Hi Moein. You did an excellent job on that simulation. I really appreciate your innovative thinking. I am pretty sure Your commitment to excellence has inspired others and your attention to detail puts you at the top. Good luck.
Hoy Moein I have another question, in particle tracing model, i try to change the dimeter of the particle but it doesn't change. i mean either with 6um or 100um there is no change in size, at least observable change. thanks Moien
Thanks a lot for this video series Mr. Naderi! It was really helpful for me as I gained a lot of interesting insights about simulating inertial focusing. I was doing a project for sorting and separating microplastics from seawater using the concept of inertial focusing in microfluidics. Unfortunately, my university has a license only for ANSYS and not for COMSOL. I tried simulating the same in ANSYS Fluent using a spiral microchannel, but due to the absence of the empirical wall lift force model in Fluent, the particles failed to focus laterally to a single stream near the inner channel wall when observed from the outlet. Anyway, thanks for this video as I got a lot of useful information from this video series.
Thanks a lot bro. You helped me a lot in my simulation. I was wondering how can I contact you. I did exactly what you did but the trajectory colors of my particles are the same (red) and I don't know hot to change them. Could you help me?
Even tho I have the wall bounce condition, most particle get stuck on the walls. It's frustrating, I can't refine the mesh much more because of my cpu, and I'm using a high discretization for the fluids (P2+P2). This particle tracing module just seems like trouble to me 😩
Thanks a lot mr.naderi. I have a question about the memory that this kinds of simulations need. Actually i have a model with 4 rows of spiral, and RAM is 40 GB. But unfortunately i accost with out of memory error and can't have more than 500 meshes. It is when some articles did such this simulation with 1 row of spiral and over 3milions meshes and just 16GB RAM. I can't even do this problem too. How can i solve this problem?
Good day, I have simulated your tutorial with positive results, however, my output gives the trajectory of the particles in low-resolution circles. It looks as if the time steps are too small so it has large straight lines instead of a circular trajectory like yours. I used your exact increments of 0.01s but I do not get such a well-rounded trajectory. Also thank you so much for this tutorial it is absolutely brilliant.
Thanks a lot for the fantastic video! Because of your video, I simulated a good result. But in the actual chip, large-sized cells flow along the inner wall, but at the fork of the outlet, they disperse and go out from the two outlets respectively. What may be the reason?
Hey, not sure if you can help but lately I have been doing some simulations very similar to this one. The problem is that the smaller particle instead of migrating to the outer wall, migrates to the inner wall. In theory this doesn't make sense, the smaller the particle, the further it migrates from the inner wall. Was wondering if you had gone through something similar. I have a good mesh (normal physics controlled), good discretization (P2+P2), made your tutorial several times and always get similar results. Only when I decrease the velocity (for instance 50 ml/min) the particles present expected trajectories. Do you have any idea what can be wrong?
Hi... What are your channel dimensions? Dean number? particle radius? It all matters. Maybe both particle sizes are significantly large and they equilibrate near the inner wall.
@@moeinpoi Hi, thanks for replying. I'm replicating the spiral in this video ruclips.net/video/GFV89cOAg9c/видео.html The spiral has five turns with channel dimension of 300 um width and 50 um height. Dimensions of particles are 5 um, 20 um and 40 um. For a flow of 100 ul/min or more, both 5 um and 40 um particles equilibrate near the inner wall while the 20 um particle is the furthest away. For 50 ul/min, 40 um particle equilibrates closer to the inner wall following the 20 um and then the 5 um, as expected. I don't know why the smallest particle equilibrates in the inner wall for velocities higher than 50 ul/min. It doesn't make sense to me, probably a higher velocity might be inducing instabilities in the simulation idk.
@@moeinpoi I think I solved the problem. I changed the dimensions of the particles to 20, 30 and 40 um and added a multyphysics that coupled the laminar flow and particle tracing. Not sure why it worked but at least I'm getting good results now.
Hi Moein, your tutorial is very useful.I'm currently doing some research, but the COMSOL simulation results are not quite right. I'm looking forward to your next video tutorial.Thank you!!!!!!!!!!!!
Hello Moein. You have done an incredible job, this really helped me. But can you tell me why I might not observe dean migration in a spiral channel with smaller cross section. I have kept both the Re and De number same FYI. Thanks again.
Hi Mohammad, In spiral channels, you have to define a cut plane that is exactly perpendicular to the flow in order to be able to visualize the dean flow, otherwise (by using predefined planes) you can't observe the vortices.
