WEBVTT Kind: captions Language: en 00:00:01.040 --> 00:00:06.560 My name is Michel Boufadel. I'm the PI  and Dr. Xiaolong (Leo) Geng is Co-PI.   00:00:07.920 --> 00:00:15.040 We collaborated with various researchers in the  nation. You can see their list here at Princeton,   00:00:15.040 --> 00:00:19.360 Duke, Rutgers, Hopkins, University of  Pittsburgh, and the University of Cincinnati. 00:00:21.600 --> 00:00:27.200 The first part of the talk is going to  focus on the number of cases in the U.S. So,   00:00:28.080 --> 00:00:34.320 you obtain these data from Johns Hopkins  University and you know they provided   00:00:34.320 --> 00:00:39.920 them daily. And then so we analyzed the special  distribution of the number of cases and then you   00:00:39.920 --> 00:00:47.360 can see here in March 2020 and then in May. And  then like if you zoom in on the certain region,   00:00:47.360 --> 00:00:52.880 let's say this is the Washington D.C. area, we  noticed that the number of observed cases you   00:00:52.880 --> 00:00:58.480 know they are spiky. So, you have like the high  number, maybe this is D.C. or Baltimore, and   00:00:58.480 --> 00:01:04.000 then you move and then there's nothing in between.  You do the smaller population and then you have a   00:01:04.000 --> 00:01:09.840 higher number. So, for us that's reminiscent  of what's observed in, you know, in turbulence   00:01:10.560 --> 00:01:16.720 and then so we thought okay that the number of  cases would be something known as multifractal. 00:01:18.640 --> 00:01:26.400 And so, we, you know, we got to investigate that.  So, the conclusion is that the number of COVID-19   00:01:26.400 --> 00:01:34.320 cases is what you call scaling, and then, but  it's not totally random. It is correlated and   00:01:34.320 --> 00:01:40.320 we analyze the correlation using what we call the  Fourier spectrum. So first because it is scaling,   00:01:40.320 --> 00:01:47.360 so the- you can find a direct relation between  what's happening at 10 kilometers up to 2,600   00:01:47.360 --> 00:01:55.360 kilometers. So initially, during the early phase  of the disease, the correlation was small you   00:01:55.360 --> 00:02:01.760 know- as you can- as one could deduce from the  slope. As the slope gets, you know, steeper, then   00:02:01.760 --> 00:02:06.960 it means the correlation increases and then we  notice that the spatial correlation of the disease   00:02:06.960 --> 00:02:13.760 converges towards the correlation- the spatial  correlation of the population. And these are other   00:02:13.760 --> 00:02:18.720 multi-factor properties that you know- they are  in the paper. I'm not going to discuss them now. 00:02:19.680 --> 00:02:23.360 And then for our investigation, we used a   00:02:23.360 --> 00:02:28.080 relatively simple model developed,  you know, more than 120 years ago.   00:02:28.080 --> 00:02:33.840 It is called SIR model. Susceptible these are the  people who could be infected- the infected ones,   00:02:33.840 --> 00:02:39.360 infectious and then removed. These could  be removed due to recover or by death. 00:02:40.160 --> 00:02:48.160 So, we are- we use this model to try to capture  what is happening and you know if you recall that   00:02:48.160 --> 00:02:54.960 spectral slope figure, this is here shown as a  time series using also our model which is the   00:02:54.960 --> 00:03:01.680 line. And then one could note that we were able to  produce the spatial correlation using that model.   00:03:03.040 --> 00:03:09.040 This is just an illustration of how our model  functions. So, we start with a population that   00:03:09.040 --> 00:03:16.320 is multifractal and then we assign, you know,  the model for infection and then you can see   00:03:16.320 --> 00:03:21.440 these are the number of new cases of course with  time- the number of new cases with the subsides. 00:03:22.880 --> 00:03:27.440 The conclusion of this is that,  you know- the first thing was   00:03:27.440 --> 00:03:32.080 the major finding for us was the population-  the special distribution of population   00:03:32.080 --> 00:03:39.440 is multifractal, so which allows us to explain why  the COVID-19 spatial distribution is multifractal.   