Second+Exercise

This model shows the wall locations. Plus where water tanks and a "bathroom" of sorts might go. I'm all for ignoring the open space at the rear of the building for the time being, as it is extremely unlike a composting toilet is going to be nicely nestled in that space.



Another version is going up a bit later this afternoon, I will explain in discussion section.

SIEM:

Attached is the Framing Sketchup of the building for exercise 2.

I will have the modified "planar" version up soon.

Why the framing model is relevant however is for when we modify the building.

Just 2 weeks ago, before a group of students went back up to Fowlers Gap to finish the building, the location of all the openings had not been finalised other than they would "fit in with the grid and frame". The final decisions were being made for aesthetic reasons (large openings towards the view, symmetry etc)

It will be interesting to investigate how if this wind analysis was incorporated into the design, working with the existing framing, how the performance of the building may have been changed for very little extra effort or change to cost.

I am trying to get in touch with the team in Fowlers Gap to see what decisions they have made about the placement of the openings (if they have).

The orientation of the building was decided due to solar access (large sloping roof, covered with solar panels, facing north to heighten efficiency... even though the building is on the top of a hill, in the middle of the desert). So also investigating the impact of the dramatic slope of the roof, and orientation on the internal ventilation should be interesting.

A note: design modification were being made to the roofing of the building due to high wind speeds experienced in the area- the roofs large overhangs and lightweight construction mean it is very likely it could become a large and very dangerous kite.

More to come soon.



Derek Here: I've experimented with the framing model of the second exercise and ran a coarse simulation, the results of which are presented in the spirit of trying provide more ideas on getting value from simple coarse simulations. 1) First, an image of the quick-and-dirty walls I placed on the framing structure. I placed these walls in SketchUp using the framing SketchUp file. In SketchUp, I grouped all the framing members into a single Group; so that upon exporting to Virtualwind's .vws format all these framing members will become I single Canopy entity. This allows me to turn-off the model for these framing members. Turning-off the model means they are not "seen" by the wind (i.e. not seen by the wind solver); and hence are shown in the viewer as semi-transparent green.



2) Second, a snapshot of instantaneous wind velocity vectors on a vertical slice through the building, showing the walls as semi-transparent, and the framing members as semi-transparent. Again, with the model for the framing members turned-off, they are not part of the simulation. Also, I turned-on the ColormapCustomLimits on the OrthoSliceVectors actor and set the upper limit to 4 m/s; hence, the strong red arrows above the building. Another similar snapshot only with the framing members visibility turned-off, and the camera looking downwind.





3) Third, a snapshot of instantaneous wind velocity vectors on a vertical slice cutting through the building lengthwise.



4) Fourth, a snapshot of instantaneous wind velocity vectors on a horizontal slice cutting through the mid-section of the building.

I have 2 .avi animations I created but unfortunately cannot seem to load them. If anyone is interested, I can e-mail them.

The above simulation was run on a resolution of 32x32x32, and took only a few minutes on a single core on my workstation. Although not detailed, running a few analogous simulations in different directions could provide some interesting comparison animations.

ANDY Thanks derek, could you please upload the skp file too. Also you'll struggle to upload the avi files, they need to be converted to m4v files using QuickTime in order to reduce the file size.

DEREK: I've uploaded the .skp file now. One further comment on reducing simulation times. In the Wind Model there is a Time Step item, which is editable. I would recommend first trying to double the size of the time step - which will reduce the simulation time in half - and see if it runs through 50-100 timesteps. If so, it's probably stable and the simulation will run through to completion. HOWEVER, it also may crash, in which case you need to return it to is predetermined value before you changed it. To get it's predetermined value again, just slightly change the wind speed which will force an update on this value. Adjusting the Time Step up by a factor of 2 or 3 will directly reduce the simulation time by the same factor. From experience, if you don't have too much "blockage" of the wind by solid walls, you can usually increase the time step by a factor of 2...sometimes 3....rarely 4.

O.K. One final comment. The best way to gain "skill" in using Virtualwind is to spend a couple of hours just playing with a COARSE resolution simulation that will run in under 5 minutes. Play with different wind directions, wind speeds, time steps, and some of the monitor data visualization actors. What you learn playing with Virtualwind under these conditions will provide you with more confidence when you start running bigger simulations. On the big simulation size, my rule-of-thumb is the overnight run. Having known a few architects in my academic days, I realize "overnight" and "sleep" are not one-and-the-same; but that concept of running a "larger" simulation while your away from your computer is part of the "simulation-running workflow".

Good luck, and have fun. Derek

ANDY Thanks for all that Derek. I can't see the skp file though!

DEREK Yes, there it is.

SIEM I created a new sketchup file with the walls in the correct locations and all the doors "open". Also the planar elements (walls/doors/roof) are a separate group to the framing allowing for the roof/walls/doors visibility to be adjusted seperately to the framing. Please see the below .skp and the virtualwind zipped folder.



I initially ran submitted a simulation with a coarse mesh as a trial, however the simulation visualisation revealed little to no airflow within the building. I activated the block actor and realised the coarse mesh was so large and blocky that the airy pavilion was reading more like a solid block, see the below animation:

media type="file" key="Coarse.mp4" width="300" height="300"

I then tried the mesh as medium, but there was not much of an improvement. After setting the mesh as FINE the building started to read in a more recognisable manner.

However after over 5 hours of computing the simulation submitted became "excited" at step 37 and stopped. I did not realise this until the morning.

37/100 steps however were completed and able to be animated - the results of this are below:

(sorry the video is a bit skizoid, I ran the simulation from a few different views, then changed the range of the results so to be easier to read)

media type="file" key="Fine37.mp4" width="300" height="300"

Below are some image grabs.







The wind paths through the building and velocities confirm what was predicted.

More conclusions to come.