At the beginning I’ve tried to simulate a winter garden, then I saw strange results so I simplified my model and finally I’ve recognized (after a few tests) that no matter how many walls I have in model or how big is the roof the results are the same. Below are pictures with results. For me it’s very strange.

Hi @igmalion,

The way Sefaira works is that it will capture the floor plate and extrude vertical walls along the perimeter of the floor and if will connect to a floor right above it. If it does not see a floor then it will also look for a roof of equal area. (± some square footage) and if it does not see one of relatively equal in area, Sefaira will create one itself. The reason why Sefaira automates this is because and energy model needs to be a “solid” (i.e., an air tight geometry) and many a times when we model we inadvertently leave openings in the buildings where we do not intend to.

In the examples above, since your floor area has not changed, and there are no windows placed in the model, Sefaira will create an exact box model for each iteration and run the analysis for you.

Hope this helps.

Yes, it helps, thanks. Sefaira works really smart to avoid thousand of gaps it is easy to make.

but…

Next I tried to create a pitched roof and I don’t know how to do that. Below are some screens. It looks like even If I create a sloping roof it doesn’t impact on result because always is created by default a flat roof. In next post showed some trials with sloped walls instead roofs but it doesn’t work. So, how to model pitched roof?Yes, it helps, thanks. Sefaira works really smart to avoid thousand of gaps it is easy to make.

but…

Next I tried to create a pitched roof and I don’t know how to do that. Below are some screens. It looks like even If I create a sloping roof it doesn’t impact on result because always is created by default a flat roof. In next post showed some trials with sloped walls instead roofs but it doesn’t work. So, how to model pitched roof?

Hi @igmalion

Thanks for sharing the detailed analysis here. The second thing to keep in mind is that, Sefaira will always convert any pitched roof model to a flat roof model on the basis of the weighted average height of the perimeter walls as the default height. So regardless of the pitch of the roof the energy model will have a flat roof.

I also suspect that the floor area of this particular building being only 100 sf, the exaggerated changes in the height of the perimeter walls is not captured because the change in air volume inside the space is negligible.

Thanks for explanation.

The floor area was 100 sm (square meters) = 1076 sf. In new model (below) is 400 square meters.

But I have still concerns (sorry about that).

For me it looks like Sefaira converts any pitched roof model to a flat roof model NOT on the basis of the WEIGHTED AVERAGE HEIGHT of the perimeter walls as the default height but on the basis of the LOWEST point of the roof as the default height of ALL walls.

Please correct me if it is not true but that is conclusion of my analysis (below).

Hi @igmalion

Thanks for sharing the results from the different iterations. I just looked at the numbers in the image that you share and it looks like the lowest wall height is infact where the flat roof is automated. I was under the impression that it was the weighted average of the height of the walls but I could be wrong. Let me test the case by creating two separate models on my end, download the .idf file to view the geometry and visually inspect how Sefaira converts the geometry into an energy model. Ill let you know once I find out more.

Hi @igmalion

So yeah I tested the different scenarios and you are correct, Sefaira actually takes the lower height of the wall and creates a flat roof on that Z plane.

And the .idf file that gets created can be seen below.

- First scenario where both walls are at 10 ft - Second scenario where one of the walls is at 30’ and the other stays the same.I do not quite know why that is the case but the model interpretation for the energy analysis seems to take the height of the shortest wall and create a roof place at that Z level. The daylight analysis however takes the model as is, since it is running Radiance as opposed to Energyplus for analysis.

Thanks for your answer.

So, it doesn’t sound good that there is no way to create and analyze pitch roof in Sefaira A lot of buildings is constructed in that way.

However I’ve tried to create a model with flat roofs on the different levels (still with one floor at the bottom) and below are results. It seems to be the same problem - Energyplus (or Sefaira) transforms geometry to with the one roof (as the last time - the lowest one). In that way the results, for example for these two model are the same but that’s should not be true.

I have checked it in Energyplus model and there is only one roof indeed.

hi @igmalion

I suspect when the model conversion algorithm was developed the decision was made to use the lowest portion of the wall since for a residential building with pitched roof, the pitch usually encompasses the attic space which is unconditioned. Now if you need to simulate a pitched roof for energy analysis then a workaround would be to have the walls at a height that will ensure that the same volume of air is within the shell of the building.

For the multiple roof scenario that you have shown, since there is only one floor in the building, Sefaira will treat any number of roof combinations as the base condition where it will take the lowest height of the wall and draw a flat roof at the z level.

Hi,

I did some calculations to find a way to create pitched roof and what I found:

- To simulate pitched roof is needed to create a secondary roof plate somewhere inside the attic to let the Sefaira to create a “default” flat roof
- The volume of air of the “default” flat roof should be the same as the volume of air in attic. It takes me to the assumptions that h1/h2 (height of secondary roof plate)/(height of pitched roof) = 0,5
- The area of the “default” envelope area = “default” roof area + “default” walls area, should be the same as area of the attic envelope area. It takes me to the assumptions that h1/h2 = 1/sin(alfa)-1/tg(alfa) where alfa is the angle of the pitched roof. The value of ratio h1/h2 depends from “alfa” and for example for alfa=30 => 0,27, alfa=45 => 0,41, alfa=60 => 0,58. Here is a calculator.
- Now we can calculate the average value between ratio in point (2) and ratio in point (3) and then we have for:

- alfa=30 => 0,38
- alfa=45 => 0,46
- alfa=60 => 0,54

Maybe it is good assumption to always have ratio 0,46.

As you can see below the results od energy simulation changed in various models.

At the end I realize than there is no matter how I define the entities in pitched roof. It is only necessary to define the secondary roof as a “roof” and the other elements can be tagged as ignored oraz as a wall or roof as well. Of course, due to light analysis it shouldn’t be tagged as “ignored”.