Spatial Temperature averaging in non-isothermal confined flow

Dear all,

I wanted to simulate fluid flow in confined channel with two parallel plates in different constant temperatures. For temperature thermostat of wall atoms I used temp/rescale with spring command to keep them together. To get temperature distribution in the channel width I used compute temp/profile to subtract the COM velocity in each region. I defined different regions in fluid domain along channel width to get spatial temperature distribution.

The problem is that for different wall temperatures, I get constant fluid temperature along the channel width, which its magnitude is close to lower wall temperature.
Does anybody faced such problem before? Velocity and density profiles look reasonable, but the output temperature is almost constant value in different bins instead of having polynomial distribution.

Any comments is appreciated.

Hard to comment based only on your description. Makes little sense that your fluid doesn’t feel the different temps from the walls unless is not interacting with them!


That was so weird for me as I think everything are correctly applied. The fluid wall interaction parameters, thermostat and averaging are as follow:

pair_style lj/cut 5.0
pair_modify shift yes
pair_coeff 1 1 1.0 1.0
pair_coeff 1 2 0.6 1.0
pair_coeff 2 2 0.36 1.0

compute fluidbin1 flow1 temp/profile 1 1 1 y 5

fix 3 upper temp/rescale 1 1.5 1.5 0.005 1.0
fix 4 lower temp/rescale 1 0.9 0.9 0.005 1.0
fix 5 wall spring/self 1200
fix 9 flow addforce 0.0091 0 0

fix temp4 flow ave/time 100 1000 100000 c_fluidbin1 file tempbin.profile

flow1 is one of the fluid region which is defined by splitting the fluid domain.

Do you think I apply anything wrong? since fluid temperature should not be constant and it must be affected by both walls.