Problem with sensor point close to wall

[Sui Hang Yan]

I got a problem on using Accelerad to simulate sDA for room which locates ~100m above ground (origin). The working plan shown below is bounded by walls with a window at south. it appears that the sensor points at edge receives a high irradiance value although i have offset the plane about 0.001m from walls. And this high irradiance problem gets eliminated after the workplane is offset 0.5m from wall. 

Also i have tried to put this room to origin. The result looks normal after all. 

So, i suspected that there is precision error on handling sensor point which is close to wall obstruction. any parameter could i tune to resolve the problem?

Thanks in advance. Yan

[Nathaniel Jones]

Hi Yan,

The Accelerad documentation mentions that your relevant geometry should be less than 100 meters from the origin. This is because GPU ray tracing uses single-precision math, which will not accurately differentiate distances as small as the ones you mention when the distance to the origin is large.

However, if you are calculating sDA, you should not have any points that close to your walls. IES LM-83 specifies that the edge of your plane should be a minimum of 12 inches and a maximum of 24 inches offset from walls.

Nathaniel

[Sui Hang Yan]

Hi Nathaniel,

Thanks for your advice!

Apart from sDA, i am also developing several scripts to study irradiation to facade, sun-hours, shadowing, etc.  so i still need to tackle with close wall problem.

Just wonder the simulation result would be affect if I set the model centriod to origin? i.e half of model would be below zero.

Yan

[Nathaniel Jones]

Hi Yan,

Setting the centroid of the model to the origin is one option. Accelerad does not care that part of the model is below the x-y plane, i.e., there is no assumption that the x-y plane is the ground.

Nathaniel

[Sui Hang Yan]

Thanks Nathaniel !