First+Exercise

First simple simulations
//** The first set of simulations should be ones of a simple box, as used to illustrate the principles in the first introductory lecture on ventilation. **//
 * The series starts with a volume 4000 x 4000 in plan, 2700 high. The walls are a nominal 200 thick. There is one opening 900 wide, on the windward face, and the wind is normal to the wall.
 * The second variation has an opening of identical size, directly opposite the ‘inlet’ opening, in the leeward face.
 * Variation 3 increases that ‘outlet’ opening to 1800 wide.
 * Variation 4 reverts to an ‘outlet’ of 900 wide, but increases the ‘inlet’ to 1800 wide.
 * Variation 5 removes the ‘outlet’ from the leeward side, and replaces it with a 900 wide ‘outlet’ in the middle of one of the ‘side’ walls.
 * Variation 6 retains Variation 5 for the outlet, but reduces the inlet back to the 900 wide original - and moves the ‘inlet’ from the centre of the windward face, in the direction away from the outlet side, such that the edge of the inlet is in the centre of the windward wall.
 * Variation 7 places a ‘blade wall’ 600 deep (nominal 100 thick and full height) on the edge of the inlet opening on the side at the centre of the wall.
 * Variation 8 ‘detaches’ this blade wall, such that there is a 100 wide gap between the blade and the wall at the edge of the window.

It would be good to see if the streamlines and velocities inside the volume approximately correspond to the illustrations in Reed’s original paper, still used to teach architecture students. I suggest using a wind speed of 3m/s (approx. 11kmph). The wind domain would have to be agreed and standardised between the variations. Between the four students, this represents two simulations each, and assumes you are cooperating on the SketchUp models. //I have not given a download link to mine, because it is drawn with redundant edges, and may be a bad one from which to start.


 * STUDENT VIRTUAL WIND CONFIGURATIONS for First simple simulations **

First exercise configurations (I hope this is the right place to put this) I re-orientated the sketchup model 90°s on red axis and found it quicker to configure the domain and wind direction. I have also re scaled the sketch up file as I was modelling in millimetres when the file was in metres
 * SEAN** - (SWP)


 * SWP- Please confirm if instances 7 and 8 are correct**

I checked the model for water tightness using the virtual wind plugin. All good. Exported the file using the (export to VWS format all) button.
 * SWP-Should the export of the exercises be made of each individual instance?**

These are the setting I used to configure the domain, atmospheric conditions and domain wind monitor.





Ran simulation of the first instance three times.

First time I forgot to put a wind monitor so I wasted 30mins with no results. Second time added a domain wind monitor with (velocity sampling) tab selected the simulation ran for roughly 40mins. Third time selected all the sampling tabs, the simulation ran over an hour.

With the results of the third exercise I generated an animation. see below media type="file" key="instance 1 SWP - Computer.m4v" width="300" height="300"

ANDY STEVE: As we discussed this morning, there seems to be something wrong. Sean pointed out that you seem to have the breeze blowing through your solid walls (see the still below).

Have checked out the sketch up model and added in the blade walls, they seemed to be missing from the file attached above.



ANDY//

try this one. see the discussion page for an explanation





i seem to have avi files yet sean's were m4v. this may be why mine don't appear to be directly playable from this window. will look into it further

SWP - hey mate I converted the video files using quicktime 10 also reduced the file size allot

ANDY

In light of Brian's comments on canopies/watertightness etc here is a new vw file which is all canopies. Also check out the comments on 'block editor' for how VW is actually understanding the modeling of opening. **SEAN:** Hey mate can you send accompanying folder structure the Assembled Raw and Texture folders with it. I just tried to do a simulation, submitted it and everything and it just disappeared from the VINE viewer. I'm thinking it didn't know where to put anything. (Has anyone else tried to run simulation with this file yet?) ANDY: would love to but the main file is 125MB and won't zip up to less than 111MB so i can't yousendit. busy now but will have another look later on.

This folder should contain everything, the previous version i tried to post also contained results data; it will make it difficult to share results laden files due to their size, hopefully that won't be an issue SEAN - This is working and am currently running simulation. It must need the enclosing file structure to run correctly.

Note: it was not as simple as just exporting the original skp file as canopies, it gave error message (see grab below). not sure why this occurred (maybe my machine) but i had to remodel everything using just single planes. it now seems to be running ok, i will post again if not the case.

