Recent enhancements in Convergent Science’s CFD software have enabled PSA Peugeot Citroën to substantially increase productivity
The company’s Converge software now offers fully automatic mesh generation, which eliminates the need for time-consuming manual meshing. The developers claim this can dramatically improve consistency between tests – with engineers on opposite sides of the planet now able to specify exactly the same criteria for the computer-generated mesh, where previously there was a degree of manual artistry involved.
Convergent Science has also integrated an adaptive mesh refinement (AMR) function in its software. Traditional packages typically generate a tetrahedral or polyhedral mesh before the start of the simulation and then stretch or skew it to suit changes during runtime. Converge uses a Cartesian cut-cell mesh, which is regenerated at every time step and coupled to both the flow and chemistry solvers.
As a result, Converge is able to continuously refine its mesh, adding more cells where increased accuracy is required, but reducing the cell count in less critical areas to speed up processing. In the case of an in-cylinder engine model, for instance, the simulation could have two million cells at bottom dead center and 100,000 cells at top dead center. More specifically, areas of interest such as the flow around the valves or the edge of the flame front can be modelled with increased accuracy without slowing down the rest of the simulation. This is claimed to provide unrivalled levels of speed and accuracy for simulations that include moving geometry.
A substantial number of major automotive OEMs now use Converge, including PSA Peugeot Citroën. Meanwhile, on the racing side, engine specialist Ilmor and NASCAR team Richard Childress Racing are among the company’s customers.
With the latest Version 2.3 release, Convergent Science is aiming to streamline the speed and efficiency of the package once again. It contains over 50 enhancements, including a variety of new geometry manipulation tools, new linear solvers, improved dynamic relaxation capabilities (relaxing the mesh in less critical areas) and a new surface chemistry solver capable of modelling molten-solid urea depletion.
PSA Peugeot Citroën has been using past iterations of the package for more than four years and the group’s manager for the modelling of energetic and combustion systems, Clément Dumand, said this self-meshing capability has substantially improved productivity.
“Generating the mesh automatically allows us to do 10 times more simulation cases than we could do previously. We needed something that efficient to evaluate the large number of configurations we want to try.”
Steve O’Connor, chief engineer at Ilmor Racing, says the introduction of Converge has prompted a step-change in the way the company manages its development process: “We’ve always used simulation, but to date it has supported our traditional approach of physically prototyping a part or concept and then trying it on the dyno. Our engineers heard about Converge and wanted to see if it really could be used to refine ideas faster to minimize our prototyping costs.
“At the moment, we are using Converge intensively to further optimize the combustion system of the Chevrolet IndyCar engine. The work is concentrating on the design of the inlet port, combustion chamber and piston crown with the aim of improving both the volumetric and combustion efficiency of the engine. The use of Converge has improved our understanding of the complex mechanisms that occur within the combustion chamber and has guided us along new avenues of development.”
Evaluating new concepts in the world virtual first and then cherry-picking a reduced number to test on the dyno has led to some dramatic improvements. On the IndyCar project alone, Ilmor says it has reduced the development time by 50% around eight weeks and resulted in a 75% reduction in prototype build costs.
February 17, 2016
