buildingSMART Forums

Level1 boundary generation to the centre line or inside face?


I am developing some routines to generate a BEM (spaces, L1 & L2 boundaries) from IFC for various admittance calculations. I am only really concerned with supplying the geometry and various attributes/properties at the moment, which I query from the IFC to perform the calculation, namely space volume, boundary surface area and associated U value for said boundaries.

Currently I generate L1 boundaries on the inside face of the corresponding element, however I have been told that boundaries should go to the centreline and not reside on the inside face of the element (specifically Energy+ always uses centreline apparently, and CIBSE requiring centreline boundaries for internal elements/partitions, but boundaries going to the inside of the face for external elements). Given the inputs required, this doesn’t make sense to me why this should be the case?

For example: Assuming an internal space with dimensions 1m x 1m x 1m (width x height x depth) which is entirely enclosed by 4 internal partitions (thickness 10cm say) with no openings or recesses (for completeness, let’s say the space is sandwiched between 2 slabs totalling 6 encapsulating elements altogether).

So this space has volume 1m^3. The total surface area of the space manifold (geometrically speaking) is 6m^2. And the total surface area of all the elements exposed/visible to the space is 6m^2 (4 partitions and 2 slabs).

I would currently generate 6 boundaries (4 partitions and 2 slabs), each boundary having a surface area of 1m^2, situated on the face of the corresponding element for that boundary. So the total surface area of all the space boundaries equals the total surface area of the space manifold, i.e. 6m^2.

If the boundary is to be to situated on the centreline, this essentially increases the surface area of each boundary resulting in the total surface area of all the boundaries being greater than the total surface area of the space manifold, however the surface area of the element exposed/visible to the space has not changed.

Since the U value in both cases remains the same the result would be different to if the boundary was situated on the face to if it was on the centreline?

If we then inflate the surrounding elements to say 20cm thickness. The space does not change, i.e. the volume and surface area of the space manifold are the same as in the previous example, the total surface area of all the elements exposed to the space does not change (it still equals 6m^2), however the U values for each element change to reflect the change in thickness of the partitions wrt to the previous example with thinner partitions.

This will obviously produce a different result.

If we have boundaries that are situated on the inside face of the element, the only quantity which has changed is the thermal transmittance, which makes sense to me. However having boundaries at the centreline means we have 2 quantities which have changed for the calculation wrt the previous example with thinner partitions. The U values reflecting the change in thickness of the elements, and the total surface areas of all the elements, however the space volume and total surface area of the elements exposed/visible to the space have not changed, resulting in a different result yet again?

This seems to be the case with the example files from Revit which have been given to me. the spaces essentially represent the volume of air in the room, however the boundaries extend to the centreline producing a larger manifold (if calculating a space from the boundaries) than the space. I know this is a setting in Revit to create boundaries to centreline, however there is no corresponding setting in Revit for the space to be extended to the centrelines of the enclosing element afaik. This runs the risk of producing wildly different results, and the error margin could increase substantially with larger rooms and spaces.

And then I get gbXMLfiles in which the space is defined by the boundaries which are at the centreline, resulting in space volumes greater than the amount of air in the room. Introducing even more difference to the original calc results.

I don’t understand why going to the centreline is important as it doesn’t appear to add any benefit wrt to the surface area’s and could produce wildly different results even though the thickness ‘metric’ wrt to the previous examples is already captured by the U value. In fact I would have thought this introduces even more error.

I appreciate that certain admittance calcs require this from the geometry in order to get accreditation, but is this correct or have I got some misunderstanding with something? And if it is correct/desired, what is the rationale behind creating BEM’s this way as it doesn’t seem accurate to me?

Many thanks in advanced for any enlightenment on this :slight_smile:

Disclaimer: I am neither a physicist, building services engineer nor seasoned Revit user. I am a developer, so please forgive my naivety and lack of correct terminology.

And to complicate matters, the approach in Belgium for Energy Level is based on gross volumes (protected gross volume) and defining building envelopes at the outside face of the walls and top side of the roof. Quiet in contrast with other tools which are often based on net-spaces.

So it seems we probably have to take multiple locations for boundaries into account, depending on the regulations and calculation methods. Many BEM tools often simplify all elements as flat faces, but you’d have to compensate in your modelling with the regulation you follow.

Disclaimer - this is not my expertise, but a constraint we often encounter when attempting to reuse information from models for energy calculations

1 Like

Would this mean that using the Belgium model, adjacent internal spaces would essentially overlap since they would both include the volume of the partitioning element as the boundary should go to the outisde face of the element for both cases?

I guess this was all resolved by many people a lot smarter and knoledgable than me over a longer period of time that I care to take a guess at, so I won’t question the rationale (and also since it seems to vary depeding on the geographic location and model required for the calc). If thats what the customers require to get the acreditation with which ever model they choose to use, then thats what the customer will get :).

Darn, I was hoping for a quick win but it seems in any case, more geometry processing and analysis would be required and there would be no escaping that in the current workflows that I am hoping to tap into.

Thankyou very much for you taking your time to reply. Much appreciated.