|
[Sponsors] |
May 27, 2017, 20:18 |
Transient Variable speed
|
#1 |
Member
jeevan kumar
Join Date: Mar 2009
Posts: 88
Rep Power: 17 |
Hello,
I need to perform Transient analysis in Fluent with variable speed for an aircraft, that is flow velocity changing with time. I have written udf and it works fine, the inlet velocity changes every time step as expected, But the problem is the flow is not developing, there is no change in velocity at outlet and the forces on the aircraft. The velocity only changed near the inlets. Please find attached he velocity contour. I have tried with PBS with different PV coupling schemes, and also DBS. The flow doesn't seams to develop. I tried with time step size of 0.01 and with different itterations/ time step, 20,50 and 100. Is there any mistake in my approach. I really appreciate for any answers/suggestions. Please find below the udf i am using. #include"udf.h" DEFINE_PROFILE(flow_at_inlet,thread,index) { real x[ND_ND],velocity; real y; face_t f; real t; begin_f_loop(f,thread) { F_CENTROID(x,f,thread); y=x[0]; t=RP_Get_Real("flow-time"); F_PROFILE(f,thread,index)=1.5*pow(t,2); velocity=F_PROFILE(f,thread,index); } end_f_loop(f,thread) } Thanks Jeevan |
|
May 27, 2017, 21:02 |
try v const?
|
#2 |
Senior Member
Svetlana Tkachenko
Join Date: Oct 2013
Location: Australia, Sydney
Posts: 407
Rep Power: 14 |
Do you also have the same "flow does not develop" problem if you use a constant velocity at the inlet? If yes, maybe the time step is too small and flow development simply takes time?
Last edited by Светлана; May 28, 2017 at 18:48. Reason: add title |
|
May 27, 2017, 21:17 |
|
#3 |
Member
jeevan kumar
Join Date: Mar 2009
Posts: 88
Rep Power: 17 |
Thanks for the Reply,
With constant velocity also the flow doesn't develop for each time step. If i run more itterations/time step say around 5000 the flow is developing. But i cannot do this with variable speed. Since i have to run for different velocity and the solution need to converge less than 100 itt. I have dont few transiant for internal flows, where i used to get convergence and flow used to develop within 20 itt. |
|
May 28, 2017, 18:47 |
v value? units?
|
#4 |
Senior Member
Svetlana Tkachenko
Join Date: Oct 2013
Location: Australia, Sydney
Posts: 407
Rep Power: 14 |
Thanks for checking. Your constant velocity value takes 5000 iterations to develop. What constant velocity value are you applying?
In your first message, you provided a contour plot with a maximum velocity of about 0.05. It does not have a scale (m) or velocity units (m/s usually but sometimes programs use mm/s and the like). Please provide this information. |
|
May 28, 2017, 21:30 |
|
#5 |
Member
jeevan kumar
Join Date: Mar 2009
Posts: 88
Rep Power: 17 |
I tried with different velocity, like 0.00015, 0.09325 and 0.375. For all the 3 velocities it use to take about Same 5000 itt to converge and for the flow to develop. Yes the velocity contours are in m/s.
Is the flow domain very big for the flow to develop? I also tried with reducing the domain size, buy it diverges. Thanks Jeevan |
|
May 29, 2017, 00:29 |
original 10,000 time steps?
|
#6 |
Senior Member
Svetlana Tkachenko
Join Date: Oct 2013
Location: Australia, Sydney
Posts: 407
Rep Power: 14 |
Then you run the non-constant velocity case for 10,000 iterations and what happens?
|
|
May 29, 2017, 17:09 |
|
#7 |
Member
jeevan kumar
Join Date: Mar 2009
Posts: 88
Rep Power: 17 |
If i run for constant velocity with 10,000 itt, I will get converged solution and the results are as expected.
|
|
May 29, 2017, 22:35 |
i mean nonconst
|
#8 |
Senior Member
Svetlana Tkachenko
Join Date: Oct 2013
Location: Australia, Sydney
Posts: 407
Rep Power: 14 |
Sorry, I was not very clear. I'm proposing to run the non-constant velocity solution for 10,000 steps.
