fix SRD and gravity on colloids

Hello fellow LAMMPS users/developers,

I would like to simulate the sedimentation of colloids in presence of hydrodynamic interactions. To this purpose, I am using the SRD implementation provided by fix srd in LAMMPS.

I am testing my results against a particular implementation of stochastic rotational dynamics [Padding and Louis PRL 93, 220601 (2004)], extensively discussed by the same authors in [PRE 74, 031402 2006].

In my LAMMPS script, the system is composed by a few hundreds repulsive colloidal particles immersed in approximately 5x10^5 SRD particles. An acceleration is imposed only on the colloidal particles via fix gravity. The SRD dynamics is with slip collisions and with the SRD thermostat turned on.

My observation is that, for a fixed total volume, systems with fewer colloids have smaller sedimenting velocities than systems with more colloids, which is precisely the opposite of what the physics should be due to hydrodynamic backflow effects [see PRL above]. I am therefore attempting to identify the source of error.

As discussed in the appendix of [PRE 74, 031402 2006] since I am applying an external field on the colloids, this pumps energy into the system, hence the temperature rises unless I use a thermostat on the SRD particles. Padding and Louis suggest to go for a local and Galilean invariant definition of the temperature, subtracting the centre of mass velocity from the velocities of every SRD cell to estimate a rescaling factor for the velocities.

From the documentation, it appears that the SRD implementation in LAMMPS has a thermostat which subtracts the mean velocity in the cell if also fix deform is on, but what happens when the flow is induced by external fields, as in my case? Is the thermostatting procedure incorrect and hence affecting the sedimentation velocities?

Thanks for your advice

Francesco Turci PhD
HH Wills Physics Laboratory
Tyndall Avenue, Bristol, BS8 1TL, UK

The fix srd doc page paragraph about “tstat”, means that the thermostatting
will work in the normal case when no fix deform is defined and also when
fix deform is defined. The case with no fix deform simply means no
deformation velocity is added, i.e. the deformation vel = 0.0. So in that
case, the “thermal” velocity of the SRD particles is rescaled, i.e. after
subtracting out the mean velocity.