Vicosity Too Hot

Dear users,

I am trying to do shear viscosity but the temperature keeps spiking too high. Is there an extra fix that could better control the temp while the shear rate is being applied?

Input:

Apply Shear Rate

fix 1 all nvt/sllod temp 298.15 298.15 100.0
fix 2 all deform 1 xy erate ${strain} remap v

Temperature output:

pastedImage.png

Thanks in advance,

Undergrad Research

Tyler

Dear users,

I am trying to do shear viscosity but the temperature keeps spiking too
high. Is there an extra fix that could better control the temp while the
shear rate is being applied?

this sounds like a rather drastic hit-it-with-the-big-hammer-until-it-gives
up solution.
before even considering this route, your should ask yourself a few other
questions:
- do you have a well equilibrated starting configuration?
- does your system conserve energy when you run without shearing (using fix
nve)? does it properly maintain equilibrium (using fix nvt)?
- is your time step adequate? when shearing, you probably should reduce the
time step from what is suitable for a regular bulk system.
- is your system large enough?
- does it maintain a suitable velocity profile with a very small straining
rate?

axel.

pastedImage.png

Hi Tyler,

Is this an unbiased temperature calculation, or the temperature from a compute temp/deform?

-David

pastedImage.png

Lammps Users,

David
-Is this an unbiased temperature calculation, or the temperature from a compute temp/deform?

the only compute command used was for pressure 'compute 1 all pressure thermo_temp' and using the 'thermo_style cutsom temp' command to get the temperature. So it should just be the Nose-Hoover style. Does that mean unbiased? And do you think adding the compute could affect the temperature?

Axel
- do you have a well equilibrated starting configuration?
>it was run for over 10 ns in npt before switching
- does your system conserve energy when you run without shearing (using fix nve)? does it properly maintain equilibrium (using fix nvt)?

ran for 2ns in nvt (it fluctuates a lot but still averages to 298K)

- is your time step adequate? when shearing, you probably should reduce the time step from what is suitable for a regular bulk system.

the time step was 0.01 in "real" units

- is your system large enough?

it has 343 molecules (7x7x7 box)

- does it maintain a suitable velocity profile with a very small straining rate?

from what I could tell, the velocity profile looked stable and the gradient didn't fluctuate much.

I did a log-scale strain sweep from 0.0000001- 100.0. It started having trouble regulating temp at 0.0001 strain rate and skyrocketed above that. The .00001 and 0.0001 were the only 2 points that were reasonable, viscosity value was .8 cP while experimental of 1.1cP.

I can get more points in this range, but was wondering if there was a better way to either: regulate the temperature to achieve higher strain rates and/or control the pressure more precisely for the lower strain rates, to get more data points?

Thanks for the Reply,
Undergrad Research
Tyler

Lammps Users,

David
-Is this an unbiased temperature calculation, or the temperature from a compute temp/deform?

the only compute command used was for pressure 'compute 1 all pressure thermo_temp' and using the 'thermo_style cutsom temp' command to get the temperature. So it should just be the Nose-Hoover style. Does that mean unbiased? And do you think adding the compute could affect the temperature?

the default temperature compute (what is output with the temp keyword)
doesn't know anything about your flow inducing deformation and will
include that flow velocity into the temperature output. however, for
such a run this should not be done. please see
examples/VISCOSITY/in.nemd.2d

Axel
- do you have a well equilibrated starting configuration?
>it was run for over 10 ns in npt before switching
- does your system conserve energy when you run without shearing (using fix nve)? does it properly maintain equilibrium (using fix nvt)?

ran for 2ns in nvt (it fluctuates a lot but still averages to 298K)

that is good, but that doesn't say anything about energy conservation,
which has to be observed without any thermostat or barostat.

- is your time step adequate? when shearing, you probably should reduce the time step from what is suitable for a regular bulk system.

the time step was 0.01 in "real" units

that is a *very* small time step for a molecular system in real units.
you should have good energy conservation with an order of magnitude
larger timestep.

- is your system large enough?

it has 343 molecules (7x7x7 box)

- does it maintain a suitable velocity profile with a very small straining rate?

from what I could tell, the velocity profile looked stable and the gradient didn't fluctuate much.

that sounds good, too.

I did a log-scale strain sweep from 0.0000001- 100.0. It started having trouble regulating temp at 0.0001 strain rate and skyrocketed above that. The .00001 and 0.0001 were the only 2 points that were reasonable, viscosity value was .8 cP while experimental of 1.1cP.

I can get more points in this range, but was wondering if there was a better way to either: regulate the temperature to achieve higher strain rates and/or control the pressure more precisely for the lower strain rates, to get more data points?

i don't understand by what you mean with "controlling pressure". there
should be no need for pressure control for this kind of simulation.

axel.