What's new in FINE/Open with Openlabs 10.1
HEXPRESS/Hybrid
- The viscous layers extrusion method and the improved projection option (EXACTMESHING keyword) have been sped up, resulting in an overall mesh generation time cut down by 4% to 63% among our database cases.
- Speed up and robustness improvements for distributed MPI version
- Global speed up for the surface to volume mesh generation
- An improved version of the optimizer for OpenFoam quality criteria is now introduced

GPU acceleration (FINE/Open) BETA
- Use Graphical Processing Units (GPU) in FINE/Open to decrease even further your convergence time.
- Combine the CPU-Booster to the new GPU technology to speed up your computations up to 10 times faster.

Fluid-Structure Interaction: flutter analysis with the Non-Linear Harmonic NLH method (FINE/Open) BETA
- FINE/Open now includes an additional module for performing Fluid-Structure Interaction simulations (FSI). The combination of the modal approach and the NonLinear Harmonic (NLH) solver ensures a fast and accurate analysis of flutter or any forced-response phenomena with limited computing resources.

Displacement of stator blades in a compressor stage (Courtesy of TU Darmstadt)

The resolution of Computational Fluid Dynamics (CFD) problemsinvolves three main steps:
- spatial discretization of the flow domain,
- flow computation,
- visualization of the results.
To perform these steps NUMECA has developed three software systems. The first software system, HEXPRESS, is an automated all-hexahedral unstructured grid generation system. HEXPRESS also includes a module HEXPRESS/Hybrid, which allows to mesh complex geometries with an isotropic hybrid mesh including mainly hexahedral cells with tetrahedral, pyramid and prism cells. The second software system, FINE/Open with OpenLabs, is a powerful CFD Flow Integrated Environment dedicated to complex internal and external flows. The third software system, CFView, is a highly interactive Computational Field Visualization system.
These three software systems have been integrated in a unique and user friendly Graphical User Interface (GUI), called FINE/Open with OpenLabs, allowing the solution of complete simulations of 3D internal and external flows from the grid generation to visualization, without any file manipulation, through the concept of a project. Moreover, multi-tasking capabilities are incorporated, allowing the simultaneous treatment of multiple projects.
FINE/Open with OpenLabs fully integrates the concept of multitasking. This means that the user can manage a complete project in the FINE/Open GUI; making the grid using HEXPRESS, running the computation with the FINE/Open solver and visualizing the results with CFView. Furthermore, the user has the possibility to start, stop and control multiple computations.
418M multidomain mesh automatically generated in 4.5 hiyrs with HEXPRESS/Hybrid. Full multiphysics CHT coupling with FINE/Open radiation and heat exchanger macro model.
Numeca International Inc.develops computational fluid dynamics (CFD) software for the simulation, design, and optimization of fluid flow and heat transfer worldwide. It offers automated and customized flow integrated environments (FINE), including FINE/Turbo for the simulation of rotating and turbomachinery flows; FINE/Open for the simulation of complex internal and external applications; and FINE/Marine for marine applications.
Product:FINE/Open With OpenLabs
Version:10.1
Supported Architectures:x64
Website Home Page :www.numeca.com
Language:english
System Requirements:PC / Linux *
Size:

