[MBttR]

Hi all,

I'm trying to simulate a simple problem of a heat source (130W) in an airflow (roughly 10 m/s). I'm doing this in OpenFOAM but this shouldn't matter. My solutions are blowing up and it's driving me into frustration to such a degree that I have to go back and look at the basics.

It is a transient problem, right? I am interested in knowing which temperature the heat source reaches while being subjected to this air flow. When googling definitions of transient vs steady state, I always find the relatively unsatisfying answer of a transient solution changing in time. This is what is happening here, right? As both the temperature of the air and the heating element will change over time.

Then I start doubting, what is NOT a transient solution? Doesn't everything change in time? Why else would we be running simulations.

Hoping someone can clear this up for me,
Thanks,

Bruno

[bighead]

check your mesh quality first. they should be good before running any calculations.

why do you say its a transient problem ?? based on what you described, it does not seem to be it. If it truely is, then you have to change your timesteps to smaller interval as well

[FMDenaro]

I am not sure about your doubt...

transient is an evolving stage that, starting from an initial condition can drive to either a steady or a ully developed unsteady state. That depends on the physics of the problem, the Reynolds number being an indicator. Furthermore, for turbulent flows that depends also on the adopted formulation.

In your case, given the heat flux and the asymptotic velocity is not sufficient to evaluate the main non-dimensional numbers. Are you considering air but what about the characteristic lenght?

[MBttR]

If transient vs steady state relates to the FLOW CONDITIONS, then I'm thinking my problem is steady, after all, it will stay at 130W and 10m/s. But I'm also not convinced that's a satisfying answer, as for example a high angle of attack airfoil will be of transient nature at a certain flow velocity, with constant flow conditions. Perhaps there the reason for it being considered transient is because the flow conditions changes over LOCATION (hence location along the chord), but the definition of transient is changing over TIME?

Sorry for thinking aloud, but clearly I'm missing something.

[MBttR]

Originally Posted by bighead

check your mesh quality first. they should be good before running any calculations.

why do you say its a transient problem ?? based on what you described, it does not seem to be it. If it truely is, then you have to change your timesteps to smaller interval as well

Thanks for the quick reply!

I thought it was transient as I wanted to see the temperature change over time, but as said in my next post (which I posted before seeing your replies, sorry about that) I am starting to doubt that it might be steady state, as the actual flow conditions don't change over time. Then again, I can't think of a problem where flow conditions do change over time (except for example a simulation involving wind gusts or a heater which increases in intensitivity).

My timesteps are automatically adjusted to have a Co number < 0.6.

[FMDenaro]

Originally Posted by MBttR

If transient vs steady state relates to the FLOW CONDITIONS, then I'm thinking my problem is steady, after all, it will stay at 130W and 10m/s. But I'm also not convinced that's a satisfying answer, as for example a high angle of attack airfoil will be of transient nature at a certain flow velocity, with constant flow conditions. Perhaps there the reason for it being considered transient is because the flow conditions changes over LOCATION (hence location along the chord), but the definition of transient is changing over TIME?

Sorry for thinking aloud, but clearly I'm missing something.

I am used to think about "transient" as a time-evolving stage. It can be of physical nature (for example an airfoil that starting from zero angle of attack reaches a certain angle) but it can also be of numerical nature (for example a flow over a cylinder; starting from arbitrary initial condition, the numerical solution has a numerical transient).

In your case 130W and 10m/s said nothing about the physical state of the solution. You can see that your Re number is = 10^6 L. Similarily, from the 130W (maybe W/m^2 ??) you should consider the Rayleigh number.
You can have a numerical transient and then reaching a steady or unsteady solution.

[MBttR]

Originally Posted by FMDenaro

I am not sure about your doubt...

transient is an evolving stage that, starting from an initial condition can drive to either a steady or a ully developed unsteady state. That depends on the physics of the problem, the Reynolds number being an indicator. Furthermore, for turbulent flows that depends also on the adopted formulation.

In your case, given the heat flux and the asymptotic velocity is not sufficient to evaluate the main non-dimensional numbers. Are you considering air but what about the characteristic lenght?

Hi Filippo, thanks. I'm not entirely sure what you mean with your last paragraph. What do you mean with asymptotic velocity? How are heat flux + velocity not enough to evaluate the main non-dimensional numbers? What exactly do you mean with non-dimensional numbers, I am only interested in temperature.

