IES-VE User Forum

Hi

I need to CFD model a large auditorium (3,500 seats) to optimize displacement ventilation system. Namely, location and number of supply and extract grilles, air volumes etc.

I am going to try a scaled down version first of 100 seats as practice.

Please can anyone let me know if they have attempted something like this before, if it is possible in the software, and any likely issues to be aware of.

thank you.

Hi,

I asked one of our experienced MicroFlo users about this one and his comments are below.

Q> I am going to try a scaled down version first of 100 seats as practice.

Don’t scale the model, i.e. don’t make a 1:10 scale model of the true geometry for example as you won’t get the correct flow regime predicted. But it is a good idea to slowly build up the complexity of the geometry at the true scale.


Q> Please can anyone let me know if they have attempted something like this before, if it is possible in the software, and any likely issues to be aware of.

Yes this has been done before but it will be difficult to produce a mesh that is fine enough to predict the flow in such a large space without bumping up to the maximum cell limit in Microflo (6-7 million cells with a dual core and 3-3.5 million cells with a quad core).

Keep the geometry as simple as possible to keep the maximum aspect ratio as low as possible. Add lots of cells - use the 6-7 million cell limit by running on a dual core.

Thanks for the info.

The proposed strategy is to have a supply grille per seat. In the model i plan to simplify so have 300 supply grilles instead of the actual 3500. With the interanl heat sources, primarily people, do I have to add a heat source component for each occupant into the model? Is there a way of simplifying?

Thanks again.

PCully wrote:Hi,

I asked one of our experienced MicroFlo users about this one and his comments are below.

Q> I am going to try a scaled down version first of 100 seats as practice.

Don’t scale the model, i.e. don’t make a 1:10 scale model of the true geometry for example as you won’t get the correct flow regime predicted. But it is a good idea to slowly build up the complexity of the geometry at the true scale.


Q> Please can anyone let me know if they have attempted something like this before, if it is possible in the software, and any likely issues to be aware of.

Yes this has been done before but it will be difficult to produce a mesh that is fine enough to predict the flow in such a large space without bumping up to the maximum cell limit in Microflo (6-7 million cells with a dual core and 3-3.5 million cells with a quad core).

Keep the geometry as simple as possible to keep the maximum aspect ratio as low as possible. Add lots of cells - use the 6-7 million cell limit by running on a dual core.

What do you mean by dual core and quad core?

This refers to the processor in the machine you are running on. Most modern machines are dual core meaning there are two processors but high powered machines will have a quad core (4 processor units).

MicroFlo is one of the few parts of <VE> that has been developed to utilise multi core processors, most other parts of <VE> such as Apache will only use one core but we are working on developing this so that it can utilise multi core in future to reduce simulation times.

Hope this makes sense, if you have a quick google you will find more information on multi core processing (if it interests you). Check the system properties on your machine to find out you are running (mine is a dual core).

Phil

Hi

I update this topic.

I don't understand why the dual core can generate 6-7 million cell, and quad core just 3-3.5 million cell. Normaly the quad core is better than the dual core. And if you only develope microflo to the dual core, the quad core should generate the same number of cell than the dual core.

I am interested by buy a quad core, but if it is less performent than the dual core, it is not interesting.