3DAirQualityPrediction User Manual

This user manual presents the available options and workflows for the 3DAirQualityPrediction pilot. For using 3DAirQualityPrediction please first contact system administrators for HPC credentials.

1. Login and create Instance

First we need to login to the MSO4SC portal, we can do it at portal.mso4sc.eu site, after successful login we can navigate to the Experiments menu.

login screen
Figure 1. Login screen of the MSO4SC portal
frontend
Figure 2. The MSO4SC dashboard

On the Instances we can add a new instance by clicking the "add a new one". In this instance we can create a simulation with user-defined datasets, and run it on the same webpage.

First we must select the Appcilaction 3DAirQualityPrediction and add a new ID for our instance.

application
Figure 3. Adding a new Instance

After that, the next session describes the available datasets, and configuration options of the 3DAirQualityPrediciton Pilot.

2. Map Data

In this case, we cannot choose different map, only gyor-mid.osm. This is the downtown of Győr.

3. Traffic simulation (1-traffic-SUMO-vehiclesph)

The traffic simulation based on SUMO (Simulation of Urban MObility, url:http://sumo.dlr.de/) simulation tool. In this scenario are using the downtown map of Győr, Hungary. After we modelled the map of Győr, we generate traffic to the streets. We can add a static Vehicle-per-hour rate of traffic.

  • If we change the option to 0, we can use a default traffic based on measurement

  • If we add a number between 1 and 10000, the simulatior will generate a number of vehicles per hour, equally spaced (not together with period or probability)

traffic
Figure 4. Defining the traffic size of the simulation

4. 3D mesh

In our pilot we can choose 3 different level of resolution of the predefinied 3D octree mesh for CFD simulation. On the Fig. 5, we can see the difference between low and mid resolution.

  • low (Lowest resolution)

  • mid (Middle resolution)

  • high (Highest Resolution)

lowmid
Figure 5. difference of the low and mid resolution

5. CFD simulation

In this section we can add the configuration for the Ansys Fluent CFD simulation. For now, there are 3 different options, which we can change. NOTE: The time step is hardcoded to 60 seconds. So choose the iteration number carefully.

  • 3-fluent-sim-start-time: Start time of the simulation (in second from midnight)

  • 3-fluent-sim-iter-1: Iteration number for initial wind field computation (steady state)

  • 3-fluent-sim-iter-2: Number of time steps for mixture formation (transient analysis)

The start time is 18000 by default (this means the simulation starts at 5:00am).

6. Wind Generation

In this section we describes the wind profile geneartor, which can generate transient wind profile from discrete measurements and surface roughness. The generator uses two different files, the first one used to add the basic configuration for the wind profile, the second contains only the measurement data.

The two files are stored in the integrated Gitlab module and can be easily modified online through the Gitlab Editor.

For this step we only provide the url for the two files.

windfiles
Figure 6. The two wind profile files

The wind profile config file with comments:

measurement.start-time=2018.09.02 00:00:00

# Measurement height in meters:
measurement.z=2

#Specific surface roughness of measurement:
measurement.z0=0.2

#Displacement height of measurement:
measurement.d=0

#Coordinate system of the measurement: descartes or polar. Descartes: x, y components, polar: magnitude, direction components:
measurement.coordinate-system=descartes

#Origo of the polar coordinate system: North, East, South, West:
measurement.coordinate-system.polar.origo=North

#The positive rotational direction of the polar coordinate-system: Clockwise, CounterClockwise:
measurement.coordinate-system.polar.positive-direction=Clockwise

#Full path of the measurement csv file. See measurement csv file specification in the wind-profile-creator-input-file-specification.md file.
measurement.data.path=/nfs/3dairq-simulation/wind/windprofile.csv

#### properties of generated profile ####

#Specific surface roughness of the application area
profile.z0=0.2

#Displacement height of application area
profile.d=0

#Vertical offset of the generated profile in meters
profile.offset=85

#Height of the generated profile in meters
profile.height=400

#Vertical resolution of the generated profile in meters
profile.resolution=10

#Rotates original measurement values (degrees, cw)
profile.modifier.rotation=0

#Multiplies original measured magnitudes
profile.modifier.magnitude-multiplier=1.0

Example wind data

time(s);x(m/s);y(m/s)
0;2.4;3.7
300;2.9;3.5
600;3.6;4.1
900;3.8;5.2
1200;4;5.1
1500;3;6.2
1800;4.1;3.2
2100;5.2;7.1

6.1. Editing wind profiles

First navigate to the main branch of the config files:

After, we have to login to the Gitlab. We don’t have to use our username/password, just hit the "Sign in / Register" button and the "Sign In with Filab".

gitlab1
Figure 7. The Gitlab public Repository for wind profile
gitlab2
Figure 8. Sign in with Filab

We have to fork a master branch for us, to enable the edit feature.

gitlab3
Figure 9. Forking the master branch

After that, we can edit the wind files, and copy-paste the corresponding new URL back to the portal.

gitlab5
Figure 10. Edit the file, and save the URL

7. Other configuration options

Because of licensing restrictions this pilot only avaliable on sze_node HPC cluster. Please ensure to select this option on "HPC: secondary" configuration.

There are some other configuration options for the developers, to ensure the proper simulation like datadir and job prefix. Please leave the default value on this section.

8. Running the simulation

After we ensured all the configuration was made, we can create the instance.

instance
Figure 11. Creating the new instance

If the instance created succesfully like in Fig. 11. simply choose the created instance and then hit the run button.

run2
Figure 12. Running Simulation

9. Postprocessing

To use built-in Post processing tool, navigate to the Visualization menu and hit the "Desktops Available" button. For 3DAirQualityPrediction we have to use the SZE-RD infrastructure.

Then hit the View Only button, for now Desktop is for developing porpuse.

visu1
Figure 13. Visualization

When we create a View-Only desktop, a remote desktop through VNC will open in a new window, you can open the simulation results with the industry-standard Paraview.

The simulation result is saved at /nfs/results/mso-3dairq-result-<SIMDATE-SIMTIME>
visu2
Figure 14. Opening simulation result
visu3
Figure 15. Viewing the simulation results

10. Donwloading simulation data

We can donwload our simulation data through SSH connection (WinSCP for Windows) using the credentials of the portal (without the "@mso4sc.eu").

The server is mso.tilb.sze.hu

download
Figure 16. Downloading the simulation results with WinSCP