**Steve:** This is consistent with Brian's advice in the Discussion, in response to the 'watertight' vs. 'canopy' issue. And consistent with the apparent ideology that the 'watertight' runs are for the simplest possible block models in the first instance. Good to find it out now, and to understand how 'realistic' the SketchUp models can be at various stages of the workflow.



ANDY: Some results. Am posting 2 videos, both of Option 7 (the one with the blade wall attached to the OUTSIDE of the building). The first has the colour limits on default setting, the second i adjusted them slightly (no real rules just a guestimate); note the differences in the patterns you can see. media type="file" key="Exercise 1 Final Final 200 option 7 - Computer.m4v" width="300" height="300" media type="file" key="Exercise 1 Final Final 200 option 7 limits altered - Computer.m4v" width="300" height="300"media type="file" key="Exercise 1 Final Final 200 option 7 limits altered Vectors - Computer.m4v" width="300" height="300"

Option 8 ( with the wall detached from the building): i have actually modeled the wall as being 200mm off to allow for the mesh to pick up the gap, looking at the following animations I am unsure that it has! media type="file" key="Exercise 1 Final Final 200 Opt 8 - Computer.m4v" width="300" height="300" media type="file" key="Exercise 1 Final Final 200 Opt 8 OrthoSliceVectors - Computer.m4v" width="300" height="300" I have attached the VW file, can someone please turn on the block editor and check visually whether the gap seems to exist; i really can't tell. Based on my previous post about predicting the gap size i'm going to move the blade 320mm from the wall theoretically the mesh cannot fail to pick up at least 1 vertical column of gap!

FIRST INSTANCE OF EXERCISE 1 media type="file" key="instance 1 orthoslice solid contours - Computer.m4v" width="300" height="300"media type="file" key="instance 1 orthoslice vectors - Computer.m4v" width="300" height="300"media type="file" key="instance 1 orthoslice vectors white - Computer.m4v" width="300" height="300"media type="file" key="instance 1 orthoslice solid contours white - Computer.m4v" width="300" height="300"

SECOND INSTANCE

media type="file" key="instance 2 orthoslice vectors white y axis - Computer.m4v" width="300" height="300"media type="file" key="instance 2 orthoslice solid contours white - Computer.m4v" width="300" height="300"media type="file" key="instance 2 orthoslice vectors white - Computer.m4v" width="300" height="300"

ANDY this is option 8 again. The model was slightly wrong before; the blade is now precisely 200mm perpendicular from the opening edge. The flow does still not seem to be as 'good' as i would have thought so i will now move the exit further to the rear of the model and re-run. Incidentally, i'm posting 2 versions the first is animation taken at slice vector 1, the other at slice vector 3. It seems that their are 31 z slice vectors, i had previously assumed they were at 1m metre intervals not quite sure why the number 31 but anyhow lifted the plane to around 1m from ground; minimal difference i think! should make more of a difference???

media type="file" key="Slice 1 - Computer.m4v" width="300" height="300" media type="file" key="Slice 3 - Computer.m4v" width="300" height="300"

The new results with the door location moved media type="file" key="Option 9 - Computer.m4v" width="300" height="300" media type="file" key="Option 9 Vecors - Computer.m4v" width="300" height="300"

ISteve: The runs for #8 and #9 are finally showing clearer behaviours that I recognise as being correlated with Reed's description of the wind tunnel experiments. Especially in the #8 animations, you can clearly see the air strem entering the inlet, and curving off to the opposite side from the gapbetween the blade and the wall.

Reed's interpretation is that the pressure gradient on the facade is such that there is a raised static pressure near the middle, and lesser raised pressures as you move towards the corners. The gap allows the higher pressure to act on the air stream, pushing it sideways as it enters the inlet, carrying further into the room before it turns eventually to exit via the side outlet.

The earlier runs are showing some fairly bizare phenomena, but at least you can see the air stream passing through between the inlet and facing outlet. The turbulance and the directions of the velocity vectors on the outside are moving through much bigger ranges than I had thought likely. In one of the last ones, there is an energetic horizontal large eddy behind the outlet. It dominates the visualisation, where I would not have expected to see it at all.

I wonder if we are getting sufficient convergence on the calculation?

