Understanding turbulent air flow in underground chamber caused by thermal gradient

JD1 · August 29, 2021, 07:14

Hi All,

Totally newbie to CFD here. I work with very precise temperature data collected from stop tap in underground chambers to identify water leaks on pipes. We have noticed that we tend to get 'noise' in the data, this is particularly more evident in the winter months.

We have a theory that this is caused by convection due to the temperature gradient from the top of the chamber to the bottom of the chamber (i.e. the cold air sinking and warm air rising). The temperature fluctuations we see tend to be on the order of 0.05-0.1 degrees C per minute. See schematic below:

thermal_gradient.png

I was hoping to use OpenFoam to test out this theory, however I've spent several days just trying to get it set up and get to grips with 1 or 2 tutorials.

I'm looking for some guidance on what is involved in getting the schematic above implemented as a CFD model to test the theory and how long you guys think this would take?

Any advice or suggestions would be greatly appreciated,

Thanks,

J

JD1 · September 3, 2021, 17:14

Anyone have any advice on this?
Thanks

Tobermory · September 4, 2021, 06:14

JD - I think that you are undoubtedly right. The signal noise is likely to be due to turbulent fluctuations/air currents. The timescale of the fluctuations will give you a hint of the scale of the currents.

For modelling - bad news first - this will be really difficult to model with any accuracy I am afraid. The simpler, RANS type solvers like buoyantPimpleFoam will provide you with a solution of the time-averaged velocity and temperature fields ... but will contain no information on the turbulent motions in the cavern.

If you want to get a time trace like the one that you are seeing from your TCs, then you'll need to run a large eddy simulation (LES) type solver to simulate the actual turbulent eddies. This will not be a trivial task, esp for someone new to CFD, and will require some significant computing resources and plenty of time, as you try and tune the heat sources around the cavern to give you the behaviour that you observe in the field.

If you want a qualitative picture of impact of thermal currents on TC readings, then you could take a simpler approach and just put a low level heat source in a box and simulate that with LES, to demonstrate that convection currents can cause TC noise ... but we know that's true, so I am not sure that that would be very informative. If you do take this approach, then you might think about using FDS, which is free and much friendlier to use for the CFD novice.

Sorry to be the bearer of bad news, but I hope that this saves you some time? All the best.

JD1 · September 4, 2021, 10:26

Tobermory, thank you very much for your reply, very helpful.

With regards to the modelling, I should have made it a bit clearer that I am not interested in recreating an exact time series to match our observations, I'm primarily interested in the average frequency and magnitude of temperature fluctuations in a chamber given a specific temperature gradient in a chamber. The goal would be to prove that the magnitude and frequency of the thermal fluctuations are caused by the temperature gradient from the top of the chamber to the bottom of the chamber. This would help prove the theory and potentially provide more insight into how to insulate the chambers.

I'm glad to hear that you believe this to be the case too.

You mention FDS, is this a fluid dynamics modelling software?

Thanks again,

JD

Tobermory · September 4, 2021, 11:34

Understood - to get the frequency and magnitude of the fluctuations, you'll need to simulate the turbulence directly though, rather than modelling it (there are no suitable turbulence fluctuation models for this type of application). Incidentally - what is the timescale of the fluctuations (will be masked slightly by the time response of your TCs I guess)?

That's assuming that the variations that you see are due to flow turbulence, and not just to "flapping"/jostling of the convection currents. If it's the latter, then modellers have been able to model somethnig similar - Rayleigh Benard convection cells (see eg https://www.sciencedirect.com/scienc...42727X99000077) using a VLES approach, but with some funky RANS closure scheme that would not be simple to reproduce in OpenFOAM.

Finally, FDS (Fire Dynamics Simulator) is a fire code developed by NIST in the US - it's a "LES" (actually VLES, I think) code with a simple interface, designed for fires. You might be able to use it with a heat sink at roof level to model the convection currents, but only in a qualitative manner I would suggest.

