Thesis M/F "DynConGrid and GridForge: From Co-Design to Adaptive Operation for Resilient Power Grids"

Job Description

Thesis Subject

Introduction
The transition toward renewable energy profoundly changes how power systems must be operated and designed. Increasing variability from solar and wind generation and more dynamic consumption patterns require the grid to become both more flexible and more intelligent.
This project combines two complementary research directions—DynConGrid and GridForge—to produce an integrated framework for co-design and adaptive operation of electrical networks. DynConGrid develops real-time Model Predictive Control (MPC) for congestion management by reconfiguring topology, curtailing generation, and ispatching storage. GridForge designs the network itself so that it supports a catalog of feasible, stability-certified topologies. During design, surrogate/proxy models approximate MPC outcomes, enabling the evaluation of how operational controllers behave under candidate topologies and thereby optimizing total cost (CAPEX + OPEX). The two topics are tightly coupled: DynConGrid supplies behavioral data and models; GridForge uses them to certify and select topologies that guarantee safe, efficient, distributed operation.

Objectives and Methodology
DynConGrid (Adaptive Operation)
• Formulate MPC combining continuous controls (curtailment, storage) and discrete topology switching.
• Develop scalable solvers and heuristics (relaxations, decomposition, learning-assisted policies) for near-real-time computation.
• Incorporate uncertainty from renewable generation and demand using robust and stochastic MPC; minimize expected operational cost (OPEX).
GridForge (Co-Design and Flexibility)
• Generate candidate network topologies and switching patterns to define a broad design space.
• Simulate DynConGrid across representative scenarios to obtain operational-cost and performance data.
• Develop surrogate/proxy models that approximate MPC outcomes (control actions, costs, constraint margins), enabling exploration of many topologies efficiently during design.
• Formulate a co-design optimization minimizing total cost = CAPEX (assets) + expected OPEX (MPC-driven operation), under stability and locality constraints.
• Apply control-theoretic certificates (Lyapunov, passivity, input–output) to guarantee fast-dynamics stability for all topologies retained in the feasible catalog.
The surrogates are not merely computational shortcuts—they are essential to include the effects of MPCbased operation during design, ensuring that the selected grid flexibility reflects realistic control behavior and total cost. Their definition will require strong collaboration between DynConGrid and GridForge.

Expected Results and Perspectives
• A mixed-integer MPC operational framework allowing topology reconfiguration for congestion management (DynConGrid).
• Surrogate-based co-design tools capturing MPC-driven operation, enabling scalable exploration of flexible topologies (GridForge).
• A certified set of feasible, stability-guaranteed topologies that balance CAPEX and expected OPEX.
• Methodological advances in surrogate-assisted optimization, distributed MPC, and stability certification.
This research bridges grid planning and real-time operation, offering operators a pathway to resilient, adaptive, and cost-efficient renewable-rich grids.

Your Work Environment

This work is funded by the CNRS-University of Melbourne Joint Ph.D. Program. Two Ph.D. students will participate in the joint research project: one will study in Australia and the other in France. Each student will spend two years at their initial institution and one year at the other institution.

Scientific Context and Originality
Traditionally, congestion management relied on preventive planning, redispatch, and curtailment under fixed network topologies. Over the past decade, Sorin Olaru and collaborators developed MPC-based congestion management and distributed control tools that account for storage and curtailment [1, 2, 3].
These works motivate extending MPC with discrete topology choices (switching) to exploit structural flexibility.
Ye Wang's recent contributions are directly relevant: real-time distributed MPC with limited communication rates and optimization-based network partitioning point to scalable, communication-aware control designs; stochastic MPC and co-design studies demonstrate methods for handling uncertainty and jointly optimizing assets and controls [4, 5, 6, 7]. The originality of this combined project lies in bridging design and operation: (i) extending MPC to include topology reconfiguration in a computationally
tractable way (DynConGrid); and (ii) integrating surrogate models that emulate MPC-driven operational outcomes, allowing GridForge to select topology sets minimizing total cost while guaranteeing fast-dynamics stability and distributed control compatibility.

Compensation and benefits

Compensation

2300 € gross monthly

Annual leave and RTT

44 jours

Remote Working practice and compensation

Pratique et indemnisation du TT

Transport

Prise en charge à 75% du coût et forfait mobilité durable jusqu’à 300€

About the offer

About the CNRS

The CNRS is a major player in fundamental research on a global scale. The CNRS is the only French organization active in all scientific fields. Its unique position as a multi-specialist allows it to bring together different disciplines to address the most important challenges of the contemporary world, in connection with the actors of change.

CNRS

The research professions