Hi CFDKareem, it is a nice tutorial and pretty interesting. Let me ask you a question, in your case, was the mesh of the first simulation equal the second simulation in the points of the profile? If the mesh of the first simulation was twice the size the second simulation you would need to click in the interpolation option, right? Another question is if you know if is possible to write expression in the boundary accordingly with a new coordinate and now global coordinate.
Yes, the two meshes are identical. I would try to maintain consistent meshes when possible, but if you can't the interpolation works well. From my knowledge, Fluent always references the global coordinate system for profiles and UDFs. You will have to manually translate your input profile in reference to the global coordinate system. If you know where the profile is going to be implemented, I try and move the geometry to an ideal location in Spaceclaim to make writing the profile more convenient. For example, if I am writing an inlet profile I will put the bottom of the inlet right at the global coordinate system so the bottom edge starts at 0,0,0.
@@cfdkareem kareem, I really appreciate your answer, thank you. If you don't mind, I would like to ask you another question about DOF in fluent. In case you know the answer, the question is: I would like to now if it's possible to set magnet field in DOF of fluent to interact with magnetic wall. This magnetic field is responsible for rotate a rotor that is set as wall in fluent. The flow of fluid is responsible for increase the rotation and temperature of this wall. Also, the magnetic properties of the wall material change accordingly with the its temperature. So, at first, the inputs of my model are the inlet velocity and temperature, outlet pressure of the flow and the magnetic field that is interacting with the wall which generates rotates. Do you know if it's possible to simulate this in fluent?
@Rau379 it's definitely quite a complex problem! It would be difficult to do through the graphical user interface, but can likely be done with a UDF. The wall motion and properties can be defined using the macros DEFINE_MOTION and DEFINE_PROPERTY. There is no defined macro for defining magnetic fields that would interact with a solid boundary. You will likely have to create a custom function for the wall motion and define it under the motion macro. Check out the Fluent customization manual for more info on UDFs.
Hi Kareem, Thansk for the explanation. It was very helpful. In my case I have more than 50 mm long straight tube, then a tapered (45 degree) section then 2mm long another straight section then the lquid introduced to multiphase system. Where do you think I should cut my nozzle region (before taper after long tube region, after taper and before last straight section or just before the nozzle exit and introduce profile to main simulation) so that I could lower my multiphase calculation time.
Ideally wherever you can cut the domain that would reduce mesh size and you can confidently assume the boundary condition at the point. You could also do a preliminary run with a single phase, and then export the profile right before the nozzle. It all comes down to your own engineering intuition on what will effectively capture your problem.
@@cfdkareem ok thanks a lot. I wasn't sure if I cut whole secondary phase injection (nozzle) and put that velecity profile as b.c. into a meshed face at the bottom of my test section it would simulate exactly the same condition. Like at the first time step would it model the liquid just like the real case(if I cut the whole scondary phase/nozzle) or should I provide some of the last section (after the taper) of the nozzle exit in to my whole simulation. But if the only important parameter is to reduce the computation time then it would be wise to divide it from nozzle exit and introduce whole nozzle exit velocity profile as b.c. to face mesh of nozzle exit. Thanks a lot
Hi! Great video. I noticed the parameters you read into fluent did not have dimensions. Are they in metric? If you change the units in the fluent menu to English units, but you read/import those profiles let's say in SI units, what happens? Is the profile data always have to in SI units ?
All profiles must be in SI units. Fluent always operates under the hood in SI. When you change to English in the GUI it's doing a "visual conversation" well still using SI under the hood.
I'm trying to find the boundary layer thickness value at that line, is it by inserting a chart of variable x velocity and y y axis, i need to match the B.L thickness value with what I have before exporting the profile
Yes you'll want to create a line extending from the wall and plot the x-direction velocity vs y. You can then find the height where the velocity along the wall achieves 99% the free stream velocity.
