[lucamirtanini]

Hi,
in the user guide it is written that the core algorithm of FDS is second-order accurate in space and time.
I have read the tech_guide pag 33-36 and it seems first order accurate in space. Maybe I misunderstood what accurate means. Do you think I am wrong?

[LuckyTran]

You can't just say "this software" is first/second order accurate. When you talk about accuracy you need to specify what term is being discretized and talk about the accuracy of those terms.

Diffusion terms are central differenced and divergence terms are also central differenced. This would normally be second order in space. But flux limiters are applied which will degrade the order (but preserve accuracy) when the limiter comes into effect.

Also it's an LES-based code, it would be a clown-fest if it was actually 1st order.

[FMDenaro]

Originally Posted by lucamirtanini

Hi,
in the user guide it is written that the core algorithm of FDS is second-order accurate in space and time.
I have read the tech_guide pag 33-36 and it seems first order accurate in space. Maybe I misunderstood what accurate means. Do you think I am wrong?

FDS is one of the CFD softwares I really would not use without having turned on the warning light!


- Apparently, it is second order accurate in space but I am not sure if the integration in time is effectively second order.
- What is more, the SGS modelling in FDS has interesting choices but ... if you read carefully the TG you will discover that a "box filter" is applied. There is no explanation about that, but I don't think is an explicit filtering. On the other hand, the FD formulation works on the quasi-linear form (non conservative) and that makes no sense to think about a box filter implicitly induced by the divergence form. I have no answer for this issue and the use of the filter width in the SGS models.

- Numerically, conservation property is doubtful.

- Last but not least, the use of the vorticity field makes quite different the real application of the SGS modelling.