ff2_logo
ec_logo

General information

This Web application was developed within the use-case (experiment) which is a part of the Fortissimo 2 EU H2020 project. It allows the end-user to configure and submit simulation/optimization jobs to the target infrastructure (HPC). Software is developed by XLAB according to the requirements of Magneti Ljubljana. Jobs that require large amounts of computing power are executed on Arctur's infrastructure.

Services

We offer three services, all of which are executed remotely on the HPC. Two of the services run in a distributed manner on the HPC cluster due to their high demands for computing power.

1. Simulating specific setups

Configure the 3D model of the pressing system, then choose an existing carbide grade material or manually enter its physical properties. After you specify the range of pressures against which you wish to test the model, simulations are run and visualization of results is performed and rendered on-screen.

Note: Although this service runs on the HPC, it is not executed in a distributed manner - it runs on a single core. Calculations are offloaded to the HPC in order to enable running accurate simulations (e.g. with a fine simulation step) while not blocking the host machine.

2. Searching for best materials

Optimization with a Genetic Algorithm (GA). Enter dimensions of your pressing system and configure constraints for physical properties of the material. The algorithm will search the parameter space of possible materials and determine which materials produce the most suitable (e.g. strongest) solutions for your model.

3. Searching for best geometries

The algorithm takes the dimensions of the green part, a material, and constraints for the geometry of the pressing system. It begins testing the yield point of most expensive (e.g. "thickest") solutions and gradually moves towards less and less expensive (e.g. "thinner") solutions. The aim of this algorithm is to reduce the cost of the pressing system, as solutions with least volume are eventually found.


Partners




©XLAB 2017-2018. All rights reserved. Terms of Use | Security Policy | Privacy Policy