@@cfdkareem Thank you, and in terms of mesh independence, how should that carried out as we have two separate domains, I mean if my parameter is the reattachment length how would I refine the first domain to export the profile then refine the second one !
Great question! In the first domain your output is the developed velocity profile. This can be calculated analytically so you could compare your simulated profile to the equation. Reducing error in the profile will give the most accurate results for the reattachment. You can then use the profile in the second and analyze the mesh sensitivity for the reattachment point. However, mesh independence mostly comes down to how much accuracy you need from the model. For me personally, in an industrial setting, I would use my engineering judgement for the first mesh, confirm the profile is acceptable, and then perform an analysis on the second, since I'm mostly interested in the reattachment and can tolerate a small amount of error. As with most modeling it's all about using your judgement, experience, and skill to determine the best way to get the information you need!
Hello Mr. Kareem. It’s not clear to me about how you created the velocity profile. did you start from space-claim, and create new geometry and meshing? you just switched to a new video, was it same goemtry as before? can you please explain the background process, some of us are beginner-level. Or is there other ways I can modify the inlet in the primary simulation to get a fully developed flow, I’m not getting this pattern. I tried to create a rectangle of 3900mm and 100mm dimensions and during the meshing it was very slow and frustrating
Hello James, yes the second simulation is created using a full new model including geometry in SpaceClaim, meshing, and a new fluent setup. The second simulation is simply a 2D rectangle with a height equal to the inlet height of the primary simulation (turbulent step). Make sure you are not using a very small mesh size for this simulation. The mesh will still be fairly large, but shouldn't take more than a few minutes to create. You can also do what you mentioned, i.e. extend the inlet of the turbulent step simulation far enough to allow the flow time to become fully developed. The benefit of splitting the simulation is that your primary simulation, the turbulent step, will have a much smaller mesh size and therefore a faster calculation time.
May I know how I can tranfer the output boundary profile from first geometry as input of second geometry for a transient case ? (As since I have columns data ,eg velocity magnitude, in a time series), how can I insert the data in time series so I get transient result from second geometry as well ?
Hello, if your velocity profile does not change throughout your transient simulation, then you can use the same profile. It will assume that the given profile is constant for each time step. If your velocity changes in the XYZ direction with time, this is more complicated. You will either have to create different profiles for each change and then use execute commands to read the new profile at the correct time. This is less than ideal if your profile is constantly changing. If that is the case then you will be better suited using a UDF to define the inlet. Thanks!
@@cfdkareem Thank you Kareem for your reply. Unfortunately, velocity profile for my case keep updating over the time. My case is turbulent in two phase flow, I am unable to define properly by simply giving an equation describing the inlet. 😮💨What I have is a big data matrix arranged in a time series which is ready to insert as input profile for second geometry. I think I should go with the method you mention or with LSTM to complete the steps. Let me know if you have any other ideas for my problem. Thank you so much.
Hi kareem, how can I transfer the volumetric data from the geometry to another fluent case with the same geometry. While writing a profile, only the profile data/ surface data is been taken not the volumetric data. For example, I have a rectangular 3D domain and the internal temperature is simulated. Now I want to give this internal temperature to another problem with the same rectangular 3D domain ( as the initial condition in the second problem). How to do it?
In Fluent you can go to File>export and export just the data file for the solution. In your new simulation you can file>Import the data file only. This will transfer the data from one simulation to another of the same domain. You can also export 3D solution data by going to file>export>solution data and choosing from an array of different file formats.
Hi Kareem, when simulating the flow through the pipe to get velocity values before the backwards step, through looking at the contours it doesn’t seem like its flowing properly as all i can see is blue at the inlet, then there is a small red vertex in the middle and then the flow in the pipe goes blue. Do you know why this might be?
If you have one vertex much higher than the rest of the flow it is usually one of two problems. Most likely there is a bad mesh element that is not converging well. Try remeshing and possibly refining it to see if it improves. The other potential is that the simulation is not fully converged. Try running the simulation for more iterations and see if it converges. If neither of these help them check your boundary conditions and make sure the inlet velocity/material properties are set correctly. Let me know how it works out!
