Re: [ESPResSo-users] Rigid aspherical object in LB fluid

[Iveta Jancigova]

Hello Rumen,

I think that the Object-in-fluid (OIF) module is not the best choice for your application.

While it is possible to use Espresso for simulations of rigid objects (either by setting the OIF elastic coefficients so that the object is very stiff or even better: by using virtual sites, where the object always keeps the same shape), it is problematic with respect to your other questions (starting with the most important one):

4. The physical friction between the disks and z-walls is not the friction we use with OIF. The friction coefficient in Espresso is a simulation parameter that represents the dissipative coupling between the immersed boundary of the object and the fluid. It needs to be properly calibrated so that the correct movement of object in fluid is achieved. We have some experience with modeling adhesion of objects to walls, but this is not the same as friction at the walls.

3. Currently, OIF cannot model objects with different density than the outside fluid.

1. Intersecting boundaries: in my experience, all nodes that are in at least one boundary, are boundary nodes and I have been using intersecting boundaries in this context. Problems may arise when your boundaries do not truly intersect but only touch, e.g. 

BBBBBBFFFFFFF

FFFFFFBBBBBBB  

where B stands for the boundary and F for fluid.

Maybe somebody else can comment in more detail on this topic. 

Best,

Iva Jancigova

 

Message: 1
Date: Mon, 5 Dec 2016 11:51:05 +0000
From: Rumen Georgiev - 3ME <address@hidden>
To: "address@hidden" <address@hidden>
Subject: [ESPResSo-users] Rigid aspherical object in LB fluid
Message-ID:
<address@hidden>
Content-Type: text/plain; charset="iso-8859-1"

Hello,

I intend to use the LB_GPU package coupled with either the Object-in-Fluid (OIF) or Immersed Boundary Method (IBM) packages and I have several questions regarding ESPResSo's capabilities.

I would like to track the motion of a rigid particle composed of two disks connected with a shaft and immersed in a flowing Lattice Boltzmann fluid. The system is periodic only in the flow direction (say, x) and has stationary planar walls in the other two (y and z) -- essentially, this a rectangular duct geometry. The particle is separated from the z-walls by thin lubrication layers. A full description of the system can be found here [doi:10.1038/ncomms3666].

After going through the UG, I have come to a couple of conclusions -- could you please verify them:

 1.  The UG states that intersecting boundaries are tricky -- could you please elaborate on that? What unphysical effects should one expect?
 2.  Both OIF and IBM place an object with a shape of choice in the LB fluid, however, they are both for soft particles. My guess is that by using the proper parameters of the elasticity moduli the particle will act as a rigid one in both cases.
 3.  The UG suggests the IBM package only supports particles which have the same mass density as the fluid and if one wishes to implement a particle with a different density, the OIF method should be used.
 4.  Since friction between the z-walls and the disks is crucial to recovering the experimentally-known motion of the particle, it seems that OIF is the natural choice, since IBM does not support friction.

Thank you very much for your time!


Best regards,


Rumen Georgiev


Rumen Georgiev  |  Intensified Reaction and Separation Systems  |  Process & Energy  |  3mE  |  TU Delft  |  34 J-0-940  |  Leeghwaterstraat 39, 2628 CB, Delft  |  tel. +316 82 946 879

—————————————————————————————————————
Mgr. Iveta Jancigova, PhD

Department of Software Technologies
Faculty of Management Science and Informatics
University of Zilina, Slovakia
cell-in-fluid research group: http://cell-in-fluid.fri.uniza.sk