 come to the talk. Patrick Römer will talk about creating a sensor network for measuring nitrogen dioxide. The talk has been come about in cooperation with Dr. Neil Seidel from the University of Hagen in the Institute for Cooperative Systems and it was based originally on a bachelor's thesis and the goal is to create a completely covering measure network that is much bigger and perhaps also better than the national measuring network. A warm welcome please for Patrick Römer. Hello my name is Patrick Römer as you have heard and I will introduce you to the sensor network for measuring nitrogen dioxide and I start with an introduction of the toxic toxicity and the problems with nitrogen dioxide and what are the measuring methods and I talk about creation and calibration of a measuring station and I talk about sensors and choice of sensors and how we build them and what the problems were and in the end I will show you some results of the measurements and how to we collect measurements and at first a nitrogen dioxide what is that? It is part of the nitrogen oxides the oxides of nitrogen and is a combination of nitrogen and oxygen a compound and there are natural sources for that by bacteria in the soil for example but in towns most of the nitrogen dioxide is created in the combustion chambers of engines normally that will create nitrogen monoxide at high temperatures and is a compound of nitrogen and oxygen it creates nitrogen monoxide in this metals table and although you put energy in it doesn't decompose by itself but it remains as nitrogen monoxide in the atmosphere and but then with ozone it combines to nitrogen dioxide which is the toxic gas that we are talking about and the legal situation is on in at the level of the European Union in the air quality directive 2008 50 EG and there are minimum requirements for the European Union and this minimum requirement they were converted into German law with the 39th emission regulation and this in this regulation there will be limit values based on WHO recommendations and there's a mean value for human health for 40 micrograms per cubic meter for during one hour the the mean value can only only be 200 micrograms per cubic meters and if you have that for several for a longer time there will be an alarm and the agencies will look into it and if there's an alarm level that is 400 micrograms per cubic meter which immediately raises an alarm and there's a an annual mean value to for the protection of plants for 30 micrograms per cubic meters and the the biggest problem is of course for for human for human health and there are data from from official official measuring systems and the mean value is the the thing that causes the most human problems human health problems and for methods there are quality standards for fixed measuring points that those are the ones that you know they're here for example at the railway station in Leipzig there's one and there's an orientating measurement and there are model calculations and there are objectives objective estimates which may have up to 75 percent error and and calculations up to 50 percent and measurements 15 or 25 percent and this national regulation in Germany also regulates other other toxins in in the air and has some other limit values for that as well in the 39th national national national regulation for natural dioxide and the measuring method is the chemical luminescence system and there's a system by the company teledyne and nitrogen monoxide is has is being combined to NO2 in a in an ultraviolet chamber and and the the light emissions then is proportional to the concentration and the gas is used with a catalyst and is is combined with oxygen to NO and this you will eventually measure and the the the process is quite specific has a has a little influence towards external noise and such such that's why it's used effectively and one such instrument is quite expensive and therefore it's rarely used because you need additional steps for processing the gas force is cooled therefore it use moisture afterwards it is heated up such that the reactions in the analytic chamber are used sufficiently fast and therefore it has a lot of processes and alternative process is the so-called passive collectors they are used for long-time average measurements they are small tubes which are filled with the chemical of the after time it just accumulates inside the tube NO2 and afterwards they are processed in laboratory and based on the collected NO2 values you can estimate the the average NO2 over the respective time and this has usually about 20 percent of it's 20 percent of error yes the the official measuring networks is controlled in this in this in this law so the measurement positions should be placed at at the positions where the the most most values of gases to be expected they should be representative representative for other places which are close around so the idea is that if you place it at one position which has a high traffic impact it is should be representative for other places with a similar amount of traffic so the question is if you're placed and if you are in if you're not living next to a next to a big street the question is what is the what is the air quality close close to where I live so in Leipzig for example there are only three places where such measurements are taken and all the other surrounding areas do not have any installed measurement place measurement stations so the question is if I'm living somewhere down towards the lower right what's the concentration of NO2 close to my home based on that I have formulated the requirements for citizen science measurement station so one is that it should be as cheap as possible because nobody wants to spend tens of thousands of euros to place such a station at home I mean it would be nice if you have so much money but well so you would like to have you would like to have good and free pieces components for such stations usually such the components you have are only available for industrial players but we want to be widely available it should be easy to build such a station and ideally it should be as precise as possible because eventually you might want to go to a government to a to an official spot and therefore we would like to stay within the framework of such orienting measurements of 25 percent of uncertainty so what are the components we have I have chosen as basis we have chosen esp 32 32 it has an integrated wi-fi four megabytes of flash space I have selected an m i c s metal oxide NO2 sensor costs about 16 euros available in the internet it just has it's available as a breakout board so you don't even have to solder just put it onto a plate