Hey Saikat, in this tutorial we used a secondary simulation to develop the flow profile. We can assume how long of a channel we will need to ensure fully developed flow by calculating the entrance length.
Hi Kareem, I am also trying to create this velocity profile based off the driver and seegmiller paper, however, the values I am getting for the fully developed profile line are constant for turbulent kinetic energy and dissipation rate. I have made sure to use the same dimensions for the geometry, any ideas why this may be? many thanks in advance
Hey George, It's hard to say without some more detail about your setup. Some simple things to check would be that your mesh is refined sufficiently along the walls, that your line you are exporting the values from is defined correctly and far enough away from the inlet, and that your simulation is running to full convergence. If you need more help please send me and email at kareemcfd@gmail.com with more details on your setup or, even better, send me your case files for the problematic simulation. Thanks!
hi Kareem, when I try to export the profile into my original profile, it does not look the same as it did after the initial simulation. My mesh is the same size and i think ive exported the correct values, so unsure why this is happening. thanks in advance
@@cfdkareem thanks for replying so fast! When I load in my data and then run the simulation, the profile does not appear as it did in the secondary sim. For example, the top of the imported profile looks like the step in terms of flow… it’s very odd. I think something to do with the global coordinates?
@@marstokrom1872 That would be my guess! Make sure the Y cordinates match between your inlet in your primary simulation and secondary simulation. If the global zero is at the bottom of the step, the secondary simulation will have to be offset from y=0 by the step height.
Kareem my guy, Your tutorial video is, without a doubt, the best I've ever seen.
Hi Kareem do you offer one-on-one tutorials, if yes, how does one get hold of you?
Hello, I do not provide live 1 on 1 consultation, but you can email (address in channel page) specific questions and I can answer them there. Thanks!
Hi CFDKareem, it is a nice tutorial and pretty interesting. Let me ask you a question, in your case, was the mesh of the first simulation equal the second simulation in the points of the profile? If the mesh of the first simulation was twice the size the second simulation you would need to click in the interpolation option, right?
Another question is if you know if is possible to write expression in the boundary accordingly with a new coordinate and now global coordinate.
Yes, the two meshes are identical. I would try to maintain consistent meshes when possible, but if you can't the interpolation works well. From my knowledge, Fluent always references the global coordinate system for profiles and UDFs. You will have to manually translate your input profile in reference to the global coordinate system. If you know where the profile is going to be implemented, I try and move the geometry to an ideal location in Spaceclaim to make writing the profile more convenient. For example, if I am writing an inlet profile I will put the bottom of the inlet right at the global coordinate system so the bottom edge starts at 0,0,0.
@@cfdkareem kareem, I really appreciate your answer, thank you. If you don't mind, I would like to ask you another question about DOF in fluent. In case you know the answer, the question is: I would like to now if it's possible to set magnet field in DOF of fluent to interact with magnetic wall. This magnetic field is responsible for rotate a rotor that is set as wall in fluent. The flow of fluid is responsible for increase the rotation and temperature of this wall. Also, the magnetic properties of the wall material change accordingly with the its temperature. So, at first, the inputs of my model are the inlet velocity and temperature, outlet pressure of the flow and the magnetic field that is interacting with the wall which generates rotates. Do you know if it's possible to simulate this in fluent?
@Rau379 it's definitely quite a complex problem! It would be difficult to do through the graphical user interface, but can likely be done with a UDF. The wall motion and properties can be defined using the macros DEFINE_MOTION and DEFINE_PROPERTY. There is no defined macro for defining magnetic fields that would interact with a solid boundary. You will likely have to create a custom function for the wall motion and define it under the motion macro. Check out the Fluent customization manual for more info on UDFs.
