Energy system modelling is a key tool for EDF R&D. It is used to evaluate the impact of energy policies (renewable deployment, EU ETS), to recommend business strategies and analyse business opportunities based on evolutions in the energy/power systems and to contribute to the public debate.

Multi-energy analysis is gaining increased attention, as more interactions between electricity, heat and cold, and gas systems create promising opportunities for decarbonisation (for instance, by developing more efficient usages or sharing flexibilities for a better integration of variable renewables). Some major challenges need to be addressed: modelling these interactions is complex, and can lead to overly complex models or, on the contrary, to the use of significant simplifications. These simplifications must be made carefully, especially when modelling power systems, as they can easily distort the results and lead to a partial understanding of the system’s complexity.

More specifically, energy system modelling is often linked to one type of modelling, namely models based on TIMES, representing the interactions between several energy vectors. These models make it possible to simulate and optimise decarbonisation scenarios over several decades, which makes them highly interesting. However, most TIMES models do not currently allow for a detailed enough representation of the electricity mix. In particular, they generally cannot give specific insights on the needs for flexibility related to the growing penetration of wind and solar sources (for example, as these models do not use hourly steps or multi-scenario approaches, rules of thumb are needed to decide on the required peaking capacity - an approach that has obvious limitations).

EDF R&D has led a major effort in recent years to study the implications of various energy scenarios on the electric power system. Various approaches were tested and developed in-house. One of these, used in “Technical and economic analysis of the European electricity system with 60% RES”, consists of using a “chain of models” (instead of a single model), with each model in the chain making it possible to study and grasp various key impacts of wind and solar integration in power systems.

The CONTINENTAL model (see Langrené et al)[1] is the main step in the modelling chain described below. The input data and hypotheses (such as the CO₂ price, demand level, etc.) come from energy scenarios, which are sometimes established using large energy systems models (such as the EDF R&D Madone/TIMES model, or the JRC model), which then constitute the first step of the modelling chain. The CONTINENTAL model’s1 outputs can also be fed into other modules/models, constituting the last steps in the chain, as described below. The following paragraphs describe in more depth the core model and the sub-modules developed.

Figure 1: Madone/TIMES model

Source: Burtin and Silva

 

In order to study the impact of wind and solar, the core power system model (CONTINENTAL) needs to feature some minimum characteristics to provide credible insights. These “minimal/recommended requirements” have been well discussed in the research community in recent years, and the most often cited are:

• Hourly base and multi scenarios of demand and variable generation: assessing the need for back up generation implies being able to take into account extreme events that can happen over a few hours, in certain years. The use of average profiles (for instance, only peak and off-peak steps, on one average day per month) cannot capture such events.
• Multi-zone modelling (i.e. modelling exchanges between countries in the European system): national studies often use a “single zone” model, using rough approximations for the level and prices of imports and export. As the European grid gets more integrated and interconnected, the impact of these simplifications should be re-assessed regularly. The CONTINENTAL model makes it possible to feature a European multi-zone market.
• Water reservoirs and pump storage management: Hydro resources and existing pump storage are key providers of flexibility – most models rely on simplified rules of thumb and a deterministic vision to dispatch these energy-constrained resources. Such simplifications might under- or overestimate the role of these resources, while a stochastic modelling of hydro (such as the one used in CONTINENTAL) gives a more realistic view.
• Detailed modelling of thermal unit constraints: dynamic constraints such as minimum on/off time, start-up costs, minimum stable generation, etc. are important when estimating the system flexibility – modelling these constraints generally implies very strong increases in problem complexity and computing time. It might therefore not be possible to always model them, but sensitivities to these parameters should be considered.

Additionally, such models should make it possible to analyse the need for different types of back-up fossil generation (e.g. the share of combined cycle versus open cycle gas turbines) – in the EDF modelling chain. The so called “investment loop” makes it possible to establish a least cost back-up generation fleet, respecting a predefined adequacy criteria (a 3/y hours Loss Of Load Expectation). Figure 2 gives an example of how thermal generation would evolve from a European system with 0% wind and solar to one with 40 %.

Figure 2 : European load duration curve of demand and net demand with 60% RES (left) and structure of the generation mix with and without wind and PV generation (right)

However, as already mentioned, adding too many features at once in one single power dispatch model might not always be a good option: the computing time is likely to increase sharply, but also (and perhaps more importantly) it might limit the possibility to understand all the phenomena at play in a high RES system (when models become overly complex, there is a risk that they become for most people a mysterious black box, instead of a tool that allows a better understanding by engineers and economists).

Therefore, for its 2015 study EDF R&D developed additional sub-modules using output data from the power system model to analyse specific power system issues, such as the need for operational margins compared to the available flexibility (“FlexAssessment” – see Figure 3), or the behaviour of frequency (“Dynamic stability module”), without modifying (and making more complex) the core power system model itself.

Figure 3 : Day-ahead operating in 2013 and the simulated operating margin for the scenario “60% RES” for France during summer

 

The CONTINENTAL model and the various sub-modules make it possible to simulate one year periods. So, it is not possible directly to propose evolution scenarios, for which TIMES models are better suited. It would, however, be possible to extend the “chain of models” and to back-feed information into the TIMES model (such as, for example, a better vision of the required thermal back-up generation). Interaction between models in this way can be a great tool to build decarbonisation scenarios that take into account both multi-energy interactions and the strong specificities of electrical power systems.

The “chain of modelling” discussed here is one example where modelling has provided significant insights into the implications and challenges of a high RES European scenario for the electricity system [EDF15]. EDF R&D, already with a long history of energy modelling, is continuing to work on energy system modelling, in an effort to keep increasing our understanding of the current and future challenges of the electricity and energy systems.