Thank you for the detailed video I have tried to simulate a particles trajectories that are charged ans soumised to a drag ad electric forces in a laminar flow in which the fluid is air But in the results the particles aren't moving during the time and their position is at the inlet can you help me please with this as soon as possible
Hi, you're welcome! So, to do that, you have to go to Results node > Particle Trajectories(fpt), and then right-click on Particle Trajectories 1 and choose Color Expression. From there, find and select Particle Radius as Expression and it will assign each particle size a different color.
Actually, I have not tried that. I am not sure whether or not we can define the mechanical properties of particles in COMSOL. However, it might be possible by creating a dynamic or moving mesh for the particles. Again, I am not sure if this approach is applicable in this case.
I'm using version 5.4... I think the lift force was introduced in a recent version. I'm not sure exactly which version though. I'd recommend you to update your COMSOL to the latest version in order to use the lift force options.
Hi Moein. As a beginner this was very helpful. May i know how to get velocity profile along a particular cross section area within the spiral channel. Thanks in advance.
Hi, I'm glad it was helpful. To get the velocity profile along some particular cross-section, you can first right-click on the 'Results' node and add a '3D Plot Group' (choose the 'Dataset' according to the study you are looking for). Then, right-click on the '3D Plot Group' and select 'Slice'. In the settings for your 'Slice' plot, locate 'Plane type' and change it from 'Quick' to 'General'. Here you can define the plane on which you would like to have velocity profile (However, it might be hard to define such a plane if you have a complex geometry). Then you can change the 'Expression' to plot velocity profile or whatever parameter you're looking for.
Many thanks for your tutorials. However, I think you made a mistake when you said that the shear gradient force is "embedded" in the velocity field. I believe you ahve to add that separately as a lift force and choose Saffman's equation. Also, in my model COMSOL don't seem to consider Drag force for some reason.. still trying to figure it out.
You're most welcome. To my understanding, Saffman is just a rotation-induced lift force, and is rather negligible in curved geometries, because of its relatively smaller magnitude in comparison with the wall-induced, shear-gradient, and the Dean drag forces. However, what I have recently found out is that the shear-gradient lift force is actually 'embedded' in the wall-Induced lift force. In fact, Comsol has defined the 'net inertial lift force' equation which is the 'superposition' of the wall-induced and the shear-gradient forces - and for some unknown reason to me - have just called it the 'wall-induced' lift force without mentioning that it indeed calculates the net effect of the two mentioned lift forces. On the other hand, I think I might have mentioned that the Dean drag force is embedded in the 'Drag Force' rather than in the velocity field. In fact, we will have lateral components of the velocity field in the direction perpendicular to the wall ( Dean vortex) in curved geometries like spirals, and the Dean drag force is simply calculated and exerted on the particles using the Drag law ( in this case, the Stokes law) that we add in the Particle Tracing physics.
@@moeinpoi Thanks for your explanation. Could you please give a reference to the informtion that says that the wall-induced lift force is actually the superposition of the wall-induced and shear gradient forces ? That would be extremly helpful. Also, is there any way to give each particle type a different colour in the plots ? I see yours appreaing in 3 different colours.
@@3dsquaduk The formula for the net inertial lift force (F wall + F shear gradient) was proposed by Ho and Leal in their 1974 paper: "Inertial migration of rigid spheres in two-dimensional unidirectional flows". This is the formula: FLift=Fw+Fs=ρU2maxa4/D2h(β2G1+βγG2) If you check the formula equation for the wall-induced lift force in COMSOL, you'll see the exact same formula. For the colors of the particles, you can simply go to the "color expression" section in the particle trajectories results, and set it so that it represents the radius of the particles. I think its default mode is set to "particle velocity".
@@moeinpoi Actually, this is not true. Saffman force arises not because the particle rotates but due to the shear rate caused by the relative speed between the fluid and the particle. Depending whether the particle leads the flow or lags behind the flow, a force will move the particle either toward the wall or the center. And this should not be confused with shear gradient lift force which is a consequence of the parabolic velocity profile in Poiseuille flows and always acts toward the wall.
Fantastic tutorial! I tried this step by step in COMSOL 5.4 but the particles trajectories are quite unstable. I even refined the mesh but it just doesn't work. I don't know what possibly could be wrong since i followed your tutorial very closely.
hi! fast reply! I started again a new simulation and turns out now the particles trajectory is fine. Did you do any post processing on that particle trajectory at the end? My particles size is way bigger than their real size. i adjusted the ratio scale factor to 1 but they disappear. thank you EDIT: when i said the trajectories were unstainble i meant that they didnt have that laminar behavior. the would pass through the wall and make abrupt changes.