i have chosen dh t 11 or dh 22 temperature and moisture sensor and the reason for this is because the the measurement you are taking are usually dependent quite a lot on temperature and moisture and therefore you need to measure those as well and find some methods to to to extract them from the from the measurements you have taken on top of that you need a few um resistance and that is it so that's such a measuring station on a simple um syncope um that's platin you don't really need to solder and so it's a very simple simple breakout board and so you don't need a pcb and so what's the the probe actually measuring and how is it working it's a resistive sensor so it's a metal oxide sensor and it changes its resistance with the change in concentration of NO2 in the air and there's a graph there for this from the datasheet and you can see that it has a double logarithmic scale and the NO2 concentration can be seen um here in this graph and there's an an influence of nitrogen monoxide if I know end of hydrogen and normally you don't have any hydrogen in your environment you shouldn't have it at least then you should move some rails um because hydrogen is very dangerous and explosive and the first thing I did was to look at the if the what the datasheet says is actually true in reality so what what does the sensor do actually so during during one month I ordered several sensors and tested them and this is the simple installation you can see this is a desiccator and normally it's used for drying things and the second is a gas mixer and have several different concentrations of NO2 and I mixed it with a with a zero gas without any two in it and the sensors or the measuring station in this container in this atmosphere and exposed it to this atmosphere and it showed that the the sensor they all had more or less the same the same reaction but shifted on the on the y-axis and this is so it's about the converter the analog digital converter output but you can see that you can live with the with this shift on the on the y-axis and it's it's pretty good because it makes a calibration much easier if you convert it to this to the scaling that has been given in the datasheet and it's more or less what you would expect or what you what what you would hope for if you look at the datasheet and then have the the results here and the next step were to filter out the influences of different different values for example the the temperature and if you have an influence of your temperature how do you do that if you have a if you test the thermometer thermometer where I work and then the thermal sensor you put it into a block and heat the block and let it cool down and you measure the temperature and compare it to to a reference thermometer and normally we don't have a turn don't have 250 degrees so the simplest solution was a styrofoam box with ice in it or with with cooling cooling elements and the the stations were were put into this cooling box and it was sealed hermetically and the concentration of the gases in the box shouldn't shouldn't change because it was really sealed and the only the only thing that changes over time will be the temperature and the result of this was a relatively linear connection and if you don't really want to measure NO2 then you can perhaps use it to measure temperature but there's a connection shown here between the the straight lines here they have they have more or less the same slope and then I did a calibration measurement to compare it with the analyzer shown at the beginning the chemical luminescence sensor and that's what it looked like and I pulled in air from the outside into the executor and through the executor and into the measuring system into the measuring device the measurement measuring device is shown as number one and the executor is number two and three and four are the computers recording the data and it showed during several days which are these clouds of points here the best NO2 influence had some some influence on the measuring value I chose some values where the temperature only was within a very small range and the moisture was also in a small range of values and the NOx concentration also were within within the same region so even during several days you you get this calibration function and if you correct the values for temperature for the from the function that I've measured before then it shows a little bit it looks a little bit better but it shows a big problem here there's this nitrogen monoxide and nitrogen dioxide are not available are not contained in a in a fixed ratio and NO is created in the engines and then changes into nitrogen dioxide but the influence of nitrogen monoxide monoxide is a reverse of that of nitrogen dioxide and it's pretty bad because now you one before really you use the measuring station you could how could we remove the nitrogen monoxide so this is the influence of temperature we looked at that again and it's very similar here too there's a cloud at the end this caused by the fact that the range of temperatures is is just a few ppb and yeah but it also showed here that the the connection to the temperature was the same as found out before and so I recap the problems here the the influence of NO2 and NO is reverse and it would be nicer if it were in the same direction then the NO influence would be much slower and NO2 and NO2 and don't always show up in the same ratio so if you have a measurement of x then you we need to have NO2 at 100 ppb and the other maybe maybe 80 ppm or ppb depending on what you have so but the ratio is not always the same so and since we have no no way of measuring NO separately there you can only really change NO into NO2 for example with an ozone generator from the aquarium supplies or with an oxidation with an oxidator like potassium permanganate and but this would increase the complexity and this is the problem that I'm working on at the moment well yes and if the measuring station works now the next step will be to create a measuring network if I've done the tests with the ozone generator then I will start creating the measuring network and so what are the requirements of the measuring network it has to be a very high availability and the sensors should ideally not break down after half a year it should have very very narrow meshes and so you can show temporal connections for example if you have a combustion somewhere and it creates a huge cloud of NO2 and then I want to see on the map how it how it spreads and so the difference the distance between measuring stations should be low and ideally so and you could then figure out defective sensors if they're close to each other and and you can put it