Towards an improved hydropower resources modelling in the Horizon 2020 Energy oriented Centre of Excellence in Computing Applications
Since autumn 2015 Prof. H. Vereecken, Institute of Bio- and Geosciences, Agrosphere (IBG-3) from Jülich Research Centre and Prof. C. Clauer, Institute for Applied Geophysics and Geothermal Energy (E.ON Energy Research Center) from RWTH Aachen University, both JARA-ENERGY members, are participants in the Energy Oriented Centre of Excellence for computing applications (EoCoE, [1]). The centre's overall goal is to use the potential of high performance scientific computing (HPSC) infrastructures and technologies for an accelerated transition towards a low carbon and renewable future electricity supply. EoCoE is one of eight funded centres of excellence (CoE) in computing applications in the Horizon 2020 programme of the European Commission and built around the Franco-German hub of the Maison de la Simulation near Paris and the Jülich Research Centre. HPSC and applied mathematics expertise forms the multidisciplinary, transversal basis of the CoE. Five thematic scientific communities, meteorology, materials, water, energy and fusion, closely interact with the HPC experts on HPC technologies, tools and services to improve, modernize and advance scientific computing and simulations codes, leading eventually to faster, more accurate usable information for stakeholders in the energy sector.
The goals within the EoCoE work package “WATER4ENERGY” under the lead of IBG-3's S. Kollet are: (i) The determination of optimal configurations of geothermal heat and power plants in an urban region characterized by different types of city quarters and a complex geological setting (RWTH); (ii) the production of high-resolution monitoring and long-term projections of streamflow, impacted by climate change, for the efficient management of hydropower, at a scale suitable for individual power plants (FZJ and partners). Regional climate change will have an impact on European precipitation regimes with clear emergent trends, for example the Rhine River basin tends towards drier summers and winters and significant shifts in climate indices such as dry spell length or precipitation intensity. Also, the Alpine domain will most likely be facing winters with increased precipitation, with a higher proportion of rainfall, leading to a change in the spring and summer meltwater regimes, crucial for hydropower production from Alpine storage lakes and hydroelectric power stations along rivers.
In the IBG-3 hydropower context, a hydrologic monitoring system of water resources for hydropower over continental Europe is being set up using the massively parallel ParFlow hydrologic model, featuring integrated surface-subsurface flow and land-atmosphere exchange processes. ParFlow is one of the few three-dimensional variably saturated groundwater-surface water flow codes that is currently applied at space and time scales up to continents (Europe, USA) and decades. Currently, monitoring runs are operational, where ParFlow is run as part of the fully coupled Terrestrial Systems Modelling Platform (TerrSysMP) at 12km spatial resolution for a pan-European model domain and 0.5km resolution for NRW, which might inform hydropower systems on the current and near term status of hydrologic states and fluxes in a real-time fashion (see Figure 1).
ParFlow-only spinup simulations using atmospheric (re-)analysis will provide states and fluxes of the terrestrial hydrologic system in hydrodynamic equilibrium and lay the foundation for a data assimilation driven reanalysis that may provide hydrodynamics of major European rivers. A complimentary focus lies currently on the setup of ParFlow in collaboration with the Simulation Laboratory Terrestrial Systems at the Jülich Supercomputing Centre (JSC) as a hyper-resolution hydrological model over the Alpine region to deliver states and fluxes (e.g., streamflow) to the hydropower system impact modelling groups at University of Trento in Italy. This Alpine ParFlow will be driven by (i) convection permitting regional climate model runs for past and future time spans and (ii) an ensemble of regional climate change projections. This allows for a validation of the hydropower generation module and an assessment of the vulnerability of the main hydropower systems (mainly storage reservoirs in the Alps), considering climate change related uncertainties as well as different reservoir management strategies.
These demanding model simulations are supported and only made possible by substantial ParFlow code modernisation and HPC developments in close collaboration with the JSC. So far, ParFlow has undergone extensive performance profiling by HPC experts and an optimum benchmarking and development environment has been implemented. To tackle big data (volume) challenges related to long integration times at high spatial resolutions over large model domains, in-situ processing in combination with optimized parallel I/O is being implemented. In addition, the numerical solver infrastructure shall be updated allowing to use hybrid parallelism on heterogeneous accelerated HPC architectures such as JSC/JURECA.
Figure 1: Example of TerrSysMP monitoring simulation result (2016-03-03 12:00 UTC) as staged on the HPSC TerrSys YouTube Channel [2] for public dissemination. Left: European model domain, 12km resolution, precipitation (boundary forcing data provided by ECMWF); right: NRW model domain, 1km/0.5km resolution, water table depth change (boundary forcing data provided by DWD). Both simulations are initialized at 2016-03-03 00:00 UTC.
Links:
- EoCoE project website: http://www.eocoe.eu/
- YouTube Channel Hpsc Terrsys: https://www.youtube.com/channel/UCGio3ckQwasR5a_kJo1GdOw