Hi Kareem,
Thansk for the explanation. It was very helpful. In my case I have more than 50 mm long straight tube, then a tapered (45 degree) section then 2mm long another straight section then the lquid introduced to multiphase system. Where do you think I should cut my nozzle region (before taper after long tube region, after taper and before last straight section or just before the nozzle exit and introduce profile to main simulation) so that I could lower my multiphase calculation time.
Ideally wherever you can cut the domain that would reduce mesh size and you can confidently assume the boundary condition at the point. You could also do a preliminary run with a single phase, and then export the profile right before the nozzle. It all comes down to your own engineering intuition on what will effectively capture your problem.
@@cfdkareem ok thanks a lot. I wasn't sure if I cut whole secondary phase injection (nozzle) and put that velecity profile as b.c. into a meshed face at the bottom of my test section it would simulate exactly the same condition. Like at the first time step would it model the liquid just like the real case(if I cut the whole scondary phase/nozzle) or should I provide some of the last section (after the taper) of the nozzle exit in to my whole simulation. But if the only important parameter is to reduce the computation time then it would be wise to divide it from nozzle exit and introduce whole nozzle exit velocity profile as b.c. to face mesh of nozzle exit.
Thanks a lot
Hi! Great video. I noticed the parameters you read into fluent did not have dimensions. Are they in metric? If you change the units in the fluent menu to English units, but you read/import those profiles let's say in SI units, what happens? Is the profile data always have to in SI units ?
All profiles must be in SI units. Fluent always operates under the hood in SI. When you change to English in the GUI it's doing a "visual conversation" well still using SI under the hood.
I'm trying to find the boundary layer thickness value at that line, is it by inserting a chart of variable x velocity and y y axis, i need to match the B.L thickness value with what I have before exporting the profile
Yes you'll want to create a line extending from the wall and plot the x-direction velocity vs y. You can then find the height where the velocity along the wall achieves 99% the free stream velocity.
@@cfdkareem Thank you, and in terms of mesh independence, how should that carried out as we have two separate domains, I mean if my parameter is the reattachment length how would I refine the first domain to export the profile then refine the second one !
Great question! In the first domain your output is the developed velocity profile. This can be calculated analytically so you could compare your simulated profile to the equation. Reducing error in the profile will give the most accurate results for the reattachment. You can then use the profile in the second and analyze the mesh sensitivity for the reattachment point. However, mesh independence mostly comes down to how much accuracy you need from the model. For me personally, in an industrial setting, I would use my engineering judgement for the first mesh, confirm the profile is acceptable, and then perform an analysis on the second, since I'm mostly interested in the reattachment and can tolerate a small amount of error. As with most modeling it's all about using your judgement, experience, and skill to determine the best way to get the information you need!
Hello Mr. Kareem. It’s not clear to me about how you created the velocity profile. did you start from space-claim, and create new geometry and meshing? you just switched to a new video, was it same goemtry as before? can you please explain the background process, some of us are beginner-level. Or is there other ways I can modify the inlet in the primary simulation to get a fully developed flow, I’m not getting this pattern. I tried to create a rectangle of 3900mm and 100mm dimensions and during the meshing it was very slow and frustrating
Hello James, yes the second simulation is created using a full new model including geometry in SpaceClaim, meshing, and a new fluent setup. The second simulation is simply a 2D rectangle with a height equal to the inlet height of the primary simulation (turbulent step). Make sure you are not using a very small mesh size for this simulation. The mesh will still be fairly large, but shouldn't take more than a few minutes to create. You can also do what you mentioned, i.e. extend the inlet of the turbulent step simulation far enough to allow the flow time to become fully developed. The benefit of splitting the simulation is that your primary simulation, the turbulent step, will have a much smaller mesh size and therefore a faster calculation time.
May I know how I can tranfer the output boundary profile from first geometry as input of second geometry for a transient case ? (As since I have columns data ,eg velocity magnitude, in a time series), how can I insert the data in time series so I get transient result from second geometry as well ?