into places that are far away from the official measuring stations and what is important is the it is to know the error of the measuring station and so no doubt has better than poor data so if you really want to do something with the measuring measured values you have to know what the error is of the values that you have measured and you need to have a way to to collect and retrieve the measured data okay and from these measurement data there are a couple of ways you can visualize them so for example the first one above is the heat map starting from a point which creates a colored surface other points in the surroundings color the surface appropriately depending on the measurement value another option is the hex pin map visualization on the lower left the surface is divided into regularly sized hexagons and they are colored depending on the concentration in the respective area a problem is if the measurement station is on a close to the border of such a hexagon but the rest of the hexagon actually has a low concentration and therefore this can this can change and and visually alter the representative visualization and the other option is to just pin appropriate spots with the measurement values which is easy to give an overview so I have programmed a platform for this which is called open knocks it's a web application with the with the instructions how to build the measurement stations well including the measurement station would have the problems in order to which doesn't have the ozone generator yet it has an option to register your own measurement stations and there's an api to send and retrieve the current and historical measurement data the source code for this measurement stations in the web applications are on github and if somebody has some idea to improve the way I analyze the measurement data I'm very happy to contribute right now there's also an api to include the official measurement stations of north rhein-west failure but unfortunately every state of germany here does use a different way to different way to provide the data and therefore it's very difficult to generalize this for the rest of data right now open knocks still links to the github repository because I have some trouble with the provider at the moment but that's how open knocks looks like you have the you have the map with the measurement data so this is mönchengladbach the green thing here is the official station on the street and the blue dots our own measurement stations which right now only provide raw data there's the instruction list with the component list to build the board because well building the board yourself and extending the board yourself I mean there are a couple of ways you could extend or include extra features for example ways that send the data automatically or something which stores the data on the usp stick and transfers the data at the data point outlook what what is there to be done so the biggest problem is the cross sensitivity with nitrogen monoxide that's one of the things where i'm currently working on so my idea right now is to is to work with ozone in order to completely completely change over the no what the nitrogen monoxide to nitrogen dioxide so the idea is that you can use this as input to calibration function to retrieve the raw you know to the complete raw you know to value of course another thing is to to spread the word that's what i'm doing right now and based on that build a whole measurement network and with this maybe identify other in or two producers which are currently unknown maybe in a way and of course the last but not least to extend the the board by additional sensors and additional functions for example to mobile to to be mobile and um mobily measure the data and send it over at some later points so that's pretty much it many thanks for for your attention yes and that has been the translation of the talk creation of a sensor network for measuring nitrogen dioxide and we were mf m what you called mf h and pink dispatcher there are two questions how many measuring stations are there already at open knocks and the other is if wouldn't it be better to have a better temperature sensor what that makes sense it's only a little bit more expensive how many are there well since we're just starting right now basically one one by myself the temperature sensor yeah the influence of the temperature is pretty much linear so a better temperature sensor yeah something one can use and maybe that's that's a useful uh in the future but what what you've seen so far it's efficient as it is so basically you're just uh you're just determining a zero for this measurement station and it's basically sufficient to only know that there has been a change by temperature by one by one degree and for the normal measurement stations where the temperature changes only about 20 degrees it's kind of sufficient so yeah but if there's something better why not it's from now from the internet please from the internet there was the question if an integration or cooperation with air Luftdaten.info because they're doing basically a similar thing and instead of having just so you could have one for NO2 and one for for aerosol and measure different different things yeah i i wrote to Darten i wrote to the mentioned website but i didn't really get an answer so somebody's waving so maybe that's the answer so we are from Luftdaten.info the three of us and we are working on on a theme like that so it was very very interesting thank you very much for the talk and we would really like to cooperate and we would like to announce that we have a sensor community that is a platform and it's only air data quality for aerosols but and it's also for a big european campaign as it has been successful in belgium with passive sensors and we are coordinating the institutions and it was a very good talk thank you very much and yeah yeah maybe we want to talk later okay you can you can continue this later continue with the questions do you know how stable sensors are in long range do they do they keep showing the same value of the same concentration in the air um there's there's a way that that poisons the sensor so some some some gases will damage the sensors permanently and make the measuring measuring much worse and i have been using one for five for five months and it didn't change a lot but i'm living on the second floor so it doesn't really get that many many pollutants as it would be on a on a big crossroads so it depends on on the individual situation so yeah it's hard to say maybe five months five months will be okay microphone number one please thank you very much for this very interesting talk we have seen in the