|Subject:||[ESPResSo-users] modelling rigid particles|
|Date:||Mon, 30 Oct 2017 09:02:13 +0000|
Dear ESPResSo users,
I want to simulate the motion of a rigid object using the ESPResSo software package. I have configured ESPResSo with all the features of the Object-in-fluid (OIF) module needed to model an object in a fluid with given elastic properties . The object has the (complex) shape of a dumbbell (as the ones studied by Uspal, Eral and Doyle in https://www.nature.com/articles/ncomms3666 ) and a mesh of its surface is generated in the right ESPResSo format . After configuration, I searched for the source code that determines the interaction forces between the object and the fluid. I think the only relevant forces are forces due to interaction with the Lattice Boltzmann fluid (that means there are no Thermostat, Coulomb interactions etcetera).
My plan is to redirect the forces that deforms the object to the propagation/rotation of the object, so it conserves its shape. I tried to track down the source code in which the deformation/elasticity forces are computed and assigned to the surface nodes of the object. But because my programming skills are not that good, I’ve troubles reading the complex c++ codes.
I think the core code “integrate.cpp” (src > core > integrate.cpp) is the most important in which the motion of the object is integrated. But from there I don’t know how the forces (or momenta) are transferred from the Lattice-Boltzmann fluid to the object nodes.
Can anyone confirm my plan or explain why it will not work out to model the motion of rigid objects? Secondly, I would like to know where the relevant codes are with which I can influence the deformation forces directly?
Some of you will wonder why I decide to take this approach for modeling rigid objects. Why can’t I just increase all elasticity parameters to approximate a rigid body? Part of my goal is to develop a fast and computationally efficient code. Omitting the deforming/elasticity properties will help to ease the computation and makes the computations more stable. At higher values for the elasticity parameters the computation becomes instable and crashes.
Ideally, I would like to implement the exact same approach employed by Mackay et al. described in https://doi.org/10.1016/j.cpc.2013.03.024 , but making use of the GPU-enabled ESPResSo LBM. Do you think this is possible?
Thank you for your time and help!
Tom Roest, BSc
|[Prev in Thread]||Current Thread||[Next in Thread]|