Nicolas Gensollen

Nicolas Gensollen

Research engineer at INRIA Paris

Posts by Collection

conference_publications

Coalition formation algorithm of prosumers in a smart grid environment

Published:

In a smart grid environment, we study the coalition formation of prosumers that aim at entering the energy market. It is paramount for the grid operation that the energy producers are able to sustain the grid demand in terms of stability and a minimum production requirement. We design an algorithm that seeks to form coalitions that will meet both of these requirements: a minimum energy level for the coalitions and a steady production level, which leads to finding uncorrelated sources of energy to form a coalition. We propose an algorithm that uses graph tools such as correlation graphs or clique percolation to form coalitions that meet such complex constraints. We validate the algorithm against a random procedure and show that, it not only performs better in terms of social welfare for the power grid, but also that it is more robust against unforeseen production variations due to changing weather conditions for instance.

Download: PDF ArXiv https://doi.org/10.1109/ICC.2015.7249262

Cite: BibteX

Submodular optimization for control of prosumer networks

Published:

We propose here a control based method for improving the storage placement in a prosumer network where generators and loads are stochastic. The particularity of our approach is to use the energy required for stabilizing the system as a criterion for the optimization of the storage placement. We use a linearized AC model for the dynamic of the system and we consider the control inputs as being the amount of power injected in the grid by the batteries. Because a prosumer may both consume and produce electricity using renewable generators (DER), power imbalances are very likely, such that the system might lose synchrony and require the controlled actions of batteries. We show that the amount of energy that has to be used is this kind of situation depends on the storage locations. We chose the placement that minimizes, over the perturbations that may occur, the average energy required for driving the system back to equilibrium. For this purpose, we propose and validate an algorithm based on a submodular optimization that includes the physical constraints of the system and has a worst case guarantee.

Download: PDF https://doi.org/10.1109/SmartGridComm.2016.7778758

Cite: BibteX

Quasi-Static Time-Series PV Hosting Capacity Methodology and Metrics

Published:

Distributed photovoltaic systems (DPV) can cause adverse grid impacts, including voltage or thermal violations. The installed capacity at which violations first occur and above which would require system upgrades is called the hosting capacity. Current methods for determining hosting capacity tend to be conservative by either only considering infrequent worst-case snapshots in time and/or only capturing coarse time and spatial resolution. Additionally, current hosting capacity methods do not accurately capture the time-dependence making them unable to capture the behavior of voltage regulating equipment and of some advanced controls mitigations. This can trigger delays from unnecessary engineering analysis or deter solar installations in areas that are actually suitable. We propose a quasi-static-time-series (QSTS) based PV hosting capacity methodology to address these issues. With this approach, we conduct power flow analysis over the course of a full year, to capture time-varying parameters and control device actions explicitly. We show that this approach can more fully capture grid impacts of DPV than traditional methods.

Download: PDF https://doi.org/10.1109/ISGT.2019.8791569

Cite: BibteX

international_talks

internships

Robustness of Web of Trust Mechanisms

Published:

Internship opportunity on the robustness of web of trust mechanisms at the Laboratoire d’informatique de Paris 6 (LIP6), supervised by Matthieu Latapy and myself.

Download: PDF

journal_publications

Stability and Performance of Coalitions of Prosumers Through Diversification in the Smart Grid

Published:

In the context of the smart grid, we propose in this paper an algorithm that forms coalitions of agents, called prosumers, that both produce and consume. It is designed to be used by aggregators that aim at selling aggregated surplus of production of the prosumers they control. We rely on real weather data sampled across stations of a given territory in order to simulate realistic production and consumption patterns for each prosumer. This enables us to capture geographical correlations among the agents while preserving the diversity due to different behaviors. As aggregators are bound to the market operator by a contract, they seek to maximize their offer while minimizing their risk. The proposed graph-based algorithm takes the underlying correlation structure of the agents into account and outputs coalitions with both high productivity and low variability. We show that the resulting diversified coalitions are able to generate higher benefits on a constrained energy market, and are more resilient to random failures of the agents.

Download: PDF ArXiv https://doi.org/10.1109/TSG.2016.2572302

Cite: BibteX

Beyond Hosting Capacity: Using Shortest-Path Methods to Minimize Upgrade Cost Pathways

Published:

This paper presents a graph-based forward looking algorithm applied to distribution planning in the context of distributed photovoltaic penetration. We study the target hosting capacity problem where the objective is to find the least-cost sequence of system upgrades to reach a predefined hosting capacity target value. We show that commonly used short-term cost minimization approaches often lead to suboptimal long-term solutions. By comparing our method against such myopic techniques on real distribution systems, we show that our algorithm is able to reduce the overall integration costs by looking at future decisions. Because hosting capacity is hard to compute, this problem requires efficient methods to search the space. We demonstrate that heuristics using domain-specific knowledge can be efficiently used to improve the algorithm performance, such that real distribution systems can be studied.

Download: PDF https://doi.org/10.1109/JPHOTOV.2019.2904540

Cite: BibteX

Do you trade with your friends or become friends with your trading partners? A case study in the G1 cryptocurrency

Published:

We study the interplay between social ties and financial transactions made through a recent cryptocurrency called G˘1. It has the particularity of combining the usual transaction record with a reliable network of identified users. This gives the opportunity to observe exactly who sent money to whom over a social network. This social network is a key piece of this cryptocurrency, which therefore puts much effort in ensuring that nodes correspond to unique, well identified, real living human users, linked together only if they met at least once in real world. Using this data, we study how social ties impact the structure of transactions and conversely. We show that users make transactions almost exclusively with people they are connected with in the social network. Instead, they tend to build social connections with people they will never make transactions with.

