Hi Barry,
No problem, good luck with your exams and hopefully we can get this rolling when you become less bogged down!
Best wish,
Darren
1 Like
Hi Darren,
Finally got the data downloaded. Regarding your concern over the force plots I’m not very clear about how to extrapolate the information as acceptable or not so they are just posted below. Do let me know what you think, I’m very interested to understand these graphs. The result controls are as what you set in your sample simulation run so they refer to the same set of geometries that you have selected previously.
Force plot of building near inlet
Force plot of structure behind building of interest
Force plot of building near outlet
Further below are the vertical slices in the X direction and then a closer look at the inlet at the velocity gradient.
X direction slice
Inlet from X direction slice
It seems that the ABL input is behaving as intended.
Regarding your question of the objective of the simulation, its two fold. The first is to deduce the average velocity of the walking spaces at level 4 of the building of interest (The very refined building in the center, do let me know if you need a screenshot for better understanding). The second, to obtain TKE values outside a region of interest close to a wall at the same level 4 under the most prominent wind inlet condition that can then be input into a drone to obtain the response that can then further be used to tune the PID control to compensate for adverse wind conditions when the drone is operating at that area of interest.
The first objective is the “side report” that I will need to present sometime next week. Meeting as much of the Greenmark simulation parameters is a big objective of mine as it equates to a more relevant and quality result. Fortunately, this ties directly to the second objective which is related to my Final Year Project hence my further interest in ensuring that the simulation results are as accurate and realistic as possible.
On the side of course I want to know the fundamentals of what I’m doing so I can apply that in the future if need be. Hence all the questions 
That sounds like a great idea Darren, I will try it out after my examination period. Currently my setup and results do seem satisfactory considering the left over time I have before submission.
Currently I’m preparing my results for submission and my fellow coursemates are also running different variations of the wind input (different directions and speeds with reference to the Greenmark document).
Hope I answered all your queries sufficiently! Do feel free to ask me more!
Regards,
Barry
Hi Barry sorry for slow response, yesterday was a busy day at the university for me too.
I am concerned that the force plot hasn’t reached a steady state (or starting to) near the area of interest, I would add a result control on your fourth-floor, maybe an area average and monitor TKE as well as putting some force controls on the fourth floor.
The force plots show good convergence at the inlet but not as good where we are actually concerned.
Kind regards,
Darren
1 Like
Hi Darren,
No worries, seem like we are all swamped!
Sure i’ll add the needed result controls upon the next set of runs. It makes sense now that I think about it when it comes to whether the area of interest has reached its steady-state. Currently the results I have, as mentioned, are quite satisfactory for my interim submission.
Thanks for the help Darren!
Regards,
Barry
1 Like
Hey Darren,
Well looks like I’m back to tackling this project again.
I’ve used your recommendation to adjust the velocity relaxation factors to get the pressure better converge. However as you can see below, increasing the relaxation factor to 0.8 actually causes the pressure to reach steady-state much further away from the ideal 1e-3 point.
U Relaxation factor 0.7 (default)
U Relaxation factor 0.8
So at this point, I decided to try out adjusting the relaxation factor for pressure itself and with some trial and error got a moderately close idealized convergence with a pressure relaxation factor of 5.25 as shown below. A factor of 5.4 gives instability and an eventual error.
P Relaxation factor 5.25
Ideally the P value would go well below the 1e-3 point. Any further ideas? I think I will adjust the velocity relaxation factor with this current P relaxation factor to see if it does anything. Oh and I’ve added the result controls that you’ve also recommend. You can see them below.
Level 4 Force Plots
Area of interest at level 4 but just outside of building
Do let me know what you think.
Cheers!
Regards,
Barry
Hi Barry, so to have no under relaxation we would set the under-relaxation factors to 1. To improve convergence we would lower these values, so 0.5 is an improvement over the default 0.7 for velocity. So, in fact, going the other way with the values should help you, convergence is sure to be slower though, so add some iterations to account for this.
Hope this helps,
Darren
Hi Darren,
Ah seems like I misinterpreted the usage of the numbers.
Will run it and see what I get.
Cheers and thanks for the input!
Regards,
Barry
Hi Darren,
I’ve dropped the value of U to 0.4 and unfortunately did not increase the maximum run time, but intermittent results were stored and convergence of p seems to be quite satisfactory as you can see below over a run time of about 2000 iterations. What do you think?
The minimum criteria for convergence is 1E-4, I was thinking of ways to get to this value without spending as much computing time and would like to hear you opinion on the viability and effectiveness of these methods:
- Dropping relaxation factor of P to 0.1 (currently 0.3)
- Dropping relaxation factor of velocity by 0.2 (currently 0.4)
- Increasing quality of mesh
Any other way to both get the P value converged or with less computing time?
Cheers.
Regards,
Barry
Hi Barry, my deepest apologies for the late reply, been a bit hectic last few days.
I think only your result control monitors will tell you if it is enough, looking at these results more iterations are required, however convergence might be reached with no further changes to relaxation. As for your suggestion of further reduction, bear in mind that each time these are reduced convergence is slower and may therefore increase time to convergence. If you do more iterations and the pressure residual doesn’t start dropping then it might be necessary to reduce the pressure relaxation factor. But do check the result controls which we previously set up, they may show that nothing further is required!!
Kind regards,
Darren
1 Like
Hi Darren,
No worries! You can take your time to reply!
Thanks for the great input, I will run more iterations and check for further convergence. Hopefully I can start extracting my results soon.
Cheers Darren, Happy Holidays!
Regards,
Barry