Hi everyone!
I created a simulation for a circular duct which is heated by a constant heat flux. Does it make a difference to simulate a 3 dimensional geometry or a 2 dimensional geometry (rectangle: radius*length; rotated around axis)? Are there different results?
I have the following problem right now:
I need to verify that my simulation reaches Nusselt number Nu=4.364 (equals a fully developped thermal flow profile) in the outlet of the duct. To calculate Nu first I need to determine the mixed temperature. There are two methods:
1. Integral from axis to wall over the local Temperature divided by radius OR
2. Volume integral over the local Temperature divided by the basis (radius^2*pi)
Due to the fact that the temperature closer to the wall has a greater influence caused by the higher perimeter than the temperature near the axis, method 2 gives a higher mixed temperature.
BUT the Nusselt number of method 2 is always too high. Only method 1 gives the correct answer although it does not weigh the influence of each local temperature considering the location.
PS: Both methods were tested with a fictional temparture profile of T(r)=r^2. Method 2 is 99.9999% of the analytical solution. Method 1 only about 67%.
I simply don't get it, that a tested method to determine the mixed temperature leads to a wrong result...
I created a simulation for a circular duct which is heated by a constant heat flux. Does it make a difference to simulate a 3 dimensional geometry or a 2 dimensional geometry (rectangle: radius*length; rotated around axis)? Are there different results?
I have the following problem right now:
I need to verify that my simulation reaches Nusselt number Nu=4.364 (equals a fully developped thermal flow profile) in the outlet of the duct. To calculate Nu first I need to determine the mixed temperature. There are two methods:
1. Integral from axis to wall over the local Temperature divided by radius OR
2. Volume integral over the local Temperature divided by the basis (radius^2*pi)
Due to the fact that the temperature closer to the wall has a greater influence caused by the higher perimeter than the temperature near the axis, method 2 gives a higher mixed temperature.
BUT the Nusselt number of method 2 is always too high. Only method 1 gives the correct answer although it does not weigh the influence of each local temperature considering the location.
PS: Both methods were tested with a fictional temparture profile of T(r)=r^2. Method 2 is 99.9999% of the analytical solution. Method 1 only about 67%.
I simply don't get it, that a tested method to determine the mixed temperature leads to a wrong result...