[CFDbeginer]

Hello everyone, I am a beginner in CFD. I am simulating a square pipe and using the following formula to verify one of the simulation results:


1.png

But the calculated result is 16.7% lower than the result calculated through the formula. Increasing the number of grids doesn't seem to have much effect. A literature with the same settings can yield correct results, The shear stress I calculated is also smaller than in the literature (0.016 vs 0.012). I want to know where the problem lies? Perhaps I should modify the k-e model parameters?

Here are my simulation details (no ribs in my case):

4.png
3.jpg
2.png

If necessary, this is the address of that literature:https://doi.org/10.1016/j.ijheatmass...er.2021.121573

Thanks！

[CFDKareem]

I would make sure you are resolving the boundary layer correctly. Check your Y+ value and confirm it is <1. Proper boundary layer resolution is required for good wall boundary conditions.

The empirical formula you gave for skin friction coefficent also assumes a fully developed flow. Based on the size of your domain the flow will not become fully developed inside the region of interest. You can either increase the length of your inlet to make sure the flow is developed in the region of interest. Or, the better way, is to use a boundary profile for the velocity at the inlet that is fully developed. You can use either an analytical equation or a second simulation to develop this profile.

[CFDbeginer]

Wrong press to reply

[CFDbeginer]

Originally Posted by CFDKareem

I would make sure you are resolving the boundary layer correctly. Check your Y+ value and confirm it is <1. Proper boundary layer resolution is required for good wall boundary conditions.

The empirical formula you gave for skin friction coefficent also assumes a fully developed flow. Based on the size of your domain the flow will not become fully developed inside the region of interest. You can either increase the length of your inlet to make sure the flow is developed in the region of interest. Or, the better way, is to use a boundary profile for the velocity at the inlet that is fully developed. You can use either an analytical equation or a second simulation to develop this profile.

Thank you for your reply

In fact, the purpose of simulating smooth square tubes is to obtain a fully developed turbulent velocity and temperature profile for the next step of unsteady simulation of ribbed channels.

I am certain that at the back of the model, Y+<1 (if the current erroneous results can serve as a reference), my model has a length of 5m, and I rely on the velocity distribution of the channel centerline to determine that the flow reaches full development after about 3m. I use a pressure drop of 4-5m to calculate the friction factor.

According to the official tutorial, the velocity distribution in my results is laminar flow, but in the tutorial, Re>4000 indicates turbulence. I would like to know if this is the reason why my pressure drop is lower than normal? If so, what should I do to obtain turbulence data?
Also, I'm sorry that I didn't include the method for calculating the friction coefficient before. The result calculated using the first formula (delta_P) in the attachment is 16.7% less than the theoretical value, while the result calculated using the second formula (τ_w) is relatively close to the theoretical value. However, in a smooth channel, the two should be equal.

centerline pressure.jpgcenterline velocity.pngtutorial.jpgcalculate.png

[CFDKareem]

Originally Posted by CFDbeginer

Thank you for your reply

In fact, the purpose of simulating smooth square tubes is to obtain a fully developed turbulent velocity and temperature profile for the next step of unsteady simulation of ribbed channels.

I am certain that at the back of the model, Y+<1 (if the current erroneous results can serve as a reference), my model has a length of 5m, and I rely on the velocity distribution of the channel centerline to determine that the flow reaches full development after about 3m. I use a pressure drop of 4-5m to calculate the friction factor.

According to the official tutorial, the velocity distribution in my results is laminar flow, but in the tutorial, Re>4000 indicates turbulence. I would like to know if this is the reason why my pressure drop is lower than normal? If so, what should I do to obtain turbulence data?
Also, I'm sorry that I didn't include the method for calculating the friction coefficient before. The result calculated using the first formula (delta_P) in the attachment is 16.7% less than the theoretical value, while the result calculated using the second formula (τ_w) is relatively close to the theoretical value. However, in a smooth channel, the two should be equal.

Attachment 95925Attachment 95926Attachment 95927Attachment 95928

Are you sure your inlet condition is correct? In the paper provided they state the inlet condition is a mass flow inlet at Re = 20000. This would put you well into the turbulent range.

[CFDbeginer]

Originally Posted by CFDKareem

Are you sure your inlet condition is correct? In the paper provided they state the inlet condition is a mass flow inlet at Re = 20000. This would put you well into the turbulent range.

Of course. The mass flow rate calculated using the physical parameters of air at 320K is 0.122075kg/s, 0.122075/(0.5 * 0.125)/1.103=1.771 m/s, which is the velocity at 0m in the figure.