Periodic Reporting for period 2 - MUSICA (Multi-Site Coupled Simulation for Advanced collaborative design)
Período documentado: 2022-07-01 hasta 2023-12-31
The objective of MUSICA project is to develop and to setup an open collaborative co-simulation platform that could be used by both academia and industry to simulate separate aircraft components (such as a fan and an airframe) designed by separate entities, in a unique CFD simulation, while ensuring that intellectual property is protected.
This involves code-to-code coupling, communication between two private networks, and co/post processing of the simulation.
The approach chosen in MUSICA relies on the new generation open platform Fidelity (former OMNIS) developed by Cadence Belgium (former NUMECA) for the management of complex projects involving multi-physics simulations.
The application of coupling approach to heterogeneous CFD solvers requires conservative interpolation of the flow quantities at the connection interface between the computation domains, which is a non-trivial task. The complexity of setting up such simulations remains however a delicate enterprise. It relies on the proper definition of the connecting interface between the solvers and on meshes that need to lie at the same geometrical position. Being usually performed in completely different environments, this is seldom the case and the users need to spend a considerable amount of time to fit the meshes at the interface.
From this work, scientists and industrials will be able to:
● couple their solvers to the platform and have direct access to the advanced project management and setup features as well as processing functionalities.
● Instrument their solver with the coupling library, enabling in-situ code-to-code coupling. The co-processing capabilities will be available with little extra effort.
Thanks to these advanced co-processing features the researchers will be able to perform large coupled unsteady simulations and process the results during the simulation, thereby dramatically reducing the storage requirements and communication between solvers.
The second objective of the project will be to extend the platform to the solvers running in different locations across public networks, in response to organisations needing to perform coupled simulations between distributed entities possessing their own tools, expertise and computing infrastructure. Several challenges are associated with this objective and will be tackled within the project:
● Ensuring reliable and secure connections between the remote entities
● Keeping the intellectual proprietary and data confidentiality of the involved organisations, requiring distributed project management and storage.
Through the challenging developments in MUSICA a unique open Fidelity platform will favor the collaboration between heterogeneous organisations possessing their own expertise and tools, thereby bringing CAE simulations to the next level.
This involves code-to-code coupling, communication between two private networks, and co/post processing of the simulation.
The approach chosen in MUSICA relies on the new generation open platform Fidelity (former OMNIS) developed by Cadence Belgium (former NUMECA) for the management of complex projects involving multi-physics simulations.
The application of coupling approach to heterogeneous CFD solvers requires conservative interpolation of the flow quantities at the connection interface between the computation domains, which is a non-trivial task. The complexity of setting up such simulations remains however a delicate enterprise. It relies on the proper definition of the connecting interface between the solvers and on meshes that need to lie at the same geometrical position. Being usually performed in completely different environments, this is seldom the case and the users need to spend a considerable amount of time to fit the meshes at the interface.
From this work, scientists and industrials will be able to:
● couple their solvers to the platform and have direct access to the advanced project management and setup features as well as processing functionalities.
● Instrument their solver with the coupling library, enabling in-situ code-to-code coupling. The co-processing capabilities will be available with little extra effort.
Thanks to these advanced co-processing features the researchers will be able to perform large coupled unsteady simulations and process the results during the simulation, thereby dramatically reducing the storage requirements and communication between solvers.
The second objective of the project will be to extend the platform to the solvers running in different locations across public networks, in response to organisations needing to perform coupled simulations between distributed entities possessing their own tools, expertise and computing infrastructure. Several challenges are associated with this objective and will be tackled within the project:
● Ensuring reliable and secure connections between the remote entities
● Keeping the intellectual proprietary and data confidentiality of the involved organisations, requiring distributed project management and storage.
Through the challenging developments in MUSICA a unique open Fidelity platform will favor the collaboration between heterogeneous organisations possessing their own expertise and tools, thereby bringing CAE simulations to the next level.
By the 30/06/2022 the following results are achieved on the MUSICA project:
● The co-simulation technology was developed on the Fidelity platform. Co-simulation gives the possibility of coupling two separate instances of a given CFD solver (in our case it was OMNIS/Open solver developed by Cadence). The management of co-simulation also requires exchanging and interpolating computational data at the coupling interfaces of the computations. This is based on the Full Non-Matching Boundary (FNMB) technology already existing in Fidelity/DSB solver (former OMNIS/Open). However, this method has been reviewed to make it independent regarding the entire 3D mesh for which a transfer between the coupled computations would be by far too expensive.
The FNMB technology also forms the core of a new interpolation API developed in the Fidelity (former Omnis) platform. This API allows any solver to be plugged in the platform to exchange and interpolate data through FNMB or rotor-stator interface connections, the coupled boundaries among them. Fidelity platform was extended to manage the definition of domain boundary connections between simulations that may lie in different documents of the same project. Each user will be able to work on his own document without sharing confidential information, i.e. only coupling structures are shared. The coupled simulation definition has been made visible and editable by all users interacting on the distributed project. Consistency of the interaction is now managed by the Fidelity platform. Synchronizations, start, stop and co-processing mechanisms were adapted to ensure the consistent behavior between the simulations. The validation of the results was performed. It was shown that the results calculated with the co-simulation approach are the same those provided by equivalent simulations running with one computation only.
