METHODOLOGY – DIGITAL TWINS, ELECTRICAL ENERGY MANAGEMENT AND SUPERVISION
Energy production systems are complex systems due to the interaction of multiple phenomena (thermal, mechanical, electrochemical, electrical, etc.). The design, sizing, and control of such systems naturally involve a modeling phase. The multiphysics nature of such systems requires a unified and graphical approach for both dynamic model analysis and supervision in terms of control, electrical energy management, and operational safety.
On one hand, L2EP and CRIStAL are developing complementary methodological and software tools (Bond Graph and REM) for modeling and control of complex multiphysics systems. On the other hand, a digital replica of these models will be created to build a digital twin, integrating artificial intelligence algorithms and machine learning for predictive maintenance purposes. These tools have been validated through numerous publications and their involvement in several projects (ANR Propice for fuel cell prognosis, European projects CHEM, E2C, Intrade for energy systems and clean intelligent vehicles supervision, European project PANDA for vehicle simulation...).
Objectives and scientific challenges
This methodological framework, which cuts across the four pillars of the COMASYS project, is based on the development of multi-scale modeling tools on one hand, and energy management and process control methods on the other hand.
The recognition of the disparity among expertise fields led us to adopt a shared approach aimed at modeling physical phenomena. This approach is undertaken at varying scales, highlighting the concept of a digital twin based on simulation. It involves the modeling of complex systems, taking into account the behaviors of their constituents with a selected level of granularity.
Given the multitude of scales and temporalities of the phenomena, model reduction methods are employed. In the domain of electrical energy converters, the L2EP develops such approaches using low-frequency finite element numerical models that capture material non-linearities. Model reduction methods rely on modal decomposition principles, enabling computation time reduction for simulating specific system components. Additionally, within the field of process engineering, the CRIStAL Research Center possesses expertise in modeling and simulating thermo-fluidic phenomena to optimize and supervise systems in terms of control and operational safety.
If establishing models and selecting their discretization scale are crucial steps in transitioning to simulation, the representation of these models (naturally multiphysics) is equally important. A common multidisciplinary approach ensures a comprehensive understanding of phenomena, particularly energy transfers and couplings, whether at the component or system level. The CRIStAL and L2EP laboratories are developing complementary causal representation approaches. Bond graphs provide a structural representation that highlights energy transfers, losses, and storage in a system. They are particularly suitable for the integrated design of production systems, encompassing modeling, simulation, and supervision using dedicated software tools. Macroscopic energy representation, on the other hand, is a functional representation that relies on integral natural causality, making it well-suited for structuring system control, including energy management.
Task description
Task 1 – Analysis of models used in each pillar (domain of expertise)
Task 2 – Proposal of reduced models
Task 3 – Bond graph and REM for Proof of Concepts (from material to system)
Task 4 – System simulation of Proof of Concepts
Task 5 – Analysis and optimization of system components
Task 6 – Integration for supervision system design
Impacts
There are numerous anticipated positive impacts:
- Provision and interconnection of simulation models tailored to the studied scales.
- Provision of dynamic models to industrial stakeholders for optimal management of multi-source platforms and technical feasibility studies.
- Implementation of online algorithms for health management and prognosis (predictive maintenance) of green hydrogen installations.
- Conducting feasibility tests and techno-economic calculations prior to the implementation of "green" electricity production and storage facilities.
- Ensuring operational availability and sustainability of equipment through early fault detection and estimation of their lifespans.
- Optimal management of operational modes using hybrid methods of artificial intelligence and multiphysics models.
Teachings on modeling methodologies, model reduction, energy management, and supervision exist within applied mathematics or systems engineering programs. In a dedicated training program, the main focus will be to tailor the content of theoretical teachings to the specific case of energy systems.
References
A. Bouscayrol, J. P. Hautier, B. Lemaire-Semail, "Graphic Formalisms for the Control of Multi-Physical Energetic Systems", dans Systemic Design Methodologies for Electrical Energy, tome 1, Analyse, Synthèse et Gestion, Chapitre 3, éditions ISTE Willey, octobre 2012.
Silva Luis, Bouscayrol Alain, De Angelo Christian, Lemaire-Semail Betty, "Coupling Bond Graph and Energetic Macroscopic Representation for Electric Vehicle Simulation", dans Mechatronics Elsevier, Vol. 24, N°. 7, pages. 906-913, octobre 2014.
Castaings Ali, Lhomme Walter, Trigui Rochdi, Bouscayrol Alain, "Comparison of Energy Management Strategies of a Battery/Supercapacitors System for Electric Vehicle under Real-Time Constraints", dans Applied Energy, Vol. 163, pages. 190-200, février 2016.
PAM Abdoulaye, Bouscayrol Alain, Fiani Philippe, Faval Fabien, "Comparison of Different Models for Energy Management Strategy Design of a Parallel Hybrid Electric Vehicle: Impact of the Rotating Masses", dans IET Electrical Systems in Transportation, décembre 2020, résumé.
Montier Laurent, Henneron Thomas, Clenet Stéphane, Goursaud Benjamin, "Model Order Reduction Applied to a Linear Finite Element Model of a Squirrel Cage Induction Machine Based on POD Approach", dans IEEE Transactions on Magnetics, mars 2021.