@@vitorhugoRH Well, you'd have to play with the 'Radius scale factor' under Point Style until your particles are the size that you want. For this case, something between 0.02 to 0.009 would work. But it might be different case by case. You can change it by trial and error until you get to the value you want.
@@vitorhugoRH Hi Vitor, my particle trajectories are not laminar, pass through the wall. When you said you started a new simulation, what exactly did you do? Did you go for a better mesh?
Thanks! I'm glad it was helpful. I have actually recently uploaded a video titled 'Visualizing Dean Flow Vortices Arrow Plot' which might also be helpful to you. Here's the link to it: ruclips.net/video/W_WRiO-MsYE/видео.html I will try to upload on various topics as well. Do let me know if there is a specific topic you'd like me to cover.
Hi Moein. You did an excellent job on that simulation. I really appreciate your innovative thinking. I am pretty sure Your commitment to excellence has inspired others and your attention to detail puts you at the top. Good luck.
Thank you very much!
Moein, i wait for the next video. good luck
My particle dont move at all from the inlet. My geometry is different, converging diverging. What could be the issue?
Can I get an explanation, why my particles are stuck at the beginning itself.
Thx bro. I should see this video early. Saved my project bro!
No problem! I'm glad it was helpful.
hoy Moein, in Drag force part you click the " wall corrections" , why? what does that mean? thanks Moein.
Hoy Moein I have another question, in particle tracing model, i try to change the dimeter of the particle but it doesn't change. i mean either with 6um or 100um there is no change in size, at least observable change. thanks Moien
Thanks a lot for this video series Mr. Naderi! It was really helpful for me as I gained a lot of interesting insights about simulating inertial focusing. I was doing a project for sorting and separating microplastics from seawater using the concept of inertial focusing in microfluidics. Unfortunately, my university has a license only for ANSYS and not for COMSOL. I tried simulating the same in ANSYS Fluent using a spiral microchannel, but due to the absence of the empirical wall lift force model in Fluent, the particles failed to focus laterally to a single stream near the inner channel wall when observed from the outlet. Anyway, thanks for this video as I got a lot of useful information from this video series.
Hi Naveen, I'm glad it was helpful!
Hope you succeed in your research.
Thanks a lot bro. You helped me a lot in my simulation. I was wondering how can I contact you. I did exactly what you did but the trajectory colors of my particles are the same (red) and I don't know hot to change them. Could you help me?
Even tho I have the wall bounce condition, most particle get stuck on the walls. It's frustrating, I can't refine the mesh much more because of my cpu, and I'm using a high discretization for the fluids (P2+P2). This particle tracing module just seems like trouble to me 😩
Thanks a lot mr.naderi.
I have a question about the memory that this kinds of simulations need.
Actually i have a model with 4 rows of spiral, and RAM is 40 GB. But unfortunately i accost with out of memory error and can't have more than 500 meshes.
It is when some articles did such this simulation with 1 row of spiral and over 3milions meshes and just 16GB RAM. I can't even do this problem too.
How can i solve this problem?
can we do this in 2d GEOMETRY...will there be any effect in results as 1 pair of walls in lift force will not be there..?
Good day, I have simulated your tutorial with positive results, however, my output gives the trajectory of the particles in low-resolution circles. It looks as if the time steps are too small so it has large straight lines instead of a circular trajectory like yours. I used your exact increments of 0.01s but I do not get such a well-rounded trajectory. Also thank you so much for this tutorial it is absolutely brilliant.
You're welcome.
So, have you tried decreasing the timesteps? Like use 0.001 instead of 0.01
Has your problem been solved? I also encountered the same problem.Can you give me some advice?
Thanks a lot for the fantastic video! Because of your video, I simulated a good result. But in the actual chip, large-sized cells flow along the inner wall, but at the fork of the outlet, they disperse and go out from the two outlets respectively. What may be the reason?
您好，可以分享这个三位模型吗，我也想在comsol 里算算，谢谢🙏
Thats great tutorial brother.. I appreciate your effort 👏🏼👏🏼👏🏼. Subscribed Hope to see such more content
Glad it was helpful.
Thanks a lot for the fantastic video! What a great tutorial that I have ever seen!!
its great i have question about magnetic force acting on the particles
I actually haven't worked on magnetophoresis myself. All I know is that you can also add it to the lift forces in the particle tracing module.
