[yamifm0f]

I am using a simple example to see the temperature behaviour in starccm+

my geometry is pipe 1m longe with a a diameter on 0.016 m
the boundary conditions are
**velocity inlet ; velocity is about 0.039 m/s (calculated using field function 0.008 mass flow) and temperature 298.15 K.
**pressure outlet; pressure is 101325.0 Pa (atmosphere pressure) and temperature is also 298.15 K.

In physics
I use single phase liquid water with IAPWS-IF97
I use segregated flow solver and segregated Fluid temperature
gravity is in the flow direction (outlet)
implicit unsteady

run time is for 50 s
with a time step of 0.1
40 Iterations

Temperature values are not as expected using both laminar model and turbulent realizable k-e two-lay (all y+ wall treatment)

based on a simple analytical solution the outlet surface average temperature should be 39.9 C
the analytical solution is (Tout-Tin = (q*L)/(cp*m))
Tin is 25 C, q is 500 w, cp is 4180, m is 0.008, and l is 1m
Re starts at 715 at inlet and ends at around 980 at outlet

This is Laminar case but with laminer as you can see in the attached file the temperature value at outlet increases to about 38.5 C and then falls to 36.4 C
with a jump at the temperature values in the first part of the pipe. the temperature jump is then decrease and becomes linear as it should from the start (the jump is unknown to me)

using turbulent realizable k-e two-lay (all y+ wall treatment) I get better temperature values at outlet but still not the hoped ones (I get 40.7 C). I also get a jump at the temperature values in the first part of the pipe. the temperature jump is then decrease and becomes linear as it should from the start (the jump is unknown to me)

I have tried decreasing the time step to 0.01 and 0.001 but no change and also have tried different mesh but still nothing.

Can you please help me with this?

[flotus1]

2 questions come to mind;
1) are you aiming for a stationary solution, i.e. is your analytical solution for a stationary case? If so, running a transient simulation is just an unnecessary error source.
2) is your numerical setup consistent with the analytical solution? Which other assumptions went into the analytical solution, like fully developed flow etc.

[yamifm0f]

Originally Posted by flotus1

2 questions come to mind;
1) are you aiming for a stationary solution, i.e. is your analytical solution for a stationary case? If so, running a transient simulation is just an unnecessary error source.
2) is your numerical setup consistent with the analytical solution? Which other assumptions went into the analytical solution, like fully developed flow etc.

1) well yes I am aiming for the flow to be steady and the 50s time step is what gets it to that state. This test is aimed for a bigger problem more complicated problem where I add change to inlet values at each time step that is why I am using unsteady. I did run the this in a steady state and the temperatures was even worse the temperature using laminar was 48 degrees and for turbulent was about 44.

2) as I stated in my description I used the equation and values stated and the flow was assumed to be fully developed and has reached steady state.

[flotus1]

Originally Posted by yamifm0f

1) well yes I am aiming for the flow to be steady and the 50s time step is what gets it to that state. [...] I did run the this in a steady state and the temperatures was even worse the temperature using laminar was 48 degrees and for turbulent was about 44.

There is a contradiction here...
First of all, how did you estimate that 50s is enough time to reach steady state?
But more importantly: if your solutions differ between transient and steady-state, the transient solution clearly has not reached steady state yet.

Originally Posted by yamifm0f

2) as I stated in my description I used the equation and values stated and the flow was assumed to be fully developed and has reached steady state.

The easier fix here is this: your analytical solution for the temperature difference is the mass-flow averaged temperature. Not just a plain surface average.
There may be more discrepancies between your numerical setup and the analytical setup though, hard to tell with the limited information at hand.

[yamifm0f]

Originally Posted by flotus1

There is a contradiction here...
First of all, how did you estimate that 50s is enough time to reach steady state?
But more importantly: if your solutions differ between transient and steady-state, the transient solution clearly has not reached steady state yet.

I use a 1-d code and run it with the same setup as starccm+ the solution reaches steady state before 50s a few seconds. I did the same setup in starccm+ and just to make sure I did had the correct time I let it run for twice that 100s but it already reached steady state at 50s like the 1-d code.


The easier fix here is this: your analytical solution for the temperature difference is the mass-flow averaged temperature. Not just a plain surface average.
There may be more discrepancies between your numerical setup and the analytical setup though, hard to tell with the limited information at hand.

I use the surface average values in starccm+ only and the analytical setup is very simple and assumes that the mass flow rate do not change (to simplify the problem) and initial setup used is initial temp and initial power

[yamifm0f]

The easier fix here is this: your analytical solution for the temperature difference is the mass-flow averaged temperature. Not just a plain surface average.
There may be more discrepancies between your numerical setup and the analytical setup though, hard to tell with the limited information at hand.[/QUOTE]

The issue is not with the analytical solution but rather with starccm+. The setup is very simple and should not have this big of an issue and even with a temperature being high or low I can use smaller time step or mesh refinement to improve it. I odd this is that the values did not change with both.

[flotus1]

I did not mean to imply that there was anything wrong with your analytical solution. I very much agree that the issue lies with the numerical setup. It does not capture all the assumptions that went into the analytical solution.
It may seem like a "very simple" setup, but getting everything right requires some thought to be put into the setup. And into post-processing. Try to forget how easy this should be, and re-evaluate what the simulation should look like in order to match the setup.

[yamifm0f]

Originally Posted by flotus1

I did not mean to imply that there was anything wrong with your analytical solution. I very much agree that the issue lies with the numerical setup. It does not capture all the assumptions that went into the analytical solution.
It may seem like a "very simple" setup, but getting everything right requires some thought to be put into the setup. And into post-processing. Try to forget how easy this should be, and re-evaluate what the simulation should look like in order to match the setup.

I understand. I only meant with "very simple" that it should not need that much work as it is not complicated. I went through the setup numerous times and they all match.

I noticed that with the same setup using laminar I get reverse flow that might effect the results but I checked the velocities and pressure on both inlet and outlet and in the inlet the velocity is less than outlet and that was normal. The inlet pressure was higher than that of the outlet and that was also normal.
The reverse flow goes up to middle or the pipe so that is the issue with this case. I have no idea why there is reverse flow and checked my boundaries (velocity inlet, pressure outlet, wall) and everything seems fine to me and there should not be any reverse flow.

for the turbulent there was no reverse flow but the issue seem to lay in the same area first half of the pipe where there is a temperature jump instead of having a liner temperature profile

[yamifm0f]

Originally Posted by flotus1

I did not mean to imply that there was anything wrong with your analytical solution. I very much agree that the issue lies with the numerical setup. It does not capture all the assumptions that went into the analytical solution.
It may seem like a "very simple" setup, but getting everything right requires some thought to be put into the setup. And into post-processing. Try to forget how easy this should be, and re-evaluate what the simulation should look like in order to match the setup.

I java record every simulation so that I can check and recheck everything