Download: PDF ArXiv https://doi.org/10.1007/s41109-020-00266-2

Cite: BibteX

local_talks

Control of Prosumer Networks

Published:

Talk on storage placement in prosumer networks using tools from network control theory, presented at the Smart-grid workshop organized by the SAMOVAR lab in Nano-Innov, Saclay, France.

Optimizing Smart Power Grids

Published:

Talk at the Conservatoire national des arts et métiers (CNAM) presenting my PhD and Postdoc work to the Complex Networks team of the computer science laboratory of Paris 6 (LIP6).

news

Nilearn DevDays 2021

Published:

We organize an international coding sprint from May 5th 2021 to May 7th 2021 around Nilearn and Nibabel, two open source tools written in Python and dedicated to neuroimaging.

open_source

The Power Grid Dataset

Published:

The Power Grid Dataset is a free dataset of real power grid system topologies. It was developed at Telecom SuParis in 2017 with Vincent Gauthier and Lester Padilla.

DiTTo (Distribution Transformation Tool

Published:

DiTTo aims at providing an open source framework to convert various distribution system modeling formats. It is the first, and currently only open source, tool to provide these capabilities.

Santa-Fe Synthetic Network

Published:

Large-scale synthetic distribution and sub-transmission dataset based on building and streetmap data for Santa Fe, New-Mexico, USA. (Not the real network) Produced using RNM-US as part of the NREL-MIT-Comillas-CYME-EDD Smart-DS Arpa-e project.

StreamGraphs.jl

Published:

StreamGraphs.jl is a Julia package to work with stream graphs and link streams. I am currently working on its development at LIP6. Feel free to drop me an email if you are interested in contributing!

Nilearn

Published:

Nilearn is a Python package for fast and easy statistical learning on NeuroImaging data with a focus on fMRI data. It leverages the scikit-learn Python toolbox for multivariate statistics with applications such as predictive modelling, classification, decoding, or connectivity analysis.

Clinica

Published:

Clinica is a software platform for clinical research studies involving patients with neurological and psychiatric diseases and the acquisition of multimodal data (neuroimaging, clinical and cognitive evaluations, genetics…), most often with longitudinal follow-up.

Leaspy

Published:

Leaspy is a software package for the statistical analysis of longitudinal data, particularly medical data that comes in a form of repeated observations of patients at different time-points.

portfolio

preprints

publications

Paper Title Number 2

Published in Journal 1, 2010

This paper is about the number 2. The number 3 is left for future work.

Download: PDF

Paper Title Number 3

Published in Journal 1, 2015

This paper is about the number 3. The number 4 is left for future work.

Download: PDF

teaching

Internet new generation

Master 1st year (M1), 20 hours, Pierre and Marie Curie University, 2014

This course introduces various networking protocols such as TCP, IPv4, or IPv6, as well as cryptographic concepts used for network security.

Programming a video game in C

License 3rd year (L3), 40 hours, Pierre and Marie Curie University, 2014

During this project, students implement a video game of their choice (like snake, pacman, or super mario) using the C programming language.

Artificial intelligence for two player games

License 2nd year (L2), 20 hours, Pierre and Marie Curie University, 2015

During this project, students learn different artificial intelligence concepts (decision trees, alpha/beta prunning, reinforcement learning…) and implement an AI for a two player board game like Awele or Othello. A tournament is organized at the end of every session to see which group managed to build the strongest AI.

Programming a video game in C

License 3rd year (L3), 40 hours, Pierre and Marie Curie University, 2015

During this project, students implement a video game of their choice (like snake, pacman, or super mario) using the C programming language.

Introduction to programming

License 3rd year (L3), 24 hours, Pierre and Marie Curie University, 2016

This course introduces basic programming concepts (variables, functions, loops…) with the Python language.

Introduction to programming in Python

License 1st year (L1), 24 hours, Pierre and Marie Curie University, 2016

This course introduces basic programming concepts (variables, functions, loops…) with the Python language.

thesis

Modeling and optimizing a distributed power network : a complex system approach of the prosumer management in the smart grid

Published:

This thesis is devoted to the study of agents called prosumers because they can, from renewable, both produce and consume electricity. If their production exceeds their own needs, they are looking to sell their surplus on electricity markets. We propose to model these prosumers from meteorological data, which has allowed us to highlight non trivial spatial and temporal correlations. This is of great importance for aggregators that form portfolios of equipments to sell services to the network operator. As an aggregator is bound by a contract with the operator, it can be subject to penalties if it does not fulfill its role. We show that these correlations impact the stability of aggregates, and therefore the risk taken by the aggregators. We propose an algorithm minimizing the risk of the aggregations, while maximizing the expected gain. The placement of storage devices in a network where generators and loads are stochastic and not fixed is complex. We propose to answer this question with control theory. We model the electrical system as a network of coupled oscillators, whose phase angles dynamics is an approximation of the actual dynamics of the system. The goal is to find the subset of nodes in the graph that, during a disturbance of the system, allows returning to equilibrium if the right signals are injected and this with a minimum energy. We propose an algorithm to find a near optimal placement to minimize the average energy control

Download: PDF ArXiv

Cite: BibteX