Learning Forum

[OSK1]

Hi there!nI am looking to replicate the model described in the paper:Submicron particle filtration in monolith filters–A modeling and experimental study by Wu et al. 2013. For the model, consisting of a 4um pore and particles ranging from 50nm to 300nm, they implemented the following forces in CFX to simulate capture efficiency:(1) Van der Waals, (2) Brownian motion where they multiplied the equation by a constant, (3) drag, and (4) electrostatic. I am interested to implement the first 3 forces and am looking at either DPM or MPM.nFrom my understanding, DPM simulates particles as a point mass, which is similar to CFX. In MPM, the volume of the particles is taken into consideration. Currently, I feel that DPM would be a better choice since I am able to activate Brownian motion and drag force. However, I would appreciate any suggestions on the following:nCan I modify the Brownian motion equation in-built in DPM? If yes, how should I go about it? How do I write a UDF to modify the in-built equation used in DPM?nIs there an in-built function to account for Van der Waals forces? Else, I believe the only way is to write a UDF on it? If so, is there a step-by-step tutorial on how to write UDF?nIs there any particle tracking test case to check if I have implemented the UDF correctly?nAs for MPM, I am facing the following issues:nWhen using Fluent poly-hexacore meshing, I am unable to get the maximum volume to be what I need:6.54e-22 [m3]. This value is derived by dividing the volume of a 50nm spherical particle by 25 times since it is stated that the particle should be well resolved across the diameter by at least 20-30 fluid cells. Hence, I interpret fluid cells as the volume of cells. (Am I interpreting it correctly?)nDespite setting trap boundary condition on the pore channel, no particles are reported as trap. Is there any test case to aid in learning how the trap BC works?nThank you in advance for any form of assistance!n

[Surya Deb]

Hello, n1.) The Brownian motion is effective for sub micron particles under Laminar Flow conditions. You will also need to use the Stokes-Cunningham Drag Law. You can modify the Cunningham correction to change the Brownian motion force. Please have a look at this link. nhttps://ansyshelp.ansys.com/account/secured?returnurl=/Views/Secured/corp/v190/flu_th/x1-62400016.7.3.html. nAlternatively, you can use DEFINE_DPM_BODY_FORCE UDF to implement the forces and customize accordingly. You will find details of the formulation in the link below:nn2.) For van Der Waals force, it is in beta stage but you will need to turn on Discrete Element Model (DEM) for inter particle collisions. You can find more details in the link below.nn3.) You can create very simple test cases to test your UDFs. Depends on what kind of checks you want to perform with the UDF. The best option would be to validate with published experimental data.4.) Why do you need MPM? As far as I understand you have quite small (sub micron sized) particles. I believe this is why you are interested with Brownian motion as well. In this case, you mention that the particle size is 50nm, right? If you divide this by 25 to 30 cells, then each division might approach the mean free path size of the fluid itself. nYou might want to check this and reconsider why you need MPM for this?.5.) Can you Sample the particles on that surface or plot the tracks nearby to make sure particles cross or touch the surface to get trapped?.Regards,nSuryan

[OSK1]

Hi Surya,nThanks for the links and reply! Indeed, I think DPM is the way to go. I am planning to simulate particle fouling of the pore channel in the future and thought that MPM could be a viable way. n1. Now, I think I should be able to accomplish this with DPM with UDF. Any idea if this is possible or if it is only possible by coupling DPM with DEM (beta stage)?nFor 2. Van der Waals forces, thanks for the information on DEM! I will read up on that. nOn the same point, I would like to implement Van der Waals forces at the pore channel. For now, I think the only way is to use a UDF and implement it as a DPM boundary condition. n2. Do you happen to know how I may call on the Cell Wall Distance value, obtained via Results>Graphics>Particle tracks>Color by Mesh...Cell Wall Distance, that is displayed? Currently, the particles traveling through the pore channel shows a Cell Wall Distance = 1.82e-8 [m]. Based on the paper, the criteria of capture by Van der Waals occurs when the Cell Wall Distance, distance from the particle center to the surface of the filter, is less than 0.7d. For the example case of 50 nm, 0.7d= 3.5e-8 [m]. Hence, I would like the particles with Cell Wall Distance < 3.5e-8 [m] to be described as trap at the pore channel. Would appreciate any guidance on this!nThanks for your assistance!nRegards,nShi Ken

[Amine Ben Hadj Ali]

Point 2) What do you want to do? To kill particles?n

[Rob]

Trap is a wall function, and can't be called in the cells. However, you can terminate a particle based on a criterion and then add a flag/tick to a scalar to tell the surface that a particle is stuck. That's all covered in various parts of the UDF manual. Note, as DPM are point masses the pore won't block so you may need to do something else. n

[OSK1]

For point 1, I tried the first suggestion and realize it does not work since the Cunningham correction (Cc) is used by both the Drag law and the Brownian motion force. By changing the Cc to account for the modification of the Brownian motion, the Drag law is also affected.nAs such, I proceeded with the second suggestion: write a DEFINE_DPM_BODY_FORCE UDF for equation 16-15 found in the theory guide:nAny idea if this is the correct approach? Cause if this is the correct equation to implement, then I believe I have to work on my UDF. The current UDF does not result in the same result as when Brownian motion is selected although I attempted to implement the exact equation. I understand that ANSYS staff are not able to comment on UDF, however, I shall post a snapshot of my UDF here in case anyone can direct me to resources to troubleshoot my UDF. nArray Yes, for point 2, I would like to kill the particles by changing their path to be PATH_ABORT. I did try writing a DEFINE_DPM_BC UDF, however, I realize that this BC is only activated when a particle hits the wall and not prior it. As such, I believe that DEFINE_DPM_BC is not the appropriate way. I am thinking to perhaps use DEFINE_DPM_SCALAR_UPDATE(VDW, c, t, initialize, p) and couple it with BOUNDARY_FACE_GEOMETRY(f,t,A,ds,es,A_by_es,dr0), where ds will be used as an indicator of when to mark the particle as killed. nAny feedback or suggestions on the above 2 points are much appreciated!nThanks!n

[Rob]

Check the documentation, but there is a kill state on particles that can be called based on a value, usually age. n

[Amine Ben Hadj Ali]

Use DPM Scalar Update Macro to kill particles. nBody Force UDF is good but the force is not accounted for when doing two-way coupling. You need to correct that by additional DPM Sources...n

[OSK1]

Thanks for the information, I will check the documentation.nNoted will check on the DPM Scalar Update Macro. In my simulation, I am doing one-way coupling, steady-state simulation, and steady-state particle tracking. In this case, is it necessary to have additional DPM sources?.