Steve again: I should have asked before about the wind velocity. After today's conversation I now understand that the simulations are still bedevilled by that issue of where you specify the wind speed - for which I thought that Brian had earlier provided some explanation.

Clearly, the 3m/s had recommended is of interest at the level of the model, not at the upper limit of the typical wind profile! If the latter is what we have in these runs, we have lovely lazy low frequency turbulance swamping the ventilation patterns that would begin to be clearer at higher velocities. I should have been more attentive to the minimum/maximum range in the velocity figures plotted.

Hopefully, when re-run with the higher wind speed the patterns will settle down to something more recognisable. But let's move on to the second model, and get this cleared up on that one.

media type="file" key="Option 9 adjusted 20ms - Computer.m4v" width="300" height="300" media type="file" key="Option 9 adjusted 20ms Vectors - Computer.m4v" width="300" height="300" (Andy) Above are the results of the huge simulation run with the wind speed increased to 20.5m/s

(Steve) These are very impressive. The model is clearly performing, and it seems to be picking up the relevant detail. The turbulence seems to be enough to shift the pressure distribution on the windward facade around, such that the air stream entering the opening is sometimes affected by the gap, and sometimes not affected. The dynamic view of that airflow through the room is very revealing......it whips around much more than I would have thought likely. It suggests there are less 'dead' areas in a typical room than most people would assume.

Now to get Brian and Kevin to advise on a less onerous output, so we can knock down computation time for the second model, even if we don't get such compelling visuals.

Sam:

3m/s wind on to option 7(two openings of equal size) media type="file" key="10ms wind.mp4" width="300" height="300"

10m/s wind on to option 6 (two openings one size bigger then the other) media type="file" key="10ms wind1.mp4" width="300" height="300"

SIEM:

Virtual Wind Exercise 1 Simulation 1: Images: Wind Resolution: Medium Wind Speed: 10m/s

Building 3 - small inlet, large outlet Windward Side, Cage visible. Leeward Side, Cage Visible:

Building 4 - large inlet, small outlet

Windward Side Cage Visible



Leeward Side, Cage Visible

Even though the cage in these simulation slightly impedes the openings, I am confident that for reasons of the experiment to test the Reed papers finding on the issue of the size of the inlet vs the size of the outlet as the opening are still in similar proportions this should not too much affect the study as the registered openings are still proportional.

Reed:

Building 4, SIMULATION 1, see v-high default time step interval

This would take too long to compute due to the extremely high number of time step intervals. I decided to try change the interval number so there would be roughly 400 time steps over the given time. This did not work : With 0.001 timesteps (still a lot) it estimates 6 hours initially. I waited half an hour for the simulation to “settle down” and give a more accurate time estimate, but the time estimate only increased.

I decided to stop and try a time step interval of 0.005, this did not work as the time step interval was now too large, and it crashed the program almost immediately.

I put the interval back at 0.001 and left it. That is 100,000 time steps over 100 seconds…

This time my computer froze.

Changed windspeed from 10m/s to 3m/s at a 10m height. And tried a 0.025 time step interval again. Again this large time step interval crashed the program.

Tried 3m/s w/ 0.0064 timestep. Estimates 56 minutes. This is good. The simulation took 3399 seconds to run or 56.65 minutes, which is exactly the estimated time!

ANIMATIONS: ANIMATION 1, BUILDING 4 - Large Inlet vs Small Outlet media type="file" key="exercise 2 building 4 3mps wind.mp4" width="300" height="300"

It appears that the wind is too slow to show what is happening. I realised my wind range was monitoring wind speeds up to 9m/s, which of course is ridiculous as my wind speed input was only 3m/s, so I adjusted the range so the animation looks more lively. I then dropped the custom limits to 2m/s, and it is much more lively. The wind now obviously SLOWS DOWN inside the building and has no clear direction.

media type="file" key="exercise 2 building 4 3mps wind maximum 2.mp4" width="300" height="300"

BUILDING 3: In comparison, same conditions, there is a clear stream throughout the building, the wind speed at times is even greater than that outside! However there is more of a range of movement than just a direct stream through the building, the wind actually flushes most of the room.

media type="file" key="exercise 2 building 3 3mps wind maximum 2.mp4" width="300" height="300"

In conclusion using virtualwind for this exercise has helped prove Reeds theory on wind movement through a room in regards to inlet and outlet sizes.