Hello, if your velocity profile does not change throughout your transient simulation, then you can use the same profile. It will assume that the given profile is constant for each time step. If your velocity changes in the XYZ direction with time, this is more complicated. You will either have to create different profiles for each change and then use execute commands to read the new profile at the correct time. This is less than ideal if your profile is constantly changing. If that is the case then you will be better suited using a UDF to define the inlet.
Thanks!
@@cfdkareem Thank you Kareem for your reply. Unfortunately, velocity profile for my case keep updating over the time. My case is turbulent in two phase flow, I am unable to define properly by simply giving an equation describing the inlet. 😮💨What I have is a big data matrix arranged in a time series which is ready to insert as input profile for second geometry.
I think I should go with the method you mention or with LSTM to complete the steps.
Let me know if you have any other ideas for my problem.
Thank you so much.
Hi kareem, how can I transfer the volumetric data from the geometry to another fluent case with the same geometry. While writing a profile, only the profile data/ surface data is been taken not the volumetric data.
For example, I have a rectangular 3D domain and the internal temperature is simulated. Now I want to give this internal temperature to another problem with the same rectangular 3D domain ( as the initial condition in the second problem). How to do it?
In Fluent you can go to File>export and export just the data file for the solution. In your new simulation you can file>Import the data file only. This will transfer the data from one simulation to another of the same domain. You can also export 3D solution data by going to file>export>solution data and choosing from an array of different file formats.
Hi Kareem, when simulating the flow through the pipe to get velocity values before the backwards step, through looking at the contours it doesn’t seem like its flowing properly as all i can see is blue at the inlet, then there is a small red vertex in the middle and then the flow in the pipe goes blue. Do you know why this might be?
If you have one vertex much higher than the rest of the flow it is usually one of two problems. Most likely there is a bad mesh element that is not converging well. Try remeshing and possibly refining it to see if it improves. The other potential is that the simulation is not fully converged. Try running the simulation for more iterations and see if it converges. If neither of these help them check your boundary conditions and make sure the inlet velocity/material properties are set correctly. Let me know how it works out!
@@cfdkareem I’ll give this a try - thank you for fast response :)
which expression you used to calculate the fully developed length
Hey Saikat, in this tutorial we used a secondary simulation to develop the flow profile. We can assume how long of a channel we will need to ensure fully developed flow by calculating the entrance length.
Hi Kareem. Are you working as researcher using CFD?
Hello Muhammad, yes I am! My research primarily focuses on multiphase simulation of molten metals.
Hi Kareem, I am also trying to create this velocity profile based off the driver and seegmiller paper, however, the values I am getting for the fully developed profile line are constant for turbulent kinetic energy and dissipation rate. I have made sure to use the same dimensions for the geometry, any ideas why this may be? many thanks in advance
Hey George, It's hard to say without some more detail about your setup. Some simple things to check would be that your mesh is refined sufficiently along the walls, that your line you are exporting the values from is defined correctly and far enough away from the inlet, and that your simulation is running to full convergence. If you need more help please send me and email at kareemcfd@gmail.com with more details on your setup or, even better, send me your case files for the problematic simulation. Thanks!
@@cfdkareem Thank you for getting back to me Kareem, I have sent you an email :)
hi Kareem, when I try to export the profile into my original profile, it does not look the same as it did after the initial simulation. My mesh is the same size and i think ive exported the correct values, so unsure why this is happening. thanks in advance
Hey Mars,
Could you explain in a bit more detail what the difference is that you are observing? Thanks
@@cfdkareem thanks for replying so fast! When I load in my data and then run the simulation, the profile does not appear as it did in the secondary sim. For example, the top of the imported profile looks like the step in terms of flow… it’s very odd. I think something to do with the global coordinates?
@@marstokrom1872 That would be my guess! Make sure the Y cordinates match between your inlet in your primary simulation and secondary simulation. If the global zero is at the bottom of the step, the secondary simulation will have to be offset from y=0 by the step height.
@@cfdkareem that worked, thank you!!!