00:03:41.040 --> 00:03:46.720 You know, there are major work where they use  big data to model the spread of the disease   00:03:46.720 --> 00:03:52.640 using number of people, using their phones, so  our approach, you know, provide a compromise   00:03:52.640 --> 00:03:59.360 between the big approach- the big data approach  and, you know, fitting models at small towns say   00:03:59.360 --> 00:04:05.200 at the scale, say, of Newark. And there's  always issues of privacy using big data.   00:04:06.800 --> 00:04:11.600 And the other, you know, again this is  maybe pure modeling but we believe that   00:04:11.600 --> 00:04:15.760 paying attention to special correlation would  constrain the model so it doesn't go wild. 00:04:17.520 --> 00:04:23.920 The next part of my talk is about the movement  of virion you know or you know just call them   00:04:23.920 --> 00:04:29.040 particles in the supermarket. Imagine this is  a supermarket that is 40 meters long, you know,   00:04:29.040 --> 00:04:37.920 20-25 meter wide. And then you have the doors  here. The red in the arrows these are the- where   00:04:37.920 --> 00:04:43.600 the air comes from- vents. And then the white  arrows are the return vents. This is hypothetical. 00:04:45.120 --> 00:04:48.000 And so, we use CFD [Computational  Fluid Dynamics] simulation, we call   00:04:48.560 --> 00:04:52.160 it RANS, to model the movement  of air in the supermarket. 00:04:52.160 --> 00:04:57.840 And I want to show here the results. Our focus was  on the attachment of the particles. There are a   00:04:57.840 --> 00:05:03.280 lot of studies that deal with the transport, but  for us we say, okay, you know, what happened? You   00:05:03.280 --> 00:05:09.040 know because we do know that the part you know the  virus or the particles they do attach to surfaces.   00:05:09.600 --> 00:05:14.640 So here you can see them attaching to the ceiling  the orange. They attached to shelves you know   00:05:14.640 --> 00:05:20.000 blue, attached on the floor which is yellow.  Whereas if you don't allow for attachment,   00:05:20.000 --> 00:05:24.560 you know even after 20 minutes you see them  spread all over the place. So therefore,   00:05:24.560 --> 00:05:30.400 the attachment on services is important when you  want to predict the indoor transport of viruses. 00:05:32.160 --> 00:05:38.640 This is one curve here where you have  one graph. You have the concentration   00:05:39.200 --> 00:05:45.040 at 5 meters from the source. This is  without attachment of 5-micron droplets,   00:05:45.040 --> 00:05:51.600 so it is 20 percent the strength of  the source. With 25 percent attachment,   00:05:51.600 --> 00:05:57.920 you can see this is like maybe 12 percent and then  with 100 percent attachment is like 10 percent.   00:05:57.920 --> 00:06:03.520 So, we conclude that the attachment doesn't  play a role which means the type of surfaces   00:06:03.520 --> 00:06:08.320 in the supermarket is not going to be-  it's not going to play a major role   00:06:08.320 --> 00:06:13.600 because there were discussions like oh should  we use you know metal or glass or plastic?   00:06:13.600 --> 00:06:17.200 We- based on these simulations, it  seems it doesn't make a big difference. 00:06:20.080 --> 00:06:24.800 One thing we investigated is also, you know,  when they said, okay, there's one way in the   00:06:24.800 --> 00:06:30.480 supermarket so people could walk one way- one-way  isles. And then we said okay well one of the   00:06:30.480 --> 00:06:36.400 things that you could reduce you know the number  of air particles the virus particles in the air,   00:06:37.200 --> 00:06:41.760 is maybe you can create "Baffles". This is- as  an environmental engineer we're used to using-   00:06:41.760 --> 00:06:47.840 to this concept for plug flow reactors. And  then we conclude that if you place these   00:06:48.640 --> 00:06:55.360 baffles in the system, you are going to reduce  the concentration of particles in the air. And   00:06:57.680 --> 00:07:03.680 the other thing that- from the study is that the  narrower the aisles, the better the air quality,   00:07:03.680 --> 00:07:07.920 which is kind of counterintuitive because every-  you know whenever you look at a supermarket you   00:07:07.920 --> 00:07:15.200 know you look at large aisles and then it gives  you the feeling that it is healthier. Thank you.