● The SU2 solver has been plugged in the Fidelity platform. This allows using the platform to prepare the geometry, generate a mesh and set up the simulation parameters required by SU2. Then the computation can be launched in serial or parallel mode directly from the Fidelity platform. This workflow has been applied for the calculation of the flow around the Onera M6 wing which is also a tutorial of SU2 solver. The results compare well between the SU2 tutorial, the simulation run with SU2 plugged in Fidelity platform and the equivalent computation run with Fidelity/Open.
● The Fidelity (former OMNIS) platform has been extended to compute aeroacoustics probes. In the context of a simulation setup, new features were made available to define the radiating surfaces and the microphones. The Ffowcs Williams and Hawkings solver (FWH) has been integrated into the platform to compute the aeroacoustics responses at the microphones. A specialized GUI has been developed to analyse the FWH results. The geometry of the airframe has been imported and repaired in Fidelity. The external boundary around the airframe was created using the CAD tools of Fidelity and the mesh generation was performed using Fidelity/Hexpress tool (unstructured mesh). The coupling boundaries defined in the simulation are consistent with the powerplant coupling boundaries.
● The co-simulation technology was developed on the Fidelity platform. Co-simulation gives the possibility of coupling two separate instances of a given CFD solver (in our case it was OMNIS/Open solver developed by Cadence). The management of co-simulation also requires exchanging and interpolating computational data at the coupling interfaces of the computations. This is based on the Full Non-Matching Boundary (FNMB) technology already existing in Fidelity/DSB solver (former OMNIS/Open). However, this method has been reviewed to make it independent regarding the entire 3D mesh for which a transfer between the coupled computations would be by far too expensive.
The FNMB technology also forms the core of a new interpolation API developed in the Fidelity (former Omnis) platform. This API allows any solver to be plugged in the platform to exchange and interpolate data through FNMB or rotor-stator interface connections, the coupled boundaries among them. Fidelity platform was extended to manage the definition of domain boundary connections between simulations that may lie in different documents of the same project. Each user will be able to work on his own document without sharing confidential information, i.e. only coupling structures are shared. The coupled simulation definition has been made visible and editable by all users interacting on the distributed project. Consistency of the interaction is now managed by the Fidelity platform. Synchronizations, start, stop and co-processing mechanisms were adapted to ensure the consistent behavior between the simulations. The validation of the results was performed. It was shown that the results calculated with the co-simulation approach are the same those provided by equivalent simulations running with one computation only.
● The SU2 solver has been plugged in the Fidelity platform. This allows using the platform to prepare the geometry, generate a mesh and set up the simulation parameters required by SU2. Then the computation can be launched in serial or parallel mode directly from the Fidelity platform. This workflow has been applied for the calculation of the flow around the Onera M6 wing which is also a tutorial of SU2 solver. The results compare well between the SU2 tutorial, the simulation run with SU2 plugged in Fidelity platform and the equivalent computation run with Fidelity/Open.
● The Fidelity (former OMNIS) platform has been extended to compute aeroacoustics probes. In the context of a simulation setup, new features were made available to define the radiating surfaces and the microphones. The Ffowcs Williams and Hawkings solver (FWH) has been integrated into the platform to compute the aeroacoustics responses at the microphones. A specialized GUI has been developed to analyse the FWH results. The geometry of the airframe has been imported and repaired in Fidelity. The external boundary around the airframe was created using the CAD tools of Fidelity and the mesh generation was performed using Fidelity/Hexpress tool (unstructured mesh). The coupling boundaries defined in the simulation are consistent with the powerplant coupling boundaries.
● The method to compute FNMB and Rotor/Stator connections has been reviewed to make it independent regarding the entire 3D mesh for which a transfer between the coupled computations would be by far too expensive. In the scope of making the method independent from 3D mesh, we reduced the memory needed for the computation by a factor 10. This achievement has a very positive impact for the Fidelity platform in general on the FNMB and Rotor/Stator computation beyond the scope of the MUSICA project and co-simulation. Those connections are used in mono simulations of Fidelity platform for Turbomachine, Automotive and Aerospace industries.
● Thanks to MUSICA project, we are able to run co-simulation between two Fidelity/DBS computations running on two machines in the same network even if they are not geographically close to each other while keeping the confidentiality of data on every side. Only the minimum required data at the interface of 2 domains needs to be shared. It is a prerequisite and opens the door to multiphysics simulations involving different solvers.
● We developed a prototype of plugin allowing to configure a computation for SU2 solver from CAD to simulation settings inside Fidelity platform. The plugin includes the GUI for SU2 solver in Fidelity. Originally, SU2 solver does not have GUI and works with text files and a command line. But it does have GUI now in Fidelity.
● Thanks to MUSICA project, we are able to run co-simulation between two Fidelity/DBS computations running on two machines in the same network even if they are not geographically close to each other while keeping the confidentiality of data on every side. Only the minimum required data at the interface of 2 domains needs to be shared. It is a prerequisite and opens the door to multiphysics simulations involving different solvers.
● We developed a prototype of plugin allowing to configure a computation for SU2 solver from CAD to simulation settings inside Fidelity platform. The plugin includes the GUI for SU2 solver in Fidelity. Originally, SU2 solver does not have GUI and works with text files and a command line. But it does have GUI now in Fidelity.