputs for on line Bayesian identification of stochastic non-linear state-space models. The proposed method relies on minimization of the posterior Cramer Rao lower bound derived for the model parameters, with respect to the input sequence. To render the optimization problem computationally tractable, the inputs are parametrized as a multi-dimensional Markov chain in the input space. The proposed approach is illustrated through a simulation example.

Keywords: Batch Process Modelling, Optimization and Control, Modelling and Identification
Abstract: Recursive system identification is an important problem in many advanced control techniques, such as adaptive control. This paper presents a new approach of two dimensional recursive least squares identification method suitable for batch processes. In this way, system identification is carried out not only using the information from time direction within the batch but also from batch to batch direction. A constraint term is incorporated in the cost function to reduce parameters varying. A guideline for selecting weight matrix in application is also provided. Furthermore, simulation results based on the data obtained from a model of injection moulding, a typical batch process, are illustrated to testify the superiority of the proposed method over the conventional recursive leasts squares.

Keywords: Control Applications, Process Optimization, Control, Modelling and Identification
Abstract: Convection dominated systems described by first order hyperbolic PDE models are common in chemical engineering. Use of such PDEs in model based control requires a large number of states for its representation and consequently requires significant computational effort. Recently, Sudhakar et al. (2013a,b) have proposed to use method of characteristics (MOC) to obtain reduced order models for such systems and have demonstrated its use in model based control. The implementation of MOC requires the use of repeated solution of initial and boundary value problem. In this work we propose to use differential transform technique to obtain approximate analytical solution for these problems, which result in significant reduction in computational effort. The technique is demonstrated in two case studies involving fixed bed reactor and plug flow reactor. The comparison of dynamic response and the computational load from using DT and numerical integration of the resulting differential equations indicates the effectiveness of using DT in MOC.

Keywords: Process Optimization, Control, Control Applications, Modelling and Identification
Abstract: An identification experiment procedure for the optimizing control of reactor/separator processes with multiple material recycle streams is proposed. Steady state relation between the cost related variables and the operation variables are approximated through response surface models. To minimize the perturbation due to the identification experiments, Latin hypercube design is applied to the selection of the sampling points and their implementation sequence is determined by the application of the traveling salesman problem. The procedure is applied to the simulated HDA (hydrodealkylation of toluene) plant to identify an unconstrained optimal operating condition.

Keywords: Batch Process Modelling, Optimization and Control, Modelling and Identification, Interaction Between Design and Control
Abstract: The paper presents development and validation of coal mill model (including the action of classifier) to be used for improved coal mill control. The model is developed by using the mass and heat balance equations of the coal mill. Genetic Algorithm is used to estimate the unknown parameters that are used in the model validation. The advantage is that the raw data used in modeling can be obtained without any extensive mill tests. The simulation results show a satisfactory agreement between the model response and measured value. The model is validated by using data collected from the power plant. Apart from the conventional PID controller, inorder to ensure tight control with less overshoot and to handle constraints Model Predictive Controller is designed to maintain outlet temperature and pulverized coal flow at desired setpoint value.

Keywords: Process Optimization, Control, Modelling and Identification, Control Applications
Abstract: For producing sulfur trioxide, sulfur dioxide is oxidized to sulfur trioxide in the presence of a catalyst. Industrially, sulfur trioxide is produced by the contact process which consists of several catalytic beds and inter-cooling stages between them. This process is highly energy intense and in exible due to long periods of start-up and shut-down caused by the thermal inertia. In order to improve the exibility of the production process, a new approach for producing sulfur trioxide is currently being investigated. A micro-structured reactor has been constructed by the Karlsruhe Institute for Technology (KIT) which consists of only one cooling passage. In our work, a three dimensional dynamic reactor model was developed that describes the dynamic behavior of the micro-structured reactor. This model was used to simulate the eciency and the distribution of temperature and reactants in the system and in particular for investigating the start-up of the reactor. A control strategy was developed and tested in simulations in order to react on disturbances and to improve the start-up time. The simulations showed that a conversion of approx. 98.5 % can be reached and by a suitable control strategy the start up time can be improved signicantly.

Keywords: Process and Performance Monitoring, Fault Detection, Supervision and Safety
Abstract: Power transformer is one of the most vital equipment in an electrical system and its failure results in huge economic losses. Amongst the various data driven techniques available in literature for diagnosing faults in a power transformer, Support Vector Machine (SVM) is one of the most promising. In this context, SVMs have typically been implemented using all the gaseous species available from dissolved gas analysis (DGA). In this work, we propose to enhance the diagnostic performance of SVMs by using them with an optimally identified subset of gaseous species available from DGA. We propose to use mutual information to identify these optimal species (features). The approach is applied on industrial datasets corresponding to various commonly encountered faults in power transformers. The results show that better diagnostic performance is obtained when CO2 concentration measurement is not used.

