Droplet Integration and "pollution heatmap"

[agrignard]

We have been discussing with @tnguyenh to see how we could implement this

A first step could be to place people at their desk and then to make them emit droplet with this law Rate of virions emitted by an infected person • Sitting quietly: 20 / min => 90% of the time • Talking: 200 / min=> 10% of the time • Singing, shouting, or exercising: 2000 / min => between 1 and 3 per hour • Sneeze: 200,000,000 => between 1 and 10 time a day?

Then from this emission I was thinking that we could easily draw a kind of pollution heatmap.

Now the question is how thoe droplet diffuse them self? Woudl it be possible to have a uniform spreading? And why not thinking of integrating the ventilation that could impact of the direction of the spreading.

[agrignard]

Maybe a good idea would be to start from only one room like CityScience like this one https://github.com/CityScope/CS_Proxymix/blob/master/ProxymixABM/includes/CityScience/building.dxf

[LAAP]

@agrignard and @tnguyenh, Integrating this knowledge is a very good idea.

I suggest to start with a simpler approach (As similar as possible as the IDSS documentation), and after we can try something a bit more complex.

With that in mind, I think that:

1) A pollution heatmap is a great idea

2) Difusión: I suggest a uniform spreading

3) Ventilation: Sounds like a dream, however some knowledge in fluid dynamics maybe needed. So I recommend to start by using the IDSS "time/volume-people" graph. I was using it assuming the curve 0.003 as the limit for "natural ventilation" (UdeG most popular setting in the classrooms)

4) If you test in a room, I recommend to test in UdeG's buildings. I Can put furniture if you wish. part of our success in the presentation was that we were simulating in their facilities. Over there, they have natural ventilation

[agrignard]

Hi @LAAP all the document comes form the IDSS doc, so the goal is to start with this, nothing more complex as it's already pretty tricky to implement

For the point 4, the reason why I wanted to start with CityScience is simply because it is small, this approach will be very time consuming and doesn't make sense to be build at the building scale as the goal is to evaluate a specific spaces. If you can produce a details typical room in UdG that would be great.

[LAAP]

@agrignard ,

Yes, this is very easy. The detail typical room at the UdeG will be a classroom. Is that OK? If so, I can put chairs and tables and push it. Please, let me know!

[agrignard]

One more sources that can be helpful

https://www.ashrae.org/File%20Library/About/Position%20Documents/Airborne-Infectious-Diseases.pdf

[agrignard]

Firs version of a basic droplet integration where any agent "emits" a certain amount of droplet

things that can be parametrized so far

• Droplet size distribution
• Droplet lifespan
• Droplet Distance

[LAAP]

@agrignard , This is looking great! Any progress in the AC ventilation issue?

[agrignard]

The first version that we have now is a good proof of concept. I close this issue for now, if anyone has idea to make the droplet more realistic please feel free to reopen this issue

[LAAP]

@agrignard ,

Some important questions before you closes this issue, I think there is a huge potential here, but we are not using it at all:

• Can we infer the areas of the building with high risk, from the amount of droplets accumulated in a m2?
• Can be reduce the droplets by assuming people following the rules of wearing masks, sanitation points, cleaning surfaces etc?
• Are agents getting "potentially" infected by the exposition to the droplets?
• Can we have a "cleaner agent", that, if we switch on the button, will emerge and clean rooms right after meetings so the droplets are reduced?

[agrignard]

Right now the droplet features is implemented as follow:

species droplet skills:[moving]{
    int live_span <- droplet_livespan update: live_span - 1;
    int size<-14+rnd(200);
    aspect base{
        draw circle(size/1000) color:rgb(size*1.1,size*1.6,200,50);
    }
}

ask people{
  create droplet{
    location<-myself.location+ {rnd(-droplet_distance,droplet_distance),rnd(-droplet_distance,droplet_distance)};
  } 
}
ask droplet where (each.live_span <= 0) {do die;}

Wold be nice to have something more realist, even in 3D, @tnguyenh I think some cool mathematical equation could be useful here

In will impact also #108

[LAAP]

Hi @agrignard ,

I think this is very related to issue #120 (Maybe, even is an overlapping issue)

[agrignard]

small improvement with droplet in 3D but still we really need some TriD here @tnguyenh ! the droplet should be closer to the people (they are too random right now)

[LAAP]

@agrignard and @tnguyenh ,

This is opening a cool door. If we have the height of the droplets, then, every droplet below 1m or 1.25m, will infect less than droplets at the level of the mouths, noses, and eyes

[agrignard]

FYI https://alkazar.com.tr/blog/indoor-air-quality-for-office-rooms/

[LAAP]

@agrignard ,

Very cool document for fluid dynamics. Air quality, when talking about virus is a bit different. Only HEPA filters can kill 99% of viruses, so AC can be a huge problem. Indoor environments can increase the risk in a 18.7% more, as it is shown in this article:

https://www.medrxiv.org/content/10.1101/2020.02.28.20029272v2

"The odds that a primary case transmitted COVID-19 in a closed environment was 18.7 times greater compared to an open-air environment (95% confidence interval [CI]: 6.0, 57.9). Conclusions: It is plausible that closed environments contribute to secondary transmission of COVID-19 and promote superspreading events. Our findings are also consistent with the declining incidence of COVID-19 cases in China, as gathering in closed environments was prohibited in the wake of the rapid spread of the disease."

[agrignard]

Droplet are implemented and working but it seems that we wo'nt integrate it in the 5 educational video. If we want to push this model further keep in mind that we have a droplet model first version working.