@@moeinpoi am stack in this problem do you recommend some tips for m
@@aliumara1205 What exactly is your problem?
Moein Naderi i cant make the particles response to external magnetic field
Hey, not sure if you can help but lately I have been doing some simulations very similar to this one. The problem is that the smaller particle instead of migrating to the outer wall, migrates to the inner wall. In theory this doesn't make sense, the smaller the particle, the further it migrates from the inner wall. Was wondering if you had gone through something similar. I have a good mesh (normal physics controlled), good discretization (P2+P2), made your tutorial several times and always get similar results. Only when I decrease the velocity (for instance 50 ml/min) the particles present expected trajectories. Do you have any idea what can be wrong?
Hi... What are your channel dimensions? Dean number? particle radius? It all matters. Maybe both particle sizes are significantly large and they equilibrate near the inner wall.
@@moeinpoi Hi, thanks for replying. I'm replicating the spiral in this video ruclips.net/video/GFV89cOAg9c/видео.html
The spiral has five turns with channel dimension of 300 um width and 50 um height. Dimensions of particles are 5 um, 20 um and 40 um. For a flow of 100 ul/min or more, both 5 um and 40 um particles equilibrate near the inner wall while the 20 um particle is the furthest away. For 50 ul/min, 40 um particle equilibrates closer to the inner wall following the 20 um and then the 5 um, as expected. I don't know why the smallest particle equilibrates in the inner wall for velocities higher than 50 ul/min. It doesn't make sense to me, probably a higher velocity might be inducing instabilities in the simulation idk.
@@moeinpoi I think I solved the problem. I changed the dimensions of the particles to 20, 30 and 40 um and added a multyphysics that coupled the laminar flow and particle tracing. Not sure why it worked but at least I'm getting good results now.
Hi Moein, your tutorial is very useful.I'm currently doing some research, but the COMSOL simulation results are not quite right. I'm looking forward to your next video tutorial.Thank you!!!!!!!!!!!!
Hi! It's great to hear that. I hope you get the results you want.
Hello Moein. You have done an incredible job, this really helped me. But can you tell me why I might not observe dean migration in a spiral channel with smaller cross section. I have kept both the Re and De number same FYI. Thanks again.
Hi Mohammad,
In spiral channels, you have to define a cut plane that is exactly perpendicular to the flow in order to be able to visualize the dean flow, otherwise (by using predefined planes) you can't observe the vortices.
Thank you for the detailed video
I have tried to simulate a particles trajectories that are charged ans soumised to a drag ad electric forces in a laminar flow in which the fluid is air
But in the results the particles aren't moving during the time and their position is at the inlet can you help me please with this as soon as possible
Hi, you're welcome!
Are you sure you have correctly defined your boundary conditions?
@@moeinpoi yes i did i am quite sure
@@charifaberkanekrachai1558 Are you sure you're looking at the results in the last time-step? Maybe you're looking at the results for t=0.
Hello! Will using the default P1+P1 discretization affect the final equilibrium positions greatly?
the same question. i didn't get that point
Hi, thanks for the tutorial. Just wanted to know how you are able to tell which particle correspond to which streamline
Hi, you're welcome!
So, to do that, you have to go to Results node > Particle Trajectories(fpt), and then right-click on Particle Trajectories 1 and choose Color Expression. From there, find and select Particle Radius as Expression and it will assign each particle size a different color.
Can we make the mechanical properties like young's module and ... for the particles? for example cancer cells are softer than RBC.
Actually, I have not tried that. I am not sure whether or not we can define the mechanical properties of particles in COMSOL. However, it might be possible by creating a dynamic or moving mesh for the particles. Again, I am not sure if this approach is applicable in this case.
What version of COMSOL multiphysics that you used? because mine does not have the lift force feature
I'm using version 5.4... I think the lift force was introduced in a recent version. I'm not sure exactly which version though. I'd recommend you to update your COMSOL to the latest version in order to use the lift force options.
Hey moem...make dean velocity contour plot please
Hi Moein. As a beginner this was very helpful. May i know how to get velocity profile along a particular cross section area within the spiral channel. Thanks in advance.
Hi, I'm glad it was helpful. To get the velocity profile along some particular cross-section, you can first right-click on the 'Results' node and add a '3D Plot Group' (choose the 'Dataset' according to the study you are looking for). Then, right-click on the '3D Plot Group' and select 'Slice'. In the settings for your 'Slice' plot, locate 'Plane type' and change it from 'Quick' to 'General'. Here you can define the plane on which you would like to have velocity profile (However, it might be hard to define such a plane if you have a complex geometry). Then you can change the 'Expression' to plot velocity profile or whatever parameter you're looking for.
