You are on page 1of 19

4/25/2021

Papers read today include

 Zhou et.al’s RANS can be successfully modeled for snowdrift. He has arguments as well
as counter arguments that support the RANS over LES and also its detrimental aspect
like over prediction of kinetic energy at front, higher deposition at frontal edge. Etc
 Another paper of Zhou states that snowdrift and snowmelt can be coupled to predict
the snow drift using meteorological data.
 Also, Tominaga’s 2011 development of snowdrift and snowmelt separate model was
studied.

One indication about new research ideas is that the number of parameters introduced in the
simulation procedure helps to increase the quality of simulation and bring it close to the
practical cases. All possible parameters related to snowdrift are listed below:

 Velocity field
 Snow density (constant assumed but is it temporally variable????)
 Snow Concentration
 Shear stress and friction velocity
 Velocity profile
 Eddy viscosity model for resolving Navier Stokes equation
 Turbulent kinetic energy profile
 TKE Dissipation profile
 Snow mass fraction profile
 Height of saltation
 Temperature
 Humidity
 Melting
 Downburst and splash effect
 RANS and URANS
 Erosion and Deposition
 Snowfall rate
 Model orientation
 Model size
 Model shape and type of roof Bluff body aerodynamics
 Grid overlay
 Grid discretization method(Finite volume vs finite difference)
 Successive accumulation and change in the snow bed profile
 Frequency of snowfall
 Time duration of snowfall
 Standing vortex formation place, size and its effect on snowdrift
 Change in convection and diffusion
 2D VS 3D
 Similarity criterion for prototype and real model
 Equilibrium and non-equilibrium of the velocity and mass flux in the saltation region
 Steady vs unsteady
 Snow pack bulk density
 Snow bond or cohesive strength
 Surface boundary condition
 Fully developed or partially developed flow in terms of atmospheric boundary layer of
snow phase
 Diameter of particle
 computational domain
 Variable of focus
 Iteration method

The purpose of calculating snowdrift is to find the variation of snowload on the structures
which ultimately helps to design the structures better as there are numerous failure cases
due to snow load.

Lets focus on improving on the cfd of steady RANS for better prediction of snowdrift by
incorporating more variables of flow into the system.
4/28/2021

In simulation,

 The code works on the stepped flat roof for snowdrift on ground.
 Analysis of snow on roof is needed to be considered in next simulation.
 Using array might be a possibility in simulation.
3/5/2021

Today and yesterday, modified the code to hook udf on all the zone ids and it worked.

Found out that the source=qtotal/height of saltation as mflux_source has the unit of kg/m3sec.
And the height of saltation is actually 0.01 for my case but the simulation goes haywire when I
try to use 0.01 as the source is magnified immensely.

Also found an article about finding out the viscosity of snow but another way. It is done using
the cross mapping of eulerian and langrangian approach.
4/5/2021

Today, tried to run the simulation using condition:

UDf code has been changed with Source= qtotal/0.01 as 0.01 is our height of saltation(hsal=hp).

Top- velocity inlet

For mixture: both TKE and TKD were kept 0

For air: udf wind velocity along x

For snow: udf wind velocity along x and sn-0.2 on z.

The simulation fluctuated bizzarely and the simulation doesnot come to conclusion.

Also at home, the simulation with following conditions gave wrong results.

simulation 1

ground - 0.00587 0.5

upper and lower roof - 0.06 0.9

height of saltation 0.01

on top- velocity inlet(both air and snow phase were with x velocity as udf and z velocity as -0.2.
5/5/2021

New Simulation

Saltation height=0.01m

Alpha_C=0.2

Alpha_J=0.15

ground - 0.06 0.9

upper and lower roof - 0.06 0.9

height of saltation 0.01

on top- velocity inlet(both air and snow phase were with x velocity as udf and z velocity as -0.2.

According to yinzi alpha_z and alpha_c are the same. They are the same parameter. But still this
simulation was run.

Result: This simulation did not even has the same contour at the lower roof. So, the analysis is
flawed.

Today, I read about standard wall functions with Ks and Cs using youtube Fluid Mechanics 101,
yes kS and Cs are the roughness height and roughness constant.
5/5/2021

Simulation results 1 with lab computer:

Mesh: cube type mesh with bottom layer 0.1 and others 0.08 on the line. This reduced the
number of cells (2248950 hexahedral cells). Maximum grid space 0.25. Both front and back 0.25
grid spacing from the model. With maximum vertical grid spacing of 0.2. Top as symmetry

Y+ value on ground= 2100, Y+ value on z of0.05m=2500, Y+ value on z of 0.1m= 3000

Reynold’s number at 0.1: 83(ground), 74(0.05), 57(0.1) at the corner of buidlidng.

Settings: same for the cube analysis with hp as 0.1 in code.

Results:

UDM – 6 Snowdepthdimensionless(udm-6/(0.00033347*150*0.2)

Snowdepthdimensionless
Friction velocity(sqrt(udm-4/1.225)/0.15)

Simulation results 2 with laptop:

Mesh: Zhang type mesh with finer grid with first layer 0.01m, maximum grid space 0.25. This is
the number of cells (2299500 hexahedral cells). Progressive vertical grid spacing with Maximum
spacing of 0.5 and minimum of 0.01. Front grid spacing 0.15 and back grid spacing 0.25. Top as
symmetry

Settings: same for the cube analysis with hp as 0.1 in code.

Reynolds number:

Y+ value on ground= 2500, Y+ value on z of 0.01m=3142, Y+ value on z of 0.005m= 2933

Reynold’s number: 4640 at the far end of field.


