[Jstempert]

Hi, I'm a French student and i have to use STAR-CCM+ to predict the aerodynamic noise produced by the nose landing gear of the Gulfstream G550.

Currently,, i'm using a simple precision of STAR-CCM+.
I wonder what are the difference between simple and double precision and what are the consequences on the predictions of the noise, the pressure and the velocity magnitude ?
Is the time simulated will be the same for the same results ?
Same question for the time-step requiered ?

Thank you.

Jerome

[phys-zephyr]

Dear Jerome,

I think that you mean SINGLE rather than SIMPLE precision.
Single precision floating point format is a computer number format that stores information in 4bytes (32 bits) in contrast to double precision that uses 8 bytes (64 bits). There are significant differences concerning the precision of the teo formats.
As single precision uses only 32bits (1bit for the sign, 8 bits for the exponent and 24 (only 23 explicitly stored) for the significand precision), the decimal digit precision obly ranges from 6-9 digits, whereas in double precision the 64 bits (1 sign, 11 exponent, 53 (52) significand precision) yield 15-17 precise digits. As you are using single precision, you can only trust 6-9 of all digits and should consider this for your results...

Hope that helped,
Cheers

[me3840]

While phys-zephyr is correct on what single and double precision means from a mathematical standpoint, it's important to note that STAR-CCM+ does not have a single-precision version.

STAR-CCM+ comes in a mixed-precision and a double-precision version. In the latter, all variables and functions are computed with 64-bit floating point numbers. In the former, all variables and functions except for position/distance (x,y,z) and pressure are computed with 32-bit floating point numbers. The four functions mentioned are computed in 64-bit, regardless of your precision choice.

If you are doing a direct noise simulation (not using a broadband model), you should use the double-precision version.

The double-precision version will be more computationally expensive, yes.

Your timestep is set based upon your physics and your highest desired frequency; the precision of the build has no bearing on this value.

[Jstempert]

Thank you for your both answers.

In fact i'm using a DES model coupled with FW-H method.

Have a good day