beginning that the measurement curves all look very similar but i just slightly shifted do you have a good idea how one could how one could calibrate that simply without without having a 20 000 euro device to calibrate against yes and and if you if you assume that they may have a certain uh deviation a certain error then my hope is that at the at night there's perhaps there's 10 ppb around so five micrograms of of NO2 in the air or NO and then you create a zero point and then shift the zero point that's something that you could try to get a daily a daily recalibration to figure out if the sensor has has big fluctuations from what you would expect is there maybe please no dialogues you may answer of course you could go and and do a test gas to to use that so maybe it's if it's in a pot or in the atmosphere and but usually you don't have a test gas at home so maybe doing that at night and and creating the zero point from that is perhaps a good idea and it's good that they are only shifted on one axis so there there's a way to do something signal angel please why do you only want the measurement value for NO2 and not for NO2 and NOx and all nitrous gases problem at night nitrogen dioxide is especially problematic because it is is water soluble and it's one of the main solutions that causes the increased mortality and in these professional devices also measure nitrogen monoxide and nitrogen dioxide but this sensor can only can only measure one right okay a remark please please be please speak into the microphone properly microphone three hi thank you very much for your presentation for cryptography there are crypto parties so maybe we should have a calibration party for such sense microphone two please hi you are using resistive sensors alternatively one could use anti-resistive sensors for example for NOx and which are sensitive for NOx and NO2 at the same time and then the optical sensors which are only sensitive for NO2 so maybe for using a calibration chain one could calibrate the sensor no that wouldn't work and that's the reason is there are optical sensors but they cost 100 to 200 euros well yeah show me the one for 30 for 30 yeah okay after afterwards and there are these electrochemical sensors which also have cross sensitivities and are also more expensive 130 or 150 euros and that's why I decided to use this resistive sensor and the calibration chain of CO2 to NO2 is not possible in this way because it is in a single measuring cell perhaps but there are different measuring elements there need to be different elements for measuring to measure different values but but please talk to me afterwards next question microphone one please thank you for your presentation for at luftarten.info you have a couple of issues regarding the particle measurements with respect to moisture so I haven't really paid fully attention have you have you thought about that moisture and temperature sensitivity the problem is I don't have a climate chamber to to simulate different moistures and create different moistures I've tried to make comparison measurements with different moistures but unfortunately the moisture was very only in a very small range and very low and there was a there was a correlation but it wasn't really enough for for this and I have to make more tests to see how it how it is connected okay I feel there could be more problems question from the internet the question is not from the internet it's for myself the price of the station you have mentioned is it all the components yes it is wouldn't it be cheaper wouldn't it be cheaper for the mass usel in Germany to to instruct a pcb delivery deliverer couldn't that be also supported from the government such that it could be distributed to the people from the government yeah I don't really have any experience with getting grant money from the state but of course that would be an option to make a pcb if we would start mass production and you could may have pcbs made quite cheaply that's yes that's something you could do but I haven't done that yet a question for mic three please it's then influenced from the height where the measurement stations are placed with respect to the nitrogen content is there some influence with respect to where it's placed on the street and is there a way to calibrate that back yes that that's the case and that's a very important point if I have green plants at the street and I've measured behind the plants there is a big difference and there's a discussion that the measuring stations were too close to the roads and put them farther away because the distance has a very big influence and it's a really important point with the sensors it needed to be part of the information that do I something like I live on the second floor because the the air will be cleaner at higher elevations than at the ground and that's one additional information that needs to be part of the of the information next question mic two please I just wanted to comment I'm Matthias and me on the project which is called open air cologne and we are doing a sensor solution with a mix of ESP correlated to two locations also with respect to particle sensors it's also now used from a Helmholtz project to correlate this with health data maybe we should maybe we should talk later yes yes I'll be here afterwards for further discussion information is placed please please do it later but right now it's for questions please micro one please what do you think what do you think if you're additionally building an oxidation into the project is it then maybe possible to do a measurement with oxidation and without oxidation and using that you could you could have a ratio between NOx and NO2 yes that is exactly the plan and that is exactly what the professional devices do and they they will oxidize or first they measure NO alone because they selectively measure NO and then afterwards they measure NOx by by by oxidizing the NO to NO2 and then create the difference and if you have the NOx value and know how much NO and NO2 is there in in total then you can from the second value you can calculate back to to the first and you have to change all the NO2 into NO another question at mic 2 please how can you only measure NO the measurement of NO alone in these professional devices is it cheaper is it is is there resistive sensors only for you know nothing that i have found would be cheaper i might be wrong but i have not found anything that would be affordable any further question maybe from the internet signal angel no no questions here yeah okay that's it that was it from the that was it from the c3 lingo live from 33 c6 thank you very much on the mic where nfh and pink dispatcher