Fluids

Fluids

Surface tension issues

    • soloviev
      Subscriber

      I am running a simulation of oil droplets (high viscosity-machine oil and low viscosity-crude oil).. The oil droplets were injected into seawater about 20 cm below the surface. When they surfaced we then observed their evolution. I have added a figure to show what happens. The black lines are what we observed in the laboratory, the red lines are what we observed with Fluent.  In both cases, the oil droplet would surface, expand a little, then contract. Then after about 0.5 to 1.5 seconds, the oil droplet would rapidly double in size. The simulations would never see this effect. I have added the information below with all relevant information. The surface tensions were the continuum surface stress (CSS) with implicit body forces.  Any ideas on what might be wrong?


      Air density = 1.225 kg m-3


      Air viscosity = 1.7894 10-5 kg m-1 s-1


      Water density = 1021.5 kg m-3


      Dynamic water viscosity = 0.00103 kg m-1 s-1


      Crude oil density = 876.5 kg m-3


      Crude oil dynamic viscosity = 0.01 kg/ms


      Machine oil density 889.0 kg m-3


      Machine oil dynamic viscosity = 0.40 kg/ms


      Water-air surface tension = 0.072 N/m


      Crude oil - air surface tension = 0.028 N/m


      Machine oil - air surface tension = 0.032 N/m


      Crude oil - water surface tension = 0.019 N/m


      Machine oil - water surface tension = 0.040 N/m


       

    • Karthik R
      Administrator

      Hello,


      Could you please elaborate a little on your model? How are you setting it up? How is your grid? What is your time-step? Perhaps attach a few screenshots in your post.


      Thank you.


      Best Regards,


      Karthik

    • soloviev
      Subscriber

      Hello Karthik,


      The settings for this model are the same as in Soloviev et al. 2016. The only difference is that the oil droplets are released from below instead of from above as in the paper. In this published work, we also experienced difficulties with oil droplet spreading.


      Soloviev, A.V., B.K. Haus, M.G. McGauley, C.W. Dean, D.G. Ortiz-Suslow, N.J.M. Laxague, T.M. Özgökmen, 2016: Surface dynamics of crude and weathered oil in the presence of dispersants: Laboratory experiment and numerical simulation. Journal of Geophysical Research–Oceans 121, 3502-3516.


      Link: https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/2015JC011533


      Thank you,


      Alex

    • Rob
      Ansys Employee

      Please can you post some images of the mesh and set up. A contour slice through the droplet & water surface with node values off would also be useful. 


      Thanks,
      Rob

    • soloviev
      Subscriber

      Attached are some images of the mesh and set up. The first shows the domain in a 2D projection (it is a 3D simulation). The blue is water and the white is air with the mesh shown on the wireframe. The mesh resolution is 0.45 mm in the horizontal and vertical. The initial oil droplet size was 3.2 mm. The second image is a contour plot of density on the center plane. This shows all three phases as the oil droplet hits the water surface. The third image is zoomed into the oil droplet.


       


      Regards,


      Alex

    • soloviev
      Subscriber

      Do you require any other information? We are trying to solve this issue before we submit a paper for publication.

    • DrAmine
      Ansys Employee

      You wrote the oil droplet will double in size: This is indicating for me that some compressibility effects need to be taken into account: an incompressible droplet miht breakup into child drops or coalescence with other droplets and might gain mass due to some reactions or interphase processes. All theses phenomena are not relevant for you. So the only one part which would let the the droplet growth or shrinks is either compressibility and pressure distribution or poor numerics.


       


       

    • soloviev
      Subscriber

      This probably is not a compressibility effect. In our previous experiment published in the JGR paper (see link above), I also noticed this same problem. I assume that this is related to the line tension coefficient at the three-phase contact line. The literature review conducted in the JGR paper showed that this coefficient is poorly known, ranging over several orders of magnitude or even becoming negative. It is not clear how this line tension coefficient is implemented in Fluent multiphase. I discussed this issue with ANSYS customer service about 2 years ago, and they mentioned that it would be fixed in the next version (V16.0). Now we are on V19.1, and the issue doesn't seem to have been addressed. 


      Best regards,


      Alex

    • DrAmine
      Ansys Employee

      Tell me the driving force which would force a droplet to grow without reactions and mass transfer?

