Type: Master assignment
Educational programme: Computer Science, Applied Mathematics
Contact (internal): Johann Hurink and Gerwin Hoogsteen
Company: TotalEnergies
Contact (external): Jules van Dijk
Context
Electric Vehicles (EVs) will play a key role in the energy transition towards a sustainable energy system. With a projection of 10 million EVs in the Netherlands in 2050, the EV fleet forms a significant source of distributed, controllable energy storage and flexibility. Through smart charging policies, EVs can be charged or discharged at various rates in order to, amongst other goals, maximize the utilization of renewable energy sources, reduce CO2 emissions, balance the grid, or prevent grid congestion.
As part of its pivot towards renewable energy, energy giant TotalEnergies operates as a Charge Point Operator (CPO), managing a portfolio of about 8,500 public charge points in the Netherlands, and aims to operate 30,000 public charge points in 2025. To reduce physical impact on the grid and due to economic incentives for participating in energy trading and flexibility markets, TotalEnergies already has developed and deployed smart charging models and optimization algorithms.
Assignment
However, in order to optimally deploy assets for the goals as indicated above, the EV user’s energy demands and departure time are essential input parameters in these optimization models. Moreover, these parameters reflect the EV user satisfaction on charging sessions. However, this information is currently not communicated (e.g. by EV standards). The absence of EV user demand information forms one of the biggest challenges faced by charge point operators in the EV industry at this time.
Your assignment is to develop an (artificial intelligence-based) forecasting model, predicting the EV user demands used to generate the individual optimized customer smart charging profile (e.g. energy demand and departure time) upon connection time. You will determine the required granularity of customer categorization based on input parameters. For the assignment you will be provided with a significant historical database of detailed charge session information one of the largest public EV charge point networks in the Netherlands, on which your models can be tested.
Workload:
- 30% theory
- 30% implementation
- 20% simulating
- 20% writing
For more information, contact Gerwin Hoogsteen or Jules van Dijk:
J.F. van Dijk (Jules)
+31621149588
jules.van-dijk@totalenergies.com