L'hypercube is proud to have contributed to the Grand Prix National de l'Ingénierie 2021, won by AREP teams, for the transformation project of the Saint-Michel Notre-Dame station (RER C).
Subsequently, we had to deal with other site constraints, such as acoustics and flood control. It is precisely this collective and iterative work of the different engineering departments that was rewarded by the GPNI!
Below is a video presentation of the main challenges of the project:
A "new" database has been made available by the EU, from the Copernicus project (free access, login required).
UTCI and T_{MRT} , August 31st, 2018
This dataset provides a complete historical reconstruction of the UTCI (Universal Thermal Confort Index, a thermal comfort index as well as the MRT (Mean Radiant Temperature, essential for the thermal comfortestimation), from 1979 up to today. Those parameters are computed by the ERA5-HEAT (Humen thErmAl comforT) model. It is based upon a reanalysis of observations from across the world, to provide a globally complete and consistent description of the Earth’s climate and its evolution in recent decades.
Technical data:
DATA DESCRIPTION
Data type
Gridded
Horizontal coverage
Global except for Antarctica (90N-60S, 180W-180E)
Horizontal resolution
0.25° x 0.25°
Vertical resolution
Surface level
Temporal coverage
1979-01-01 to near real time for the most recent version.
Temporal resolution
Hourly data
File format
NetCDF
Conventions
Climate and Forecast (CF) Metadata Convention v1.6
Versions
UTCI v1.0
Update frequency
Intermediate dataset updated daily in near real time, Consolidated dataset monthly updates with 2-3 month delay behind real time.
Thermal comfort and the simulation of natural ventilation, for us that means a lot! Our researchers Edouard WALTHER and Antoine HUBERT presented their latest advances at the IBPSA World #bs2019 conference in partnership with École Normale Supérieure Paris-Saclay.
Since 2017, The Hypercube has been performing detailed airflow simulations and implementing numerical optimization procedures. These methods require significant computing power, which until now has been provided by "virtual machines" in Microsoft's Azure cloud. The team thus had 24 bodybuilt "machines" (8 ultra-clocked logic cores and 56 to 112 GB of RAM), switched on on demand and allowing a high reactivity.
This solution, which has been in use for 2 years, limited us in terms of reactivity (IT maintenance) and connectivity (data transfer speed) with a non-negligible associated cost.
The question then arose, with the AREP IT team, of an opportunity to invest in an "on-premises" calculation server... It is now done! It has just been delivered, in kit form.
Montage
les 4 processeurs physiques
It is therefore a computing rack with 4 processors, 20 cores each capable of hyperthreading (virtualization of a second core), in other words, 160 cores dedicated to simulating building physics phenomena! This represents the equivalent of more than 25 high-performance computers (for example for 3D visualization).
160 cores in action !
Today, this "beast", with 256 GB of RAM memory, extensible up to 3TB, runs on the open source Linux operating system (Debian 9).
Many thanks to AREP IT Services for their advice, support and energy!