1
$\begingroup$

Given a flow inside a square duct with constant temperature at the walls $(T_{w1} = T_{w2} = T_w)$ the physical property in terms of temperature that repeats itself in a periodic fashion is the $\textbf{Dimensionless Temperature} ~ \mathbf{\theta}$ defined as shown in the picture below :- enter image description here

I am interested in defining $\theta$ that is valid for inhomogenous Dirichlet boundary conditions i.e. when $T_{w1} \neq T_{w2}$ which shall show a periodic behavior in fully developed flow.

P.S. In an actual 3D flow through square duct there would be four faces(or walls) and all can be at different fixed temperature. For brevity purpose I have shown a case here which shows two walls in a 2D system (can be considered to be a flow between parallel plates with adibatic/free slip boundary condition for temperature/flow respectively).

EDIT

So I am looking to extend the implementation of streamwise periodic boundary condition for non-equal wall temperatures. In case of equal wall temperatures the $\theta$ can be defined as depicted in the picture which then can be transcended to the periodic boundary condition requirement for Dirichlet conditions to be

$\theta_{0,y,z}$ = $\theta_{L,y,z}$

where $L$ in the fully developed periodic length.

Substituting for $\theta$ and doing simplification one gets

$T_{0,y,z} = c_1 T_{L,y,z} + c_2$

where

$c_1 = \frac{T_b(0) - T_{wall}}{T_b(L) - T_{wall}} $ and $c_2 = T_{wall} (1 - c_1)$

Now for a constant and equal wall temperature using this formulation gives the right nusselt number values as well as temperature values (fixing $T_b(0)$ to be some reference tempearture).

In the case of two constant but, different wall temperatures we have two possibilities

$T_{wall} = T_{wall1}$ or $T_{wall} = T_{wall2}$ or $T_{wall} = avg(T_{wall1}, T_{wall2}) $

Choosing one among the first two makes the temperature bounded by either of the wall temperatures but not both. I observe then a few cells near the wall2 or wall2 where the temperatures go below the cold wall temperature or hotter than the hot wall temperature (depending on which wall is taken as T_{wall} in the expression the problem occurs at the opposite wall). Choosing the third definition, if we see the expression for $c_1$ it is really possible to have a bulk temperature at the outlet to be that of the average wall temperatures which is then a condition of singularity.

So the question remains how should we define $\theta$ so as to have a consistent representation of periodicity and also having a bounded solution as well.


References

Streamwise periodic boundary conditions for Constant $T_{wall}$

Asymmteric Graez Nusslet problem

$\endgroup$
1
$\begingroup$

You could define $T_w$ as $$T_w=\frac{1}{mes(\partial\Omega)} \int_{\partial \Omega}T_W\,dS$$ Where $T_W$ is the wall temperature and $T_w$ is the average wall temperature.

For a 2D problem with constant wall temperatures, this integral reduces to $$T_w=\frac{1}{2}(T_{W1}+T_{W2})$$ This is so, because in this context $\partial\Omega$ is the upper and lower boundary line of the channel with total length ($mes(\partial\Omega)$) equal to $2L$ and if its temperature is constant the integral over the whole boundary is equal to: $T_{W1}L+T_{W2}L$. Divide them and you will obtain the derired result. For the 3D case it will be straightforward. Now the measure ($mes(\partial\Omega)$) will be the area of a plane, and you will be integrating the wall temperature over this plane...

Or you can simply put $$\theta=T/T_b$$

Edit

What if I say you that the nondimensionalisation with a difference is the same as with a single temperature?

In your definition of $\theta$ change $T\to T+c$, with $c$ arbitrary. You see that this change also results in $T_b\to T_b+c$. If you put $c=T_w$ you will have: $$\theta=T/T_b$$

$\endgroup$
  • $\begingroup$ Well from what I presume to be different about the Temperature as a state variable is that normally in all implementations, temperature is non dimensionalised as a ratio of two differences rather than just a ration of two temperatures (which is different in case of velocity). The non-dimensionalisation should be such that $\theta$ is periodic over the length of representative channel. $\endgroup$ – datapanda Mar 5 '18 at 9:40
  • $\begingroup$ Also, I am a bit confused about the integral that you presented. I am not so deep into maths and hence I don't get the physical interpretation of the integral. Since there are just two discrete different wall temperatures what does the integration actually mean and over which surface $dS$ are you referring to integrate upon. Could you elaborate. Thanks for the reply though. $\endgroup$ – datapanda Mar 5 '18 at 9:43
  • $\begingroup$ See the edit please. $\endgroup$ – HBR Mar 5 '18 at 9:59
  • $\begingroup$ Thanks @HBR for the edit, I will check if the $\theta$ definition provided by you is indeed periodic. What it seems to me is that even if I choose to define $\theta$ as (a) ratio of temperature deferences and (b) ratio of temperatures, there won't be a dependence of boundary condition on the value of $\theta$ (as you depicted in the answer) and it should be periodic in nature. $\endgroup$ – datapanda Mar 5 '18 at 13:22
  • $\begingroup$ returing to this thread after a long while. To give you a perspective the problem I am investigating is asymmetric graetz nusselt problem. Numerically when I define my periodicity with this definition I see unbounded temperatures though the Nusselt number does converge to the value neeeded. On a contrary using my definition if I take $T_w$ to be the hot wall temperature then numerically after long simulation time the temperature becomes unbounded(starts dropping below T cold) and vice versa. $\endgroup$ – datapanda Aug 20 '18 at 19:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.