|
|
May 31, 2017, 01:20 |
Summary
|
#9 |
Senior Member
Svetlana Tkachenko
Join Date: Oct 2013
Location: Australia, Sydney
Posts: 407
Rep Power: 14 |
Hello,
Sorry, I kept asking questions in short one by one aiming to get answers and see the picture from there. It probably is not easy to be faced with these questions without an explanation of what is really going on. What I think is happening is simply the flow taking time to develop. As you said - the 0.00015, 0.09325 or 0.375 (presumably m/s) constant velocity flow takes 5000 steps to develop - your time step is 0.01 (presumably seconds) - with non-constant velocity, after _______ iterations the flow becomes as shown on your picture, with a maximum velocity of about 0.06 m/s - the non-constant velocity equation is 1.5 * t^2 - this could happen at 0.2 seconds - this means you probably ran your non-constant simulation for only 20 time steps - 20 is much much less than 5000 which might explain your results provided in the contour plot Please also understand that if the inlet velocity keeps growing very quickly then the inlet velocity will always grow more quickly than the velocity in the rest of the domain and then the flow might actually never finish developing in the sense that you expect. So we have these questions: - what happens if you run the non-constant velocity simulation for 5000 steps? 15000? - is there a reason for using 0.01s time step? what if you try 0.1s? |
|
May 31, 2017, 15:58 |
|
#10 |
Member
jeevan kumar
Join Date: Mar 2009
Posts: 88
Rep Power: 17 |
If I run 5000 steps with constant velocity then basically I will get a steady state solution with so many time steps and my solutions will converge.
I tried different time steps 0.0001,0.001,0.01,0.1,0.25 and 0.5. With higher timestep the solution is diverging basically the change in velocity is very high for higher timestep. So i stick to 0.01. I have done slimler analysis for internal flow for variable velocity and variable pressure. I used to get convergence in lessthat 50 itt and the flow used to develop for each timestep, the domain was much smaller in that cases. May be the domain is very big and fluent in incapable of given same results for big domain. Thanks Jeevan |
|
May 31, 2017, 20:19 |
|
#11 |
Senior Member
Join Date: Dec 2016
Posts: 152
Rep Power: 10 |
You might want to try going back to the basics. What is the overall size of the fluid domain, what does your mesh look like, how fine is it, what are your boundary conditions, what initialization method did you use, and have you tried running a steady-state problem. I think it might be one of these little things simply because you have high velocity around the edges of the domain rather than all throughout.
Sent from my iPhone using CFD Online Forum mobile app |
|
May 31, 2017, 21:58 |
|
#12 |
Member
jeevan kumar
Join Date: Mar 2009
Posts: 88
Rep Power: 17 |
I tried all the possible combinations available in fluent, different solver, different schemes. Mesh is good all sense, skewness is less that 0.85. I have done many aerodynamic analysis, So i am aware the requirements of mesh for fluent. I did run steady state for different velocity and i used to get converged solution within 2000 itt. If it was constant velocity i would have not worried, if i run for enough time step i will get converged solution. But my problem involves velocity changing with time and if i start transiant form my steady state data, the solution changes on the first time step and the flow is not developing and solution not converging for subsequent time steps. As i said i tried all possible combination in fluent, but nothing seams to work, I even tried to reducing the flow domain, kept around 2 times the aircraft length. But still the flow doesn't seams to develop. Only thing remaining is trying it in different CFD solver
|
|
June 1, 2017, 00:11 |
Non-constant please
|
#13 |
Senior Member
Svetlana Tkachenko
Join Date: Oct 2013
Location: Australia, Sydney
Posts: 407
Rep Power: 14 |
What happens if you run the non-constant velocity simulation for 5000 steps? 15000?
|
|
June 1, 2017, 14:58 |
|
#14 |
Member
jeevan kumar
Join Date: Mar 2009
Posts: 88
Rep Power: 17 |
If I run with non-constant velocity with time step size of 0.01 sec for 5000 steps with max iterations per time step of 100. The flow is not developing and the solutions is not converging for each timestep. The mass imbalance is very high(of the order of 1000) and there is hugh difference in inlet and outlet velocity(since the flow is not developing) and the forces are on the aircraft are close to zero.
|
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
AMI speed performance | danny123 | OpenFOAM | 21 | October 24, 2020 04:13 |
Transient heat transfer simulation with variable heat source | rdr | CFX | 3 | July 31, 2015 04:33 |
Input File for transient cavitating flow | Badi | CFX | 13 | July 1, 2013 09:28 |
a transient simulation question | rystokes | CFX | 3 | June 3, 2009 07:44 |
variable fan speed | nilay | FLUENT | 1 | January 30, 2009 07:58 |