Many thanks for your tutorials. However, I think you made a mistake when you said that the shear gradient force is "embedded" in the velocity field. I believe you ahve to add that separately as a lift force and choose Saffman's equation. Also, in my model COMSOL don't seem to consider Drag force for some reason.. still trying to figure it out.
You're most welcome.
To my understanding, Saffman is just a rotation-induced lift force, and is rather negligible in curved geometries, because of its relatively smaller magnitude in comparison with the wall-induced, shear-gradient, and the Dean drag forces. However, what I have recently found out is that the shear-gradient lift force is actually 'embedded' in the wall-Induced lift force. In fact, Comsol has defined the 'net inertial lift force' equation which is the 'superposition' of the wall-induced and the shear-gradient forces - and for some unknown reason to me - have just called it the 'wall-induced' lift force without mentioning that it indeed calculates the net effect of the two mentioned lift forces.
On the other hand, I think I might have mentioned that the Dean drag force is embedded in the 'Drag Force' rather than in the velocity field. In fact, we will have lateral components of the velocity field in the direction perpendicular to the wall ( Dean vortex) in curved geometries like spirals, and the Dean drag force is simply calculated and exerted on the particles using the Drag law ( in this case, the Stokes law) that we add in the Particle Tracing physics.
@@moeinpoi Thanks for your explanation. Could you please give a reference to the informtion that says that the wall-induced lift force is actually the superposition of the wall-induced and shear gradient forces ? That would be extremly helpful. Also, is there any way to give each particle type a different colour in the plots ? I see yours appreaing in 3 different colours.
@@3dsquaduk The formula for the net inertial lift force (F wall + F shear gradient) was proposed by Ho and Leal in their
1974 paper: "Inertial migration of rigid spheres in two-dimensional unidirectional flows". This is the formula: FLift=Fw+Fs=ρU2maxa4/D2h(β2G1+βγG2)
If you check the formula equation for the wall-induced lift force in COMSOL, you'll see the exact same formula.
For the colors of the particles, you can simply go to the "color expression" section in the particle trajectories results, and set it so that it represents the radius of the particles. I think its default mode is set to "particle velocity".
@@moeinpoi Actually, this is not true. Saffman force arises not because the particle rotates but due to the shear rate caused by the relative speed between the fluid and the particle. Depending whether the particle leads the flow or lags behind the flow, a force will move the particle either toward the wall or the center. And this should not be confused with shear gradient lift force which is a consequence of the parabolic velocity profile in Poiseuille flows and always acts toward the wall.
thanks a lot for this video 🙏
Fantastic tutorial! I tried this step by step in COMSOL 5.4 but the particles trajectories are quite unstable. I even refined the mesh but it just doesn't work. I don't know what possibly could be wrong since i followed your tutorial very closely.
Hi! What exactly do you mean by the particle trajectories being 'unstable'?
hi! fast reply! I started again a new simulation and turns out now the particles trajectory is fine. Did you do any post processing on that particle trajectory at the end? My particles size is way bigger than their real size. i adjusted the ratio scale factor to 1 but they disappear. thank you
EDIT: when i said the trajectories were unstainble i meant that they didnt have that laminar behavior. the would pass through the wall and make abrupt changes.
@@vitorhugoRH Well, you'd have to play with the 'Radius scale factor' under Point Style until your particles are the size that you want. For this case, something between 0.02 to 0.009 would work. But it might be different case by case. You can change it by trial and error until you get to the value you want.
@@vitorhugoRH Hi Vitor, my particle trajectories are not laminar, pass through the wall. When you said you started a new simulation, what exactly did you do? Did you go for a better mesh?
Hello Praneetha,
I think you should decrease your time-steps.
Great Work Moein... Your video was very helpful for my understanding. What will be the next topic for your video if I may ask?
Thank you. God Bless
Thanks! I'm glad it was helpful.
I have actually recently uploaded a video titled 'Visualizing Dean Flow Vortices Arrow Plot' which might also be helpful to you. Here's the link to it:
ruclips.net/video/W_WRiO-MsYE/видео.html
I will try to upload on various topics as well. Do let me know if there is a specific topic you'd like me to cover.
thanks very much you're the best
You can help me in my M.tech project.
Is that a question or a fact? 😃