Supported platforms and operating systems Windows
NUMECA software is supported in 64 bits on x86_64 processors only (64 bits processors that are compatible with Intel 8086 set of instructions).
Formally tested on
- Windows 8.1 Professional -
- Windows 10 Professional -
Please also notice that the following limitations apply:
- Computations under Windows (except Windows Server) are limited to one PVM daemon running at a time. As a consequence, access to multiple users on a given machine is not allowed,
- Parasolid uses SSE2 instruction for high performance. If your platform is not supporting the SSE2 instruction, the FINE™ GUI cannot be started.
- AMD EPYC and RYZEN processors are slower on Windows 10 compare to Linux.
Multiprocessors calculations
Multiprocessors calculations are allowed on shared and homogeneous distributed platforms. Non-homogeneous configurations are not allowed.
The following limitations additionally apply:
- Multiprocessors calculations are allowed on distributed platforms in local user configurations provided that the same account is used for all machines,
- Domain administrators are not allowed to run multiprocessors calculations.
- MPICH2 library needs Microsoft .net Framework package before installing NUMECA software otherwise execution of mpiexec.exe will lead to a Windows error message.
Hardware requirements
Next to the standard hardware (monitor, keyboard and mouse), some specific recommendations apply in the use of NUMECA software:
- a mouse with scroll wheel is strongly recommended.
- the monitor should support 24-bit color graphics and have a 1280 x 1024 pixel resolution for adequate visualization.
- an Ethernet card should be installed and properly configured.
- NUMECA Software does not support CXFS (Clustered XFS) and IBRIX disk file systems.
Memory and disk space requirements
The Random Access Memory (RAM) required to run NUMECA software depends upon several factors, in particular including the number of grid points inserted in the grid, the physical models selected (turbulence, adaptation,...), the introduction of non-matching boundary conditions and the selected calculation mode (mixed or double precision).
Similarly, a minimum disk space is required on the hard disk to allow the storage of the project files. The size of these files largely depends upon the number of grid points generated.
More product specific details
HEXPRESS
The use of minimum 1 GB RAM is strongly recommended. As a general guideline, about 0.5 to 0.7 GB RAM are required to run a one million nodes project. The suggested swap space should be equivalent to at least 3 times the RAM installed.
As a general guideline, a minimum of 300 MB is required to store the project files (provided that the geometry is not defined by Parasolid™ or CATIA files) for a one million nodes project.
HEXPRESS/Hybrid
The Random Access Memory (RAM) required to run HEXPRESS/Hybrid depends upon several factors. In particular, the complexity of geometries and the settings of the CONF file. As a general guideline, the Random Access Memory (RAM) required to run HEXPRESS™/Hybrid are:
- for mesh without viscous layers - between 500 MB and 700 MB RAM per million of cells
- for mesh with viscous layers - between 700 MB and 900 MB RAM per million of cells
The use of minimum 4 GB RAM is mandatory, but the installation of 8 GB is recommended. The suggested swap space should be equivalent to at least three times the RAM installed.
About 1 GB is required to store a two million nodes mesh in SPH format.
OpenFINE/Open with OpenLabs
The use of minimum 1 GB RAM is mandatory, but the installation of 2 GB is recommended. As a general guideline, about 1.5 GB RAM is generally required to run a one million nodes project in double precision mode. The suggested swap space should be equivalent to at least 3 times the RAM installed.
Mesh partitioning for parallel computations requires about 1 GB memory per million cells. When using serial partitioner, the user must ensure that the memory on the head node is sufficient, because the partitioning is performed only on the head node in case of serial partitioning.
As a general guideline, a minimum 1.4 GB is required to store the mesh and solution files (one file, no full non-matching connections, solution stored after one iteration) when running a one million nodes project.
CFView
The use of minimum 1 GB RAM is strongly recommended. As a general guideline, about 250 MB RAM are required to visualize a one million nodes project. The suggested swap space should be equivalent to at least 3 times the RAM installed.
No specific disk space is required on the hard disk for CFView except if images or data files are saved or if the meridional average is created in CFView (turbomachinery application).
Graphics requirements
NVidia graphics cards are fully supported, installed together with the latest drivers.
NUMECA software makes use of the advantages of the available Graphics Card. Some trouble may however come up with the default driver OPENGL since it may not be supported by the user host computer and machine display. In order to set the system compatible with a more portable driver, typically MSW driver is used under Windows instead. Note however that the proposed drivers are usually less efficient in terms of graphical speed than the default driver.
Compilation requirements for OpenLabs
Customization of the CFD models using OpenLabs requires the possibility for compilation.