UDM – 6 Snowdepthdimensionless(udm-6/(0.00033347*150*0.2)

Snowdepthdimensionless

Friction velocity(sqrt(udm-4/1.225)/0.15)
6/5/2021

Simulation results 3 with desktop:

Mesh: cube type mesh with bottom layer 0.1 and others 0.08 on the line. This reduced the
number of cells (2248950 hexahedral cells). Maximum grid space 0.25. Both front and back 0.25
grid spacing from the model. With maximum vertical grid spacing of 0.2. Top as symmetry.

Settings: same for the cube analysis with hp as 0.1 in code.

Code setting:

Ht of saltation(hp or hsal) = 0.1

Density = 185kg/m3

Alpha_c and alpha_j are 0.2

Velocity 1.1 at H=1

Ut and Wf are 0.2 AND -0.2

Vol fraction = 6e-7 instead of UDF volume fraction

Roughness height(ks) and roughness constant(cs) = 0.000587 and 0.5.

Y+ value on ground= 459, Y+ value on z of0.1m=596, Y+ value on z of 0.05m= 505

Reynold’s number: 87.9 at the far end of fieldground. Ar 0.1- 48. At 0.05- 65

Result:

Solution didn’t go beyond 10^-5 in residuals for vof.

UDM – 6
Snow depth dimensionless (udm-6/(6e-7*185*0.2)

Snow depth dimensionless


Dimensionless Friction velocity(sqrt(udm-4/1.225)/0.2)

Simulation results 4 with laptop:

Mesh: cube type mesh with bottom layer 0.1 and others 0.08 on the line. This reduced the
number of cells (2248950 hexahedral cells). Maximum grid space 0.25. Both front and back 0.25
grid spacing from the model. With maximum vertical grid spacing of 0.2. Top as symmetry.

Code setting:

Ht of saltation(hp or hsal) = 0.1

Density = 185kg/m3

Alpha_c and alpha_j are 0.2

Velocity 1.1 at H=1

Ut and Wf are 0.2 AND -0.2

Vol fraction = 6e-7 instead of UDF volume fraction

Roughness height(ks) and roughness constant(cs) = 0.06 and 0.9.

Y+ value on ground= 459, Y+ value on z of0.1m=596, Y+ value on z of 0.05m= 505

Reynold’s number: 87.9 at the far end of fieldground. Ar 0.1- 48. At 0.05- 65
UDM – 6

Snow depth dimensionless (udm-6/(6e-7*185*0.2)

Dimensionless Friction velocity(sqrt(udm-4/1.225)/0.2)

8/5/2021
Simulation results 5 with laptop:

Mesh: cube type mesh with bottom layer 0.1 and others 0.08 on the line. This reduced the
number of cells (2248950 hexahedral cells). Maximum grid space 0.25. Both front and back 0.25
grid spacing from the model. With maximum vertical grid spacing of 0.2. Top as velocity inlet.

Code setting:

Ht of saltation(hp or hsal) = 0.1

Density = 185kg/m3

Alpha_c and alpha_j are 0.2

Velocity 1.1 at H=1

Ut and Wf are 0.2 AND -0.2

Vol fraction = 6e-7 instead of UDF volume fraction

Roughness height(ks) and roughness constant(cs) = 0.000587 and 0.5

Y+ value on ground= 457.28, Y+ value on z of0.1m= 503.01, Y+ value on z of 0.05m= 640.2


Reynold’s number: 87.47at the far end of fieldground. At 0.05- 74.37 And 0.1- 61.27

UDM – 6

Snow depth dimensionless (udm-6/(6e-7*185*0.2)

Snow depth dimensionless


Dimensionless Friction velocity(sqrt(udm-4/1.225)/0.2)
A new finidng has been made whci says that the hroughness height(ks) cannot be greater than
the first cell height(hp).

Simulation results 6 with laptop:

Mesh: zhang type mesh with hp 0.01, growing 0.045spacing and 0.06 last length.

Hoz spacing=0.2, near body=0.1, above roof top: hp=0.01, spacing=0.1, last length=0.2(plz
confirm) Top as symmetry

Code setting:

Ht of saltation(hp or hsal) = 0.01

Density = 185kg/m3

Alpha_c and alpha_j are 0.2

Velocity 1.1 at H=1

Ut and Wf are 0.2 AND -0.2

Vol fraction = 6e-7 instead of UDF volume fraction

Roughness height(ks) and roughness constant(cs) = 0.00006 and 0.9

Y+ value on ground= 234.13583, Y+ value on z of0.1m= 439, Y+ value on z of 0.05m= 439

Reynold’s number: 305 at the far end of fieldground. At 0.05- 251 And 0.1- 251

UDM – 6
Snow depth dimensionless (udm-6/(6e-7*185*0.2)

Snow depth dimensionless

Dimensionless Friction velocity(sqrt(udm-4/1.225)/0.2)


Simulation results 7 with desktop:

Mesh: zhang type mesh with hp 0.01, growing 0.045spacing and 0.06 last length.

Hoz spacing=0.25, near body=0.1, far- 0.3

above roof top: hp=0.01, spacing=0.25, last length=0.3

no of cells: 2350950 hexacells

Top as symmetry

Code setting:

Ht of saltation(hp or hsal) = 0.01

Density = 185kg/m3

Alpha_c and alpha_j are 0.2

Velocity 1.1 at H=1

Ut and Wf are 0.2 AND -0.2

Vol fraction = 6e-7 instead of UDF volume fraction

Roughness height(ks) and roughness constant(cs) = 0.00006 and 0.5

Y+ value on ground= 234.13583, Y+ value on z of0.1m= 439, Y+ value on z of 0.05m= 439

Reynold’s number: 305 at the far end of fieldground. At 0.05- 251 And 0.1- 251

You might also like