Learning Forum

[Batuhan Özbay]

First of all, hello, my name is Batuhan Özbay. I am a third-year mechanical engineering student at Sakarya University. Although I have not taken a fluid mechanics course since my second year, I have been working very hard to develop my skills in Ansys Fluent. I know that Ansys Fluent struggles with measuring heat flux and viscous stress. By experimenting with the RNG, Standard, and Realizable models within the k-epsilon turbulence model, I calculated the turbulent Nusselt number between two plates. While the Standard and RNG models provided correct results, the Realizable model showed heat flux higher than normal. Of course, I am not as knowledgeable as you, and maybe I am doing something wrong, but would you have the chance to review it anyway?https://drive.google.com/file/d/1BWUm6M4L_tlDdxnDGIDmL0dsTawAh2nS/view?usp=sharing

[Rob]

Staff aren't permitted to open/download files, so we're reliant on you posting images etc.  

I would be surprised if it was a bug, and if the other two results are correct suspect it's more likely convergence or a model setting somewhere. Can you post some of the images/data so we can comment? 

[Batuhan Özbay]

[Batuhan Özbay]

After solving the process using the above values, the Nusselt value is found to be 72.685. I chose the 'Scalable Wall Function' for the wall function. By changing the mesh sizes and solving with the RNG model at the same Reynolds number, I obtained the following results. I dont use y+ and i choose quadratic mesh. I try 1mm, 2mm, 4mm, 0.5mm, 0.25mm mesh size(menter lecnher method doesnt convergence 0.5mm and 0.25mm)

[Batuhan Özbay]

[Batuhan Özbay]

I called this an error because it was mentioned in the Ansys_Fluent_Theory_guide that the k-epsilon realizable model is newer, as I used the default settings, and it has a fairly simple structure. At the same time, the convergence of the standard and RNG methods led me to this thought. And my geometry and analysis is so simple like that photo.

[Rob]

rke is a similar age to RNG and they generally behave in much the same way. Have a look with Enhanced Wall Treatment and see how it looks. 

[Batuhan Özbay]

okey i tried and i found this result

[Rob]

And convergence? How does the flow field look - in detail not just a number or cursory image. 

[Batuhan Özbay]

okay, this is my convergence graph and my convergence values

[Batuhan Özbay]

And since people misinterpreted Ansys's nusselt reading, I wrote my own equations using "expression" 
Wall Heat Flux(qconstantx100): 

maxVal(Wall Heat Flux)@x100

LengthTemperature(Tx100):

(lengthInt(Velocity u*(Y-0.01[m])*Temperature)@x100/lengthInt(Velocity u*(Y-0.01[m]))@x100)

Nusselt:

(qconstantx100 *L)/(k*(Twall-Tx100))

and as I said before, RNG and Standard gave correct results under the same conditions.

[Rob]

And the flow field & mesh?  80 iterations isn't very many; under the Residual panel "advanced options" set check convergence as off. That'll monitor the residuals but not stop the calculation when the threshold is set. A residual of 1e-7 is very small, so I'd expect a good result but I also don't like models that have run so few iterations. 

[Rob]

And how does the temperature & heat flux graph look for that surface? Using a maximum value can be problematic if the area is tiny. 

[Batuhan Özbay]

Typically, in academic studies, values between 10^-6 and 10^-12 are used. Therefore, I chose 10^-7 for everything except energy, which I set to 10^-12. This is because heat flux can vary significantly. I adjusted my mesh size to 0.125 mm. When writing the expression, we use 'maxVal' for heat flux because this is a two-dimensional geometry, and 'wall heat flux' only passes through the wall. If we were to take the average of the line, it would include the midpoint of the line, which means adding many zeros to the arithmetic average. When looking at it pointwise, it gives us the same value as 'maxVal'. i did the same thing as the rng and standart method I would understand there. would be a difference but it was more than normal.The geometry channel is 20 mm wide and 1100 mm long. The first 600 mm shows the hydrodynamic entry length, and the next 500 mm shows the stable state. The part labeled 50-100 in the tables I previously sent to you indicates the location. And inlet temp 300 K, wall temp 310 K.

[Batuhan Özbay]

 

This is the temp graph

And this is wall heat flux graph(It might be poor at the beginning, but we are analyzing the flow after 600 mm anyway.)

 

[Batuhan Özbay]

My geometry is very long, so if I include the entire flow field, it would be unclear. Therefore, I will exclude the section starting from 600 mm.

[Rob]

And if you plot the three results onto one graph? How does the profile vary across the channel for the three models?