I am considering air, yes. Characteristic length is a good one.. I hadn't considered this to be a factor here (sorry if that's stupid, I'm used to external flow problems, not CHT). Characteristic length here I suppose is 30mm (length of the heating element in streamwise direction). In OpenFOAM though, you define the kinematic viscosity, after which it will determine the Reynolds number from that. I have the kinematic viscosity of air (1.8e-5), leading to a Reynolds number of (I believe) roughly 17000.

*edit*
You posted a second comment while I was writing this one, will consider that now and come back, thanks!

[MBttR]

Originally Posted by FMDenaro

I am used to think about "transient" as a time-evolving stage. It can be of physical nature (for example an airfoil that starting from zero angle of attack reaches a certain angle) but it can also be of numerical nature (for example a flow over a cylinder; starting from arbitrary initial condition, the numerical solution has a numerical transient).

In your case 130W and 10m/s said nothing about the physical state of the solution. You can see that your Re number is = 10^6 L. Similarily, from the 130W (maybe W/m^2 ??) you should consider the Rayleigh number.
You can have a numerical transient and then reaching a steady or unsteady solution.

How do you come at a Re number of 10^6 L ? Is this just a quick estimation? Shouldn't it be (10/1.4e-5) = 7e5 L? Only a factor 2, I know, but want to make sure.

Was not familiar with the Rayleigh number. It is deifnitely W, not W/m2. The heating element is an electric motor, I know the power from measurements, roughly know the efficiency and from there considered how much power (hence, 130W) is lost as heat. Looking up the definition of the Rayleigh number, I'm not sure I can define it. It has the surface temperature in its definition, which is what I'd like to calculate. But the order of magnitude I believe will be (with characteristic diameter 0,08 cause the motor is cylinder shaped and an expected 80deg temp difference) 5.2e6. Does that figure make sense? How does it come into play in determining whether I have a transient or steady case?

[MBttR]

Originally Posted by bighead

check your mesh quality first. they should be good before running any calculations.

why do you say its a transient problem ?? based on what you described, it does not seem to be it. If it truely is, then you have to change your timesteps to smaller interval as well

Are there any rules of thumb for the domain size around a cht problem? I have a pretty small domain right now, with zeroGradient boundaries on the walls.

[FMDenaro]

I would think to a turbulent flow condition

[MBttR]

Originally Posted by FMDenaro

I would think to a turbulent flow condition

But transient or steady state? And why?

[MBttR]

If I go back to the textbook definition, it must be transient. If I keep my eyes fixed at a certain point with a certain position x,y,z, the properties there (T, V) are still a function of time. Sure, it might reach an equilibrium at some point and be steadystate then, but I think it still requires me to solve with a transient solver. If anyone disagrees please do tell me.

By the way, I have just tried solving with a steady state solver and my solution still blows up, so it must be due to my boundary conditions, mesh, ... Something I will ask further in the OpenFOAM forum. Thanks all for the quick replies.

[FMDenaro]

Turbulence is ..turbulence, it IS unsteady by definition.
You can only think to a STATISTICAL steady state in case of energy equilibrium (production equal to dissipation). In such a case, you can use the statistally averaged form of the equations, the so-called RANS that are steady and solve only for the zero-th order statistic part of the flow (i.e. averaged variables).

[MBttR]

Originally Posted by FMDenaro

Turbulence is ..turbulence, it IS unsteady by definition.
You can only think to a STATISTICAL steady state in case of energy equilibrium (production equal to dissipation). In such a case, you can use the statistally averaged form of the equations, the so-called RANS that are steady and solve only for the zero-th order statistic part of the flow (i.e. averaged variables).

I've always learned that a steady flow can be fully turbulent as well, as it only refers to the transient nature of the flow. Anyway! I got what I came for; reassurance that I should stick to transient flow and hence my problems are of an entirely different nature. Thanks for all the help

[FMDenaro]

Originally Posted by MBttR

I've always learned that a steady flow can be fully turbulent as well, as it only refers to the transient nature of the flow. Anyway! I got what I came for; reassurance that I should stick to transient flow and hence my problems are of an entirely different nature. Thanks for all the help

A turbulent flow can be only STATISTICALLY steady ... that means that the statistical averaged of any variable does not vary in time.