Skip to main content

Energy oriented Centre of Excellence for computer applications

Periodic Reporting for period 2 - EoCoE (Energy oriented Centre of Excellence for computer applications)

Reporting period: 2017-04-01 to 2018-09-30

The Energy oriented Centre of Excellence in computing applications (EoCoE, pronounced “Echo”) was created with the goal of tapping the huge potential offered by the rapidly advancing High Performance Computing (HPC) infrastructure to foster and accelerate the European transition to a reliable low-carbon energy ecosystem. EoCoE is organised around a central Franco-German hub coordinating a pan-European network, assembling twenty-one partners from a total of eight countries, all strongly engaged in both the HPC and energy fields.

Thematically EoCoE is composed of four domain pillars, each addressing a specific research community in sustainable energy: Energy Meteorology, as a means to predict variability of solar and wind energy production; Energy Materials, dedicated to photovoltaic cells, batteries and super capacitors for energy storage; Energy Hydrology, as a vector for thermal or kinetic energies associated with geothermal and hydropower respectively; and Fusion for Energy, geared to electricity plants as a long-term sustainable energy source. The thematic pillars are connected via a strong transversal basis specialising in the high-end computing demands of scientific and industrial research, with HPC-related expertise in numerical methods, applied mathematics, and advanced programming methods and tools for Exascale.

The EoCoE project is guided by four high-level objectives:
1. To enable scientific breakthroughs in the Energy domain by redesigning existing codes used for energy-related applications.
2. To develop cutting-edge mathematical, numerical and computational methods and tools.
3. To adapt service activities (outputs of the first two objectives) to public laboratories, industries and SMEs, including training activities for reducing the skill gap
4. To foster HPC and the Energy-oriented scientific and industrial communities’ ecosystem.

As documented in the following sections, considerable progress has already been made on all four of these fronts during the first half of the project.
So far, much effort has been invested in building a community around the topic of renewable energy modelling. These efforts have paid off, as witnessed by the organistion of a session at the European HPC Summit, three face-to-face meetings, three performance evaluation workshops and many other dedicated events. The matrix structure of the project has facilitated strong collaborations between HPC experts of WP1 and application experts of various energy fields (WP2-5). The EoCoE pan-European network is now fully operational and capable of providing interdisciplinary expertise to external clients.

For the transversal basis (WP1), an important achievement was, in collaboration with the POP Centre of Excellence, to implement a rigorous evaluation procedure for application performance. This benchmarking process gives a clear view of the application performance in several key areas (I/O, communications, memory, …) and establishes its precise status when EoCoE experts start to examine it. The impact of each code modification during the optimization process can then be seamlessly monitored and the impact of subsequent support activity quantitatively assessed. It constitutes an important asset to for pushing EoCoE applications towards exascale.

Several important results were obtained in WP2 (Meteorology for energy). For example, the software Solar_Nowcast providing the short-term forecast of global irradiation from ground-based or satellite images was greatly optimized. Optimization efforts reduced the run time by a factor of 10 and enables the application to meet the real time constraints imposed by data acquisition. This work was performed in collaboration with a team from EdF R&D and constitutes a strong step towards using the code in the EdF operational production chain.

Material science (WP3) is also a fundamental issue for many energy applications and often requires substantial computing resources. Therefore, a lot of effort was dedicated to improve models and numerical approaches for the first-principles, atomistic description of materials. Promising results were also obtained for perovskite solar cells, a good example of where energy materials research is essential to promote sustainable alternative to silicon for renewable energy generation by solar power.

Concerning energy hydrology (WP4), the EoCoE teams are involved both in geothermal energy and in hydropower monitoring. In both domains, the numerical production models were again optimized significantly, as for example the I/O of the SHEMAT-suite which was improved by almost two orders of magnitude. These tools combined with the creation (or access) to large and pertinent data-set are enabling tangible progress in the assessment of the influence of climate change on the hydropower production in the Alps or on the geothermal potential in the region around the city of Geilenkirchen.

Controlled nuclear fusion (WP5) is already a very massive user of HPC. Simulations at ITER relevant parameters are not yet achievable but advancing toward this objective is one of EoCoE’s goals. Innovative meshing techniques were developed to properly simulate turbulence and MHD instabilities in tokamaks. Efficient coupling between kinetic and fluid codes were also studied in detail. This coupling is absolutely necessary in order to simulate the full tokamaks including the influence of the vessel wall, and raises important issues, both on the technical and physical sides.
All the work carried out in the framework of EoCoE is directly linked to the ongoing energy transition and as such has strong economic and societal potential in the medium to long term, Even if it is still heavily HPC-oriented. EoCoE has achieved a very important step in defining and implementing a code performance auditing procedure and initiating a series of follow-up actions. This has already had a very strong impact both on very (HPC-) experienced users and have also served to disseminate performance awareness and appetite for more among less experienced ones. A number of codes have been improved by several orders of magnitude, initiating the road to exascale for energy.

The two examples below are illustration on how numerical simulation can effectively contribute to the energy transition:

* Improving efficiency of photovoltaic cells by designing materials at the atomic scale: a new procedure to model silicon hetero-junction (SHJ) solar cell from the atomic-scale material properties to the macroscopic device characteristics has been designed. This approach opens the way to the simulation of very large interfaces fully exploiting the power of HPC infrastructures. Moreover it provides input for mesoscale numerical approaches devoted to the assessment of the charge carrier dynamics affecting the overall efficiency of the photovoltaic device.

* High Resolution River Discharge Modeling for Hydropower Energy Applications: a continental scale high resolution modeling system for the investigation of river discharge in the European region has been developed at 3 km resolution and successfully calibrated and validated against observed and modeled discharges data for a given geographic region and time frame. This new model framework allows to assess simulated time series data using visualization tools and post-processing analysis chains developed in collaboration with EoCoE HPC experts.