Keywords: Process and Performance Monitoring, Fault Detection, Supervision and Safety
Abstract: The ability to monitor large numbers of variables and the flexibility to assign alarms to each variable led to a substantial increase in the numbers of alarms in industrial plants. This, in turn, increased the numbers of false and redundant alarms. In plant operations, the numbers of annunciated alarms regularly exceed the acceptable rates that operators can handle. To reduce the number of assigned alarms, a risk-based alarm system has been proposed in the literature (Ahmed et al. 2011) where alarms are assigned to groups of variables instead of individual variables. This articles explores the options for grouping variables for alarm allocation. Several grouping methods are discussed and an event-based grouping procedure is detailed. Selection of the key variables for a group is performed using the information that the variables can have to distinguish between an abnormal and a normal condition. The concept of mutual information is used to quantify the information. Variables with high information gain are grouped together for each respective abnormal event. To identify the redundant variables within the groups to further reduce the number of variables to be monitored, the maximum cross-correlation between pairs of key variables are used. A case study using the example of a continuous stirred tank reactor is used to demonstrate the methodology.

Keywords: Process and Performance Monitoring, Fault Detection, Supervision and Safety, Inferential sensing, State Estimation and Sensor development
Abstract: An optimally designed sensor network is essential to ensure optimal and safe process operation. Several approaches for designing sensor networks for efficient fault diagnosis have been presented in the literature. Most of these utilize signed digraph (SDG) based process representation and assume that the SDG is accurately known. For a nonlinear system, the signs on the edges in a SDG depend on the operating conditions and can thus change with time or operating mode and hence may not be accurately known. However, such uncertainties in SDG modeling have been largely ignored in sensor network design literature. In this work, we propose a design approach that considers such uncertainties while selecting optimal sensor networks. The resulting network is optimal in the sense that it results in the lowest number (or lowest cost) of sensors that can ensure observability and resolution of faults while simultaneously identifying the signs on the uncertain edges. Similar to faults, the concepts of observability and resolution are defined for such uncertain edges as well and used in the design procedure. The utility of the proposed approach is demonstrated by applying it on a five tank system.

Keywords: Process and Performance Monitoring, Fault Detection, Supervision and Safety
Abstract: For process fault detection and diagnosis, a real time hybrid method based on Principle component analysis (PCA) and Bayesian belief network (BBN) is described. Upon successful identification of fault from PCA residual plot and Q statistics, information from the PCA contribution of each variable is passed to the BBN for root cause analysis. Pearl`s message passing algorithm is used for belief updating. Early detection of fault, makes the methodology more reliable and robust during the process fault occurrence. The aim of this monitoring tool is to incorporate prior process knowledge along with the present observed evidence to come up with the most plausible explanation of how the process is behaving. The effectiveness of the proposed method is demonstrated for a Dissolution tank model for different simulated scenarios by detecting and diagnosing the fault accurately.

Keywords: Process Optimization, Control, Control Applications, Process and Performance Monitoring, Fault Detection, Supervision and Safety
Abstract: For a long time, companies in the Chemical Process Industry have turned to Advanced Process Control (APC) as a means to increase profitability and production capacity. Among the most widely used APC techniques is Model Predictive Control (MPC). Many service and software vendors have become knowledgeable in executing projects and delivering MPC applications throughout a diverse range of processes. The focus of this work is on what happens after project conclusion, when the project team is dismissed and the application is handed over to the customer operations. It has been observed in practice that without proper maintenance, APC applications will tend to progressively lose their ability to capture benefits. This work will present how we developed a program aimed at sustaining the value of our APC applications. The program reflects our experience in supporting APC applications from multiple vendors installed in 11 different polymerization lines in seven different sites. It will be presented which steps were taken in building such program starting with the cause-effect analysis, where we identify the causes for application malfunction, going through the identification of stakeholders, key performance indicators and assignment of roles and responsibilities. Although the outlined program was built based on the characteristics of the plants and APC applications found in the polymers business, it should be equally applicable to other industries. It is also interesting that the program does not depend on the choice of APC vendor.

Keywords: Process and Performance Monitoring, Fault Detection, Supervision and Safety, Interaction Between Design and Control
Abstract: To ensure safe operation of technical processes, faults have to be reliably detected and isolated to provide information for process maintenance, shutdown, or reconfiguration. Fault detection and isolation can be achieved by invalidation of fault candidates, i.e. models of the system in fault-free and faulty condition. In order to enhance the performance of fault detection and isolation, so-called active approaches use input signals with the objective that the resulting system outputs are consistent with at most one fault candidate. Guaranteeing or analyzing robustness of active fault diagnosis with respect to input, output, and process uncertainties and nonlinearities is challenging. This paper provides certificates of robustness of input sequences with respect to the aforementioned uncertainties and nonlinearities. The certificates enable the determination of input and output uncertainties for which unique fault diagnosis results can still be guaranteed. In addition, a method is presented to select a minimal number of outputs that still guarantee robust fault diagnosis, thus reducing the measurement setup and cost. The approach employs nonlinear mixed-integer feasibility problems and a relaxation framework and does not require the explicit computation of reachable sets. The approach is applicable to polynomial discrete-time systems and is demonstrated for a numerical example.