Supercomputing for Industry


Our researchers could offer  partners their knowledge and expertise in numerical modeling and simulation in the areas of computational fluid dynamics (CFD), computational structural dynamics (CSD), and computational biomechanics. The examples of successful projects could be found in our Portfolio.  We could offer expertise and experience in using commercial codes such as ANSYS, MSC, Fluent and the open source codes such as Open FOAM, Code_Saturne, Code_Aster, and Elmer as well. For the all above mentioned codes, we could also offer  technical support for our clients. More details could be found in section Applied Research.

One of the main reasons why companies seek services of HPC center is that those systems allow them to solve their problems with required accuracy and in acceptable time. To solve aerodynamics of a car without supercomputer i.e. using common work station, relatively coarse numerical model with approx. 10 million unknowns have to be used. Using our HPC system, much more detailed model with approx. 1.3 billion of unknowns could be solved.

The second equally important reason is that using the supercomputer could dramatically decrease the computational time needed. As an example, we are presenting a simulation in a hydraulic valve. For this problem, two different numerical models consisting of 14 and 52 millions of cells were created. The computational time for both models using different number of cores is listed in tables below. From those tables a reduction of the computational time could be observed. For 8 cores, this is equivalent of common workstations; 12 hours of computational time were needed to solve this problem. Computational time was reduced to 17 minutes when 1024 cores were used.

Table for first option Table of second option
N cores solution time [s] speedup N cores solution time [s] speedup
8 12124 1 8 42344 1
16 6947 1,75 16 27000 1,57
32 3900 3,11 32 11806 3,59
64 1939 6,25 128 3088 13,71
128 814 14,89 128 3088 13,71
256 497 24,39 256 1630 25,98
512 296 40,96 1024 991 42,73

Another example of the computational time reduction is illustrated by car crash test simulation. In this case, the computational time was reduced to 1 hour from 5 hours needed when a work station was used.