Title :

Distributed control of dynamical network systems – Application to a counter-current heat exchangers network

Etudiant :Bertrand ZITTE

Directeur ou Directrice :Françoise COUENNE

Début :10/2017

Fin :10/2020

Financement :Bourse ministérielle

Date de la soutenance :03/09/2021

Commentary :

The current industrial context tends to find the most efficient ways to save energy. In this perspective, the optimization of heat exchanger networks in real time and thus the use of control synthesis become a necessity. In this work, our approach is based on a compartmental representation of heat exchangers in finite dimension.

The main contributions of modeling concern the iterative construction of heat exchanger models based on the cascade interconnection of elementary blocks (an elementary block corresponds to a compartment of the hot fluid which exchanges heat with a compartment of the cold fluid). This approach also allows to represent the interconnection in series of two exchangers as well as the most common topologies in heat exchanger networks: collectors and splitters. These two topologies correspond to interconnections in parallel on the input and output flows of the exchangers respectively. .

For the control part, we propose control laws of the inlet flow of the hot fluid, the other inlet being assumed to be fixed. The state representation of the exchangers as well as of the exchanger network lead to bilinear models.

The contributions related to the control of the exchangers concern:

– Robust stabilization around an operating point of a heat exchanger using the forwarding method on a n-block model. Simulations are proposed on a three-blocks exchanger model using a state observer. The observer uses the measure of the outlet temperature of the two fluids leaving the exchanger.

– Two control syntheses are proposed on a parallel exchanger topology. The control objective is to maximize the energy supplied to the customers. The control syntheses are based on the Extremum-Seeking method with a distributed estimator. The estimator allows to compute locally for each agent (in our case, the exchanger and its control) an estimate of the global criterion from the information available at its level as well as those of its neighbors as defined by the communication graph. The syntheses are based on two different scenarios of the communication graph: the first control law is synthetizing in the case where the communication graph is complete and the second one in the case where it is connected. Simulations are proposed on a network of three exchangers for the first scenario and five exchangers for the second.

**Title:**Distributed control of dynamic systems networks. Application to a cogeneration system.

Name:Bertrand Zitte

Graduate School:EEA

Directors:Françoise Couenne, Boussad Hamroun

Beginning of the thesis:1er octobre 2017

Expected end of the thesis:30 septembre 2020

Funding: Bourse ministérielle