Automation and supervision of intensified processes – ASPI

In the first part, existing safety criteria were evaluated and limitations for the detection of faults and the evaluation of situations at risk of thermal runaway were highlighted. A new criterion was then developed, based on a dynamic threshold of the heat released by the reaction. Monitoring this heat makes it possible to detect faults when it exceeds a set threshold. The collaboration between process and automation specialists has led to the application of a method that uses a high-gain status observer to estimate, in real time, the heat released from simple temperature measurements. The performance of this new approach was compared to a detection method based on the dynamic temperature threshold.
Another aspect of safer operation is to implement fast and exothermic reactions in heat exchanger reactors (HEX). The control of this equipment is fundamental with a view to intensification to optimize production and ensure safety. Control systems based on fault-tolerant control are widely used for the regulation of such nonlinear systems. An approach for isolation and fault identification (FDI) has been developed from adaptive observers based on simplified HEX model. These techniques developed by the automation partners were first validated in simulation and then tested in real time on an experimental pilot set up in a process laboratory.

A new method for on-line fault detection based on a dynamic threshold of the heat released by the reaction, evaluated in real time using a status monitor, has been developed and tested on two different reaction systems. This method does not require knowledge of kinetic and enthalpy parameters and is mainly based on temperature measurement and information about the reactor geometry and cooling system. It can thus be used in real time when operating chemical reactors.

From a supervisory perspective, detection, isolation and fault identification (FDI) algorithms were developed by integrating nonlinear adaptive observers using a nonlinear reactor model. The algorithm developed, whose effectiveness was demonstrated in simulation, was implemented on a control system of an experimental pilot unit built around a HEX reactor.

The effectiveness of the detection criteria could be improved by integrating observers with filtering capabilities to reduce noise when estimating the heat released and reduce detection time.
These criteria, which have been shown to be effective for traditional batch or semi-batch reactors, could be transposed to HEX continuous reactors.
The detection of faults combined with diagnostic methods will make it possible to identify the origin of the drift. As was done in simulation, the FDI results can be used in real-time on the pilot for the development of a fault-tolerant control system.
Continuing the experiments on the pilot could lead to the application of the algorithms developed to industrial reactors with the aim of reducing accidents related to thermal runaway, the consequences of which are very often disastrous.

Farza, M.; et al. Improved high gain observer design for a class of disturbed nonlinear systems. Nonlinear Dyn. 2021, 106 (1), 631–655.

Han, X.; et al. Dynamic and sensor fault tolerant control for an intensified heat-exchanger/reactor. Eur. J. Control. 2023, 69 (1), 100736.

Kouhili, Y.; et al. Performance of Runaway Detection in a Batch Reactor using Thermal Runaway Criteria. Chem. Eng. Trans. 2022, 91, 553–558.

He, M.; et al. The Simplified Modeling and Experimental Verification of a Heat Exchanger/Reactor. In Proceedings of the 11th International Conference on Sustainable Chemistry. 2023, 677–682.

The ASPI project is mainly motivated by the development of an added value system engineering concerning the control problem of intensified chemical processes. These issues range from the control system design to its supervision with a particular emphasis on those fundamental functions regarding the diagnosis, the prognosis as well as predictive maintenance. Supervision will be elaborated from the best available results on fault diagnosis and identification, observer’s synthesis, system identification, faults tolerant control and efficient procedures for signal processing. These methodologies will allow man to be unloaded from a part of the process monitoring while increasing the operational safety. For this, it will be necessary to anticipate and correct any drifts or dysfunctions that could lead to accidental situations.
These techniques will be applied to an industrial field that is particularly critical from the point of view of safety and the catastrophic consequences that accidents can cause: chemistry and more specifically fine chemistry, pharmaceutical chemistry and new syntheses for the valorization of biomass. Another aspect of the project concerns the transformation of production processes through the implementation of innovative processes in the field of process intensification that prefigure the chemical plant of the future. In the production line, the reactor occupies a central place because chemical reactions occur in it. Reactions are highly non-linear operations with respect to the different operating conditions and whose control is crucial in relation to productivity and safety. In the project, we will focus more specifically on new types of multifunctional, continuous reactors that are an alternative to traditional 'batch' reactors. These intensified reactors radically improve the transport and transfer properties (thermal and mass) and allow implementing reactions by approaching the intrinsic limits of their kinetics.
This interdisciplinary fundamental and methodological research work will be performed in collaboration between laboratories of Chemiacl Engineering and Automatic: LGC, LAAS, LAC and LSPC. Fundamentally, this involves the development of experimental modeling, observer synthesis, identification and error detection and fault-tolerant control approaches for intensified reactors. These approaches will be particularly used to develop a control system with a predictive ability to anticipate accidental situations.
From chemical engineering point of view, the control system must guarantee the operational safety of the intensified processes. Particular attention will be paid to the development of a realistic simulator of static and dynamic behaviors and the creation of a fault database for the supervision of the control system and its reconfiguration if necessary. This requires good modeling of reactors in degraded mode with management of model changes in case of fault detection.
Experimental validations will be carried out on pilots available in laboratories to highlight the added value of the developed approaches. Two situations will be addressed: an implementation in a case already studied both from the point of view of the apparatus and the chemical system and the generalization to a system (reactor and chemical synthesis) for which the characterization is still incomplete.