August 29, 2021, 07:14	Understanding turbulent air flow in underground chamber caused by thermal gradient	#1
JD1 New Member Join Date: Aug 2021 Posts: 3 Rep Power: 4	Hi All, Totally newbie to CFD here. I work with very precise temperature data collected from stop tap in underground chambers to identify water leaks on pipes. We have noticed that we tend to get 'noise' in the data, this is particularly more evident in the winter months. We have a theory that this is caused by convection due to the temperature gradient from the top of the chamber to the bottom of the chamber (i.e. the cold air sinking and warm air rising). The temperature fluctuations we see tend to be on the order of 0.05-0.1 degrees C per minute. See schematic below: thermal_gradient.png I was hoping to use OpenFoam to test out this theory, however I've spent several days just trying to get it set up and get to grips with 1 or 2 tutorials. I'm looking for some guidance on what is involved in getting the schematic above implemented as a CFD model to test the theory and how long you guys think this would take? Any advice or suggestions would be greatly appreciated, Thanks, J

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Simulation of a turbulent flow into a diffuser	Sam6789	SU2	1	August 13, 2021 16:42
Problem with divergence	TDK	FLUENT	13	December 14, 2018 06:00
pisoFOAM (LES) - internal pipe flow - convergence	gu1	OpenFOAM Running, Solving & CFD	0	January 11, 2018 16:39
Flow analysis from a pressurised chamber to a larger chamber	WT4	Main CFD Forum	3	November 4, 2015 22:19
fluid flow fundas	ram	Main CFD Forum	5	June 17, 2000 21:31

September 3, 2021, 17:14		#2
JD1 New Member Join Date: Aug 2021 Posts: 3 Rep Power: 4	Anyone have any advice on this? Thanks

September 4, 2021, 06:14		#3
Tobermory Senior Member Join Date: Apr 2020 Location: UK Posts: 670 Rep Power: 14	JD - I think that you are undoubtedly right. The signal noise is likely to be due to turbulent fluctuations/air currents. The timescale of the fluctuations will give you a hint of the scale of the currents. For modelling - bad news first - this will be really difficult to model with any accuracy I am afraid. The simpler, RANS type solvers like buoyantPimpleFoam will provide you with a solution of the time-averaged velocity and temperature fields ... but will contain no information on the turbulent motions in the cavern. If you want to get a time trace like the one that you are seeing from your TCs, then you'll need to run a large eddy simulation (LES) type solver to simulate the actual turbulent eddies. This will not be a trivial task, esp for someone new to CFD, and will require some significant computing resources and plenty of time, as you try and tune the heat sources around the cavern to give you the behaviour that you observe in the field. If you want a qualitative picture of impact of thermal currents on TC readings, then you could take a simpler approach and just put a low level heat source in a box and simulate that with LES, to demonstrate that convection currents can cause TC noise ... but we know that's true, so I am not sure that that would be very informative. If you do take this approach, then you might think about using FDS, which is free and much friendlier to use for the CFD novice. Sorry to be the bearer of bad news, but I hope that this saves you some time? All the best.

September 4, 2021, 10:26		#4
JD1 New Member Join Date: Aug 2021 Posts: 3 Rep Power: 4	Tobermory, thank you very much for your reply, very helpful. With regards to the modelling, I should have made it a bit clearer that I am not interested in recreating an exact time series to match our observations, I'm primarily interested in the average frequency and magnitude of temperature fluctuations in a chamber given a specific temperature gradient in a chamber. The goal would be to prove that the magnitude and frequency of the thermal fluctuations are caused by the temperature gradient from the top of the chamber to the bottom of the chamber. This would help prove the theory and potentially provide more insight into how to insulate the chambers. I'm glad to hear that you believe this to be the case too. You mention FDS, is this a fluid dynamics modelling software? Thanks again, JD

September 4, 2021, 11:34		#5
Tobermory Senior Member Join Date: Apr 2020 Location: UK Posts: 670 Rep Power: 14	Understood - to get the frequency and magnitude of the fluctuations, you'll need to simulate the turbulence directly though, rather than modelling it (there are no suitable turbulence fluctuation models for this type of application). Incidentally - what is the timescale of the fluctuations (will be masked slightly by the time response of your TCs I guess)? That's assuming that the variations that you see are due to flow turbulence, and not just to "flapping"/jostling of the convection currents. If it's the latter, then modellers have been able to model somethnig similar - Rayleigh Benard convection cells (see eg https://www.sciencedirect.com/scienc...42727X99000077) using a VLES approach, but with some funky RANS closure scheme that would not be simple to reproduce in OpenFOAM. Finally, FDS (Fire Dynamics Simulator) is a fire code developed by NIST in the US - it's a "LES" (actually VLES, I think) code with a simple interface, designed for fires. You might be able to use it with a heat sink at roof level to model the convection currents, but only in a qualitative manner I would suggest.