    • soloviev
      Subscriber

      After the initial moment of the oil drop contact with the water surface, a triple junction surface of oil, water, and air phases occurs assuming that the liquid phases are immiscible [see e.g., Miksis and Vanden-Broeck2001]. The triple junction is a point in two dimensions and a line in three dimensions. The surface tension forces at the triple junction influence the process of oil drop spreading on the water surface.


      For more details please see the Introduction section in Soloviev, A.V., B.K. Haus, M.G. McGauley, C.W. Dean, D.G. Ortiz-Suslow, N.J.M. Laxague, T.M. Özgökmen, 2016: Surface dynamics of crude and weathered oil in the presence of dispersants: Laboratory experiment and numerical simulation. Journal of Geophysical Research–Oceans 121, 3502-3516.


      Link: https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/2015JC011533


       


      Best regards,


      Alex

    • DrAmine
      Ansys Employee

      The droplet or bubble might spread and deform but the mass remains constant until it breakup to child particles. The driving force for me is surface tension duality with pressure. 


      Why I was talking about compressiblity: Consider a bubble rising up in quiescent water. The bubble needs to change its volume as the pressure outside is changing. This can only be done if a density dependency on pressure is fulfilled as EOS. So expansion and compression require a high end EOS.


      Let's stick to this problem: I guess you require here a proper force balance. If you ran the case only in 2D planar then you have at the intiali stage a cylinder and not a droplet. Force balance can be obtained only in 3D or at least 2D axissymmetric (if the geo and flow allows that which is by far not your case).  

    • soloviev
      Subscriber

      Does anyone know how the line tension at the triple junction line is implemented in the multi-phase Fluent model? It might be the cause of the difference between the model and our lab observations of oil drop dynamics at the air-water-oil interface.


      Thanks,


      Alex


       

    • DrAmine
      Ansys Employee

      The surface tension implemented in Fluent is implemented as pair (Volumetric force CSF or Stress CSS). Please carry out a 3D analysis.

    • soloviev
      Subscriber

      We do run 3D simulations...


      Amirfazli and Neumann (2004) (and others referenced in this paper) have shown that the triple line tension can be a factor in sessile lens dynamics.There is no information about the triple line tension in the 19.1 or previous Theory Guides. The Volumetric force CSF or Stress CSS are apparently for the interfacial tension, not for the line tension. Is the triple line tension accounted for in the Fluent Multiphase Model? If not, can we develop a UDF to enter this line tension in the model?


      Reference: Amirfazli, A., and A. W. Neumann (2004), Status of the three-phase line tension, Adv. Colloid Interface Sci., 110, 121–141.


      Thank you,


      Alex


       


       


       

    • Karthik R
      Administrator

      Hello,


      The VOF approach in Fluent uses surface tension (interfacial tension). It is capable of handling contact dynamics near the wall using what is referred to as the wall adhesion approach. Users can specify contact angles - either static or dynamic (UDF required) to model the movement of the solid-liquid-gas contact line. You are asking about liquid-liquid-gas contact line. 


      Having said this, I do not completely understand why you feel that line tension is the missing piece in your analysis, as long as the forces around your oil drop are conserved. If I understand your problem correct, shouldn't the pressure difference across your drop interface be balanced by the surface tension forces (which eventually results in a certain curvature)? Anyways, I hope this helps.


      Btw, I could not open the review paper you shared in the previous post as we do not have access to it. I'd have loved to read more about it.


      Thank you.


      Best Regards,


      Karthik 

    • soloviev
      Subscriber

      Hi Karthik,


      The paper of Amirfazli, A., and A. W. Neumann (2004), Status of the three-phase line tension, Adv. Colloid Interface Sci., 110, 121–141


      is available at 


      https://ac.els-cdn.com/S0001868604000417/1-s2.0-S0001868604000417-main.pdf?_tid=9407cf6b-9379-4aea-96e0-2353751ba56a&acdnat=1534910468_585e24388314db31ba65bd0fe8709c3a


      t can also be found just by googling....


      Best,


      Alex


       

    • soloviev
      Subscriber

      I still would like to have an answer on the original question: Does the Fluent Multiphase model account for the three-phase line tension?


      Thanks,


      Alex


       


       

    • DrAmine
      Ansys Employee

      I will get back to you asap.