On Windows 64 bit, Microsoft Visual C++ Build Tools 2015 must be installed: make sure to select the component Windows 10 SDK 10.0.10240 during the installation.
Supported platforms and operating systems Linux
NUMECA software is supported in 64 bits on x86_64 processors only (64 bits processors that are compatible with Intel 8086 set of instructions).
Formally tested on
- Fedora Core (a) 28
- (K)Ubuntu (bc) 18.04
- CentOS (d) 7
a. The GNU C and C++ compiler (g++) must be installed. The gcc version should be 4.8.# or 4.9.# (type the command gcc -v in a shell to get your gcc version). The officially tested version is g++ 4.8.5.
b. Debian based operating systems like Ubuntu, even if LSB compliant, are not recommended as a license server as it is known to diverge occasionally from the LSB standard.
c. For HEXPRESS/Hybrid pay attention to the changed NVIDIA driver location. It is necessary to specify its location defining the environment variable LD_LIBRARY_PATH to /usr/lib/nvidia-current/.
d. CentOS 7.5 with vmware is not supported by HEXPRESS/View.
Multiprocessors calculations
Multiprocessors calculations are allowed on shared and homogeneous distributed platforms. Non-homogeneous configurations are not allowed.
Hardware requirements
Next to the standard hardware (monitor, keyboard and mouse), some specific recommendations apply in the use of NUMECA software:
- a mouse with scroll wheel is strongly recommended.
- the monitor should support 24-bit color graphics and have a 1280 x 1024 pixel resolution for adequate visualization.
- an Ethernet card should be installed and properly configured.
Memory and disk space requirements
The Random Access Memory (RAM) required to run NUMECA software depends upon several factors, in particular including the number of grid points inserted in the grid, the physical models selected (turbulence, adaptation,...), the introduction of non-matching boundary conditions and the selected calculation mode (mixed or double precision).
Similarly, a minimum disk space is required on the hard disk to allow the storage of the project files. The size of these files largely depends upon the number of grid points generated.
HEXPRESS
The use of minimum 1 GB RAM is strongly recommended. As a general guideline, about 0.5 to 0.7 GB RAM are required to run a one million nodes project. The suggested swap space should be equivalent to at least 3 times the RAM installed.
As a general guideline, a minimum of 300 MB is required to store the project files (provided that the geometry is not defined by Parasolid™ or CATIA files) for a one million nodes project.
HEXPRESS/Hybrid
The Random Access Memory (RAM) required to run HEXPRESS™/Hybrid depends upon several factors. In particular, the complexity of geometries and the settings of the CONF file. As a general guideline, the Random Access Memory (RAM) required to run HEXPRESS™/Hybrid are:
- for mesh without viscous layers - between 500 MB and 700 MB RAM per million of cells
- for mesh with viscous layers - between 700 MB and 900 MB RAM per million of cells
The use of minimum 4 GB RAM is mandatory, but the installation of 8 GB is recommended. The suggested swap space should be equivalent to at least three times the RAM installed.
About 1 GB is required to store a two million nodes mesh in SPH format.
OpenFINE/Open with OpenLabs
The use of minimum 1 GB RAM is mandatory, but the installation of 2 GB is recommended. As a general guideline, about 1.5 GB RAM is generally required to run a one million nodes project in double precision mode. The suggested swap space should be equivalent to at least 3 times the RAM installed.
Mesh partitioning for parallel computations requires about 1 GB memory per million cells. When using serial partitioner, the user must ensure that the memory on the head node is sufficient, because the partitioning is performed only on the head node in case of serial partitioning.
As a general guideline, a minimum 1.4 GB is required to store the mesh and solution files (one file, no full non-matching connections, solution stored after one iteration) when running a one million nodes project.
CFView
The use of minimum 1 GB RAM is strongly recommended. As a general guideline, about 250 MB RAM are required to visualize a one million nodes project. The suggested swap space should be equivalent to at least 3 times the RAM installed.
No specific disk space is required on the hard disk for CFView except if images or data files are saved or if the meridional average is created in CFView™ (turbomachinery application).
Graphics requirements
NUMECA software makes use of the advantages of the available Graphics Card. By default, the driver OPENGL is used.
In case a problem appears with the default driver OPENGL, the user should change to the X11 driver under Linux. Note however that the X11 driver is usually less efficient in terms of graphical speed than the default driver.
The drivers OPENGL and OPENGL2 may not work when using VNC or remote access by "ssh". In case of problems, the X11 driver should be used.
Compilation requirements for OpenLabs
Customization of the CFD models using OpenLabs requires the possibility for compilation:
On Linux 64 bit, the GNU C and C++ compiler (g++) must be installed. The gcc version should be 4.8.# or 4.9.# (type the command gcc -v in a shell to get your gcc version). The officially tested version is g++ 4.8.5.