Innovation & Quality


FINE™/Turbo 13.1, FINE™/Design3D 13.1 and FINE/Agile™ version 13.1 are now available.

These releases come with documentation included in the software package, available in our customer area on www.numeca.com.
FINE/Agile™, AutoBlade™, AutoGrid5™ and CFView™ are also available as stand-alone.

Highlights

AutoGrid5™: Reduction of the ZR effect default mesh size 

The default mesh size of the ZR effect has been significantly reduced, while keeping an accurate boundary layer, the requested expansion ratio and an optimum cells quality. A new clustering type ZR Effect Boundary Layer is available in the ZR effect edit mode.

AutoGrid5™ 12.2: 42,111 cells

AutoGrid5™ 13.1: 24,819 cells

Reduction of 41% of the number of cells for the same accuracy

 
AutoGrid5™: New gap topology

For turbomachinery blades thicker at the trailing edge than the leading edge, AutoGrid5™ now proposes a new gap topology that allows arbitrary numbers of points on both ends. With this new option, you will be able to accurately capture the flow in the wake, with a reduced number of points upstream the blade.

  

 

FINE™/Turbo: New turbulence quantities for inlet boundary conditions and initial solution

With FINE™/Turbo 13.1, you can now specify the turbulence inlet boundary conditions and initial solution in terms of more convenient quantities such as turbulence intensity, turbulent viscosity ratio, hydraulic diameter and turbulence length scale.

New turbulent quantities for 2-equations turbulence model
 

FINE™/Turbo: Multi-GPU compatibility

Multi-GPU architectures are becoming increasingly common at industrial sites and leadership scale supercomputing centers. These systems achieve a high amount of FLOPS per node through the use of a single high-power, multicore CPU, paired with multiple GPU accelerators.
To answer the market’s need, NUMECA has extended his GPU acceleration capability to multiple GPU accelerators per node.

 

FINE™/Design3D: Design Space Reduction BETA

The new method implemented in FINE™/Design3D, reduces the dimension of the design space based on the Principal Component Analysis (PCA). The reduction of dimensionality has a small impact on the value of the optimization objectives, but the computational cost of the total optimization process can be significantly reduced.
The process is described below:

 

Test case: Rotor 37
  • Structured mesh with 700k cells
  • RANS + Spalart-Allmaras
  • Rotating Hub: 17188 rpm
  • ~3 minutes on 16 procs
Parametric model: 50 parameters
  • Stacking law
  • Thickness law
  • Camber law
Isentropic efficiency:
  • Initial design: 0.845
  • Original design space              (50 parameters): 0.887
  • Reduced design space (22 parameters): 0.886

Using design space reduction before starting the optimization has saved more than 50% of the computational cost!

 

Fixes and improvements

Admin Tool

  • New Diagnostics & License Management functions

AutoGrid5™

  • Blocks conversion to right-handed after generation
  • Python generation commands in real-batch
  • Python command: ZR effect unmapped edges
  • Python command: ZR effect auto blocking parameters
  • Relax partial wake prolongation clustering in bypass configuration

FINE™/Turbo

  • Improve GPU performance
  • Upgrade of the real gas polynomial definition
  • New outputs to the .mf file

FINE™/Design3D

  • Interface and kernel adaptations for FINE/Marine
  • Various Usability Improvements

FINE/Agile™

  • Upgrade of meanline tools to 8.6.7
  • Upgrade of AxCentⓇ to 8.6.7
  • General bug fixes and improvements (contact your support representative to get the full list of new features)

 

Share this article
  • Would you like to know more?
    Click below to get in touch.