Optimising energy for the future
The chosen topic for the second Semaine des mathématiques (Maths Week)
, 18th - 24th March 2013, is "Mathematics for planet Earth". This provides us with the perfect opportunity to promote research projects focussing on this subject.
Interview with Olivier Teytaud, from the Tao project team, which is embarking on the POST project with the aim of creating a software platform enabling investment in the electrical system to be optimised on a continental scale.
What is the aim of the POST project?
Alongside the company Artelys, an SME specialising in optimisation, we have just received funding from the ADEME, which will enable us to work for four years on simulation and optimisation in the field of electricity generation and transmission . Recent technological innovation in the field of high-voltage direct current transmission opens up the possibility, which has previously been extremely limited, of constructing large direct current networks in forthcoming decades. This represents a new opportunity for the development of so-called "Supergrids", linking centres generating electricity from renewable sources with consumer markets. We are examining three fundamental questions: What do we do about generation? What do we do about transmission? and What do we do about storage ? as, when it comes to renewable energy, where you don't have the option of choosing when it is generated, there is an issue of energy storage. Ultimately, it should be possible to devise a tool, which allows us to see clearly how energy is generated and circulates, on the basis of multiple criteria, in order to provide a guide for long-term investment.
How will the project progress?
The simulator will be populated with three kinds of inputs.
- First of all, data , which requires us to understand how each technique works, in some detail, in order to convert it electronically (for processing by computer) and enter it into a model. We will also add information obtained from scientific literature on the subject, on the way in which different generation, transmission and storage components react.
- We will then establish different scenarios , with an economic slant (if there is development requiring consumption or, on the contrary, if there is a crisis where there is no longer any consumption) and a geopolitical slant (a more or less keen desire for energy self-sufficiency in each country), the possibility of ecological penalties (for using petrol or coal, or the choice to discontinue nuclear generation), as well as incorporating technological hypotheses/assumptions. In this respect, we will make assumptions for each generation and storage option.
- Finally, we will incorporate and compare different investment scenarios , for both power stations and the network.
These three main types of information will allow the simulator to determine costs. The advantage of our approach is that we have included an optimiser, in order to allow for adjustment, i.e. closed loop feedback. This adds new information and provides new investment strategies, in order to improve costs (costs that are both economic and ecological).
When it comes to simulation, what are the problems?
The advantage of simulation is that it provides a better understanding of the system , in particular for renewable energies, which are hard to manage within a network. In reality, there will not necessarily be any wind when there is a peak in consumption, or enough sun at the end of the afternoon when everyone is returning home from work! There is therefore a need for something to absorb this variability , and for us to be able to simulate this offsetting/balancing process in sufficient detail. Another example is provided by large-scale wind generation in Northern Europe, which poses the problem of transmission, as energy is generated quite some distance from where it is consumed. We therefore need to take account of energy losses on transmission lines. These two aspects, of matching generation in terms of time and space, need to be simulated . Electricity storage, in particular hydraulic, will be modelled. Pumped energy transfer stations (STEP), for example, allow the water in a dam to be raised in order to generate new electricity at the right time.
Energy is the most significant optimisation problem of this century
And in terms of optimisation, what are challenges?
The aim of optimisation is to achieve the best possible result on the basis of different criteria . Our basic tools are dynamic stochastic optimisation and reinforcement learning. These optimisation techniques are ideal for problems with a time component. However, this kind of optimisation also has a number of objectives, as a result of the vast array of possible scenarios and different criteria.
What attracted you to this project?
Our way of working differentiates us from other teams. For example, other studies have a "copper plate" approach, i.e. they assume that electricity flows perfectly and at no cost, or that daily storage is not a problem. In our model, we want to go further in terms of the sophistication of the model . We have decided to work on this project in the most open manner possible . Therefore, people outside the project are able to respond throughout the entire duration of the project, and we are able to take account of criticism in order to improve the model and limit the risk. Finally, having the right energy is a major asset for both industry and private individuals; we know that the landscape is going to change greatly in the energy sector and, therefore, the decisions taken in forthcoming decades will be extremely important. I have really come to like having an application for my work, and rather than having 10 small projects, I prefer to tackle a major issue that will result in something real. I believe that this will be a source of enormous problems, but if it did not pose a challenge, there would be no point in it .
These articles could interest you:
Olivier Teytaud , Tao project team Research Scientist