Improving Thermal Modeling Efficiency of Curtain Walls

One of the things that most engineers would agree on is that modeling is fun. Structural, Thermal, DFM, DFA, the list seems endless.  Most engineers would also agree that importing the dimensional information, cleaning up the model to prepare it for modeling is about as fun as the root canal that I just had.

As a mechanical engineer at Enclos corP (not a misprint, the trendy marketing way of spelling), I led the thermal and energy analysis of curtain wall assemblies.  For those not familiar with the  term, a curtain wall is the exterior glass and aluminum on a modern building. It is the "face" of the building that most of us associate with the design of the building.   It is called a curtain wall because it must support itself only, not the building.

Transient

Enclos designs curtain walls for large 40 to 50 story buildings. Each unique structural element of the curtain wall must be modeled to determine its thermal resistance and interior temperatures.  Most projects had between 8 to 10 elements to be analyzed.

Enclos had adopted software and analysis methods that required between 2 and 3 hours to model and solve each thermal detail.  A large portion of this time was the cleaning up of details created in AutoCad for the successful import into their thermal modeling software called THERM and prior to THERM, Algor.

One of the facets that most design engineers learn about themselves and other designers is that no one of us creates layouts in the same manner.  Power users may have the best technique and create models that are easily imported . But others can leave a great deal of features  in a mode that the analysis software may have a hard time interpreting.

This was the case at Enclos.  Designers created the details that were then given to engineers who then dutifully recreated the details in a level of accuracy that THERM and Algor would accept.  The best analogy that I have is that we would burn down the house - and rebuild it.

Efficient? No. Enjoyable? No.

I recall when I arrived at Enclos, I asked the engineer who had developed the method if he enjoyed the current modeling method. He firmly said No.

One of the great things about the Internet is that designers can share with one another.  This is where I started.  On a CFD support website, I explained my situation to my peers.  A voice from Germany suggested trying software that they had had great success with called Flixo.

What was great about Flixo was that it wasn't flustered by all of the personal loose ends that we as designers frequently leave.  It effectively said " I know what your intent is.  I'll take it from there."  Importing the model ,applying materials and running the model became a 15 minute pleasure instead of a 3 hour ordeal.

 

Transient

More time for modeling. Less time on clean up.

As October rolls around, I think about how grateful I am that another engineer was kind enough to share. And I think about Octoberfest and Beer.