    • DrAmine
      Ansys Employee

      Solid-Fluid-Fluid system is dealt by providing contact angle at solid boundary and fluid-fluid surface tension coeff. Fluid-Fluid-Fluid system requires three values of surface tension coefficients. Intrinsic contact angle is the output which Fluent should ideally predict based on Force balance during equilibrium. However, non-zero spurious currents at equilibrium might affect true behavior.


       We do not have the paper you are referencing. We can just see what it has been summarized in the abstract.


       

    • soloviev
      Subscriber

      I have attached a more recent reprint on the subject of line tension. There is a statement in this paper "...the rather unfortunate situation where the concept of the line tension is sometimes completely ignored in situations where it plays an important role."


      I still would love to get an answer to my original question. Is line tension included in the Fluent multi-phase model?


      Thanks,


      Alex

    • DrAmine
      Ansys Employee

      Perhaps my last answer was not clear enough.  Line tension is supported but the handling might be prone to spurious currents due to numerical approximation.

    • soloviev
      Subscriber

      I didn't understand that you were talking about line tension in your answer yesterday since you never mentioned "line tension". Now I see that it was about the line tension.


      The ANSYS Fluent Theory Guide does not mention line tension. According to the paper that I attached in the previous message, the line tension can differ by six orders of magnitude.


       Can you provide the parameterization formula to see how the line tension is included in the ANSYS Fluent multiphase model? 


      Thanks,


      Alex

    • DrAmine
      Ansys Employee

      Perhaps I have some different understanding of the line tension: Contact angle is the output which Fluent should ideally predict based on Force balance during equilibrium.  This will be for each phase pair. The force balance is pair based and might be related to the curvature or directly be understood as a stress term. Once again non-zero spurious currents (gradient approximation with or without surface tension) at equilibrium might affect the reality. If the line tension is the result one would expect at the contact of the three phases then this could be considered as a result from Fluent. Ideally, at the line tension the net force will be zero (Newmann's triangle). A zero is very hard to achieve. Brackbill force and surface stress are described in the manual. I will share the paper internally as this might require deeper investigation.

    • soloviev
      Subscriber

      Line tension is different from the Force balance during equilibrium. The line tension has the dimension N, in contrast to N/m for interfacial tensions between different phases. There is a long chain of publications about line tension starting from Gibbs. The line tension can be six orders of magnitude different (See paper attached in my previous message.)


      I believe that the difference between the multiphase model and our laboratory experiment with oil drops is due to the line tension issue in the model.


      Thanks,


      Alex

    • DrAmine
      Ansys Employee

      I apologize that I required some time to understand what you are actually looking for. If one neglects line tension then the Newmann triangle is fulfilled. The latter is not the case if line tension effects are considered. I have large doubts that those are implemented in VOF. I remember that these effects like disjoining pressure were only feasible as I was a student with DNS codes with explicit VOF.


      So the line tension is nothing else but the cumulative deviation of surface tension compared to the values far away from the contact line. The line tension is due to imbalances of intermolecular forces.  If you can write us a summary and add more references so I will forward that to development. You can directly write your local support in NA or even if you like forward everything to me per mail. Especially how to quantify it theoretically is very important for us and line tension has been usually object if controversy. 


       

    • soloviev
      Subscriber

      Hello Amine,


      The most recent information that I have found is in the Law et al. (2017) paper that was provided in a previous message. This publication contains a comprehensive review of previous publications on this subject. I am not really involved in the theoretical side of the line tension problem. We have run a laboratory experiment in parallel with a VOF model and see discrepancies. 


      My lab may be able to help with verification of the possible VOF modification using laboratory data, which we have collected during previous experiments.


      We are planning to publish our laboratory and computational results on oil drop dynamics on the sea surface as a part of our ongoing Gulf of Mexico Research Initiative Project involving the University of Miami and Columbia University. We are open for collaboration on the implementation of line tension in the VOF model. If you will provide your email address, it might be easier to communicate.


      Thank you,


      Alex


      soloviev@nova.edu

    • DrAmine
      Ansys Employee

      Dear Alex,


       


      I have forwarded your contact internally. Thanks a lot for the interesting discussion.

    • soloviev
      Subscriber

      What is the status of this request? Have there been any updates on the line tension in multiphase Fluent? We were told several months ago that someone would contact us about this issue and I have not heard any response. We are submitting a publication in the very near future, which relates to this issue.


      Thank you,


      Alex

Viewing 27 reply threads
  • You must be logged in to reply to this topic.