Implement processes optimization with AI

The law on energy transition for green growth requires that by 2030, 10% of gas consumption in France should come from biomethane sources. Various projects for injecting biomethane into the network are and will therefore be selected over the next few years to be integrated into the French gas network.

In this context, our client has asked Sia Partners to develop a model for optimising connections, taking into account the constraints linked to the different types of projects and connections, in order to minimise the overall budget. To this end, four types of connections to the system are considered, and the costs and associated technical constraints are incorporated into the model.

As only a part of the projects is currently known, the model includes a module for randomly positioning future projects based on the methanisable deposits and the target volume for 2030. 

The positions of future so-called "fictitious" projects, whose exact position is not yet known, are predicted stochastically based on the methanisable potential of each canton and the characteristics of existing injection projects.

In a second step, the operator sets the target injection volume defining the quantity of "fictitious" projects to be created with their respective injection capacities. The optimization model builds a connection network by considering for each project four types of possible connections at different points of the network with the associated installation costs and technical constraints of the network. 

The model, associated with a solver, calculates the least expensive technically possible solution according to constraints imposed by the operator. A metaheuristics algorithm finally vibrates the solution so as to connect the projects together and avoid any local minimum.

• Integration and processing of our client data in a geographical database ;
• Simulation of the positions of the points to be connected from existing projects and available reservoir data ;
• Formalisation, then implementation and resolution of the mathematical optimisation problem, parameterisable as input by an operator (resolution of the linear integer problem, implementation of Steiner trees and then vibration of the solution using a metaheuristic algorithm);
• Realization of a complementary tool allowing the realization of a Monte-Carlo study on the results with as random parameter the placement of the projects according to the methanizable deposit data by canton in France.
• Development of geographic visualization maps allowing dynamic observation and generation of exports of detailed model results.

Depending on the technical constraints and costs considered, the multiple simulations made possible by the model revealed a distribution of the integration of these new projects on the system between the distributor and the transmission provider. 

These various simulations are used to fuel discussions with the Commission de Régulation de l'Énergie on the management rules and limits of the contracts offered to new producers when they are connected to the grid.

Our teams of experts in Data Science, Optimization and Data Engineer will help you to integrate the tool into your Information System. Once integrated, it turns out to be a complete decision support tool, a facilitator and accelerator in the realization of your studies. Any addition of functionality or new data is tailor-made and according to your needs, so that in the long term, the tool is fully integrated into your structure and independent of the intervention of our teams.