J.A. Romagnoli

Model predictive control based on Wiener models

Wiener models, consisting of a linear dynamic element followed in series by a static nonlinear element, are considered to be ideal for representing a wide range of nonlinear process behavior. They are relatively simple models requiring little more effort in development than a standard linear model, yet offer superior characterization of systems with highly nonlinear gains. Wiener models may be incorporated into model predictive control (MPC) schemes in a unique way which effectively removes the nonlinearity from the control problem, preserving many of the favorable properties of linear MPC. This paper examines various model structures including ARX and step-response models with polynomial or spline nonlinearities and their corresponding identification strategies. These techniques are then applied to an experimental pH neutralization process where the performance of Wiener MPC is compared with that of the linear MPC and the benchmark PID control to showcase the salient features of this new approach.

Operation of semi-batch emulsion polymerisation reactors: Modelling, validation and effect of operating conditions

A detailed dynamic model was developed for a styrene emulsion polymerisation semi-batch reactor to predict the evolution of the product particle size distribution (PSD) and molecular weight distribution (MWD) over the entire range of monomer conversion. A system exhibiting zero-one kinetics was employed, with the model comprising a set of rigorously developed population balance equations to predict monomer conversion, PSD and MWD. The modelling equations included diffusion-controlled kinetics at high monomer conversion where the transition from the zero-one regime to a pseudo-bulk regime occurs. The model predictions were found to be in good agreement with experimental results. Both particle growth and the PSD were found to be strongly affected by the monomer feedrate. Reactor temperature had a major influence on the MWD which was, however, insensitive to changes in the monomer feedrate. These findings were confirmed experimentally. As a result, it seems reasonable to propose that the use of the monomer feedrate to control the PSD and the reactor temperature to control the MWD are appropriate in practical situations. Consequently, an optimal monomer feed trajectory was developed off-line (using the validated reactor simulation) and verified experimentally by producing a polymer with specific PSD characteristics.

Economic impact of disturbances and uncertain parameters in chemical processes—A dynamic back-off analysis

Abstract Optimal operating conditions in chemical plants are characterized by operation on a number of active constraints. The presence of disturbances and model uncertainties can easily cause constraint violations. Thus, it is necessary to move the operating point away from the active constraints into the feasible region (back-off). Recently a strategy for detemining the necessary open-loop steady-state back-off from the nominal optimum has been developed by the authors, which consists of solving a joint steady-state optimization-flexibility problem. This paper extends these ideas to consider dynamic situations, thus leading to a joint dynamic optimization-flexibility problem. Having determined the economic penalty associated with open-loop back-off in dynamic systems, the next step is to estimate the potential recovery of this penalty that various control schemes might provide. The proposed approach also includes the optimization of the controller parameters. Results are given for a simple flowsheet example.

A strategy for simultaneous dynamic data reconciliation and outlier detection

The presence of outliers corrupts the procedure of dynamic data reconciliation. In this note, a cluster analysis technique is suggested as a way for discriminating outliers and normal observation data. Furthermore, the formulation of the dynamic data reconciliation problem is modified to incorporate the outlier information. In this way, dynamic reconciliation can be carried out simultaneously with outlier detection. The performance of the proposed approach is demonstrated by simulations on a chemical engineering example from literature.

Detection and classification of abnormal process situations using multidimensional wavelet domain hidden Markov trees

Abstract This paper addresses the detection of abnormal process situations during plant operation via an effective trending strategy. Wavelet-domain hidden Markov models (HMMs) are exploited as a powerful tool for statistical modeling and processing of wavelet coefficients. We focus on the multivariate problem as many variables contribute to the decision regarding process status. A simulation study illustrates the salient features of the proposed framework.

Paper: Variable structure control with a second-order sliding condition: Application to a steam generator

A method for the design of second-order sliding mode controllers is developed for a class of nonlinear systems. The key idea is the necessity of nullifying both the auxiliary output as well as its time derivative when the system is sliding. It is found that if the auxiliary output has a strong vector relative degree one, it is possible to design a continuous control input. This allows the application of the sliding mode control to systems that do not allow discontinuous control signals generated by the classical variable structure control. Robustness of the controllers is analyzed and dynamic bounds for the uncertainty are found for a class of nonlinear systems. The design technique could be extended to nth-order sliding controllers, though the complexity of the calculations increases and the robustness bounds become more restrictive. The application of the technique to a drum-type steam generating unit to solve a robust tracking problem is presented.

On optimising control and the effect of disturbances: Calculation of the open-loop backoffs

Abstract An algorithmic approach is presented for determining the necessary open-loop back-offs from a steady-state optimum, so as to ensure that disturbances cause no constraint violation. The approach consists of defming a joint optimisation-flexibility problem that can be solved within an optimisation framework based on a cutting-plane approach. After calculating the open-loop back-off cost over the disturbance range of interest, bounds on the value of closed-loop control may be found by selecting the worst economics for the uncontrolled case.

A framework for on-line full optimising control of chemical processes

Abstract An increasing demand for improved productivity and better quality control has shifted the interest of the research community to nonlinear model-based control, which has a better chance to meet these requirements due to the intrinsic nonlinear nature of chemical and physical processes. Recent progress in modelling, simulation and optimisation environments (MSOEs) and open software architectures (OSAs) have created the conditions to conceive novel paradigms for advanced process control (APC) of large-scale complex process systems. However, large-scale mechanistic models have scarcely been used in control algorithms and, therefore, issues arising from embedding these process models in APC applications have not been addressed satisfactorily. In this manuscript we propose a novel framework for advanced nonlinear model-based control of process systems which aspires to bring the latest advances in model-based technology closer to the Process Industries.

On the use of gap metric for model selection in multilinear model-based control

One way to design the control of a nonlinear system is to use a set of linear models that are close to the nonlinear system. This gives rise to a need to define the concept of closeness. Since systems can be visualized as input-output operators, a natural distance concept would be the induced operator norm. Yet, the norm cannot be generalized as a distance measure. The aim of this paper is to discuss the application of a distance measure between systems, the gap metric, in order to select a reduced set of models that contain nonredundant process information for robust stabilization of feedback systems based on multimodel controller design.

Robustness analysis and control of a pilot-scale distillation column

A pilot-scale distillation column has been connected to an industrial Distributed Control System (ABB MOD 300). The connection of this DCS to a VAX-cluster through Ethernet Gateway facilitates data-collection from the column. This set-up has been used to investigate different control strategies as well as advanced operational techniques. This paper considers the preliminary study of the robustness analysis of this pilot-scale distillation column. Using a combination of plant data and non-linear simulations, both a nominal column model and an associated model uncertainty are identified. This information is used to carry out a complete analysis of robust stability and performance for the case where a (Ziegler-Nichols tuned) multiloop control scheme is employed. To perform robustness analysis of the control, both structured and highly structured uncertainty characterization approaches are used and it is shown that the latter case provides less conservative results.