OPEN HARDWARE, LOW COST, AIR QUALITY STATIONS FOR MONITORING OZONE IN COASTAL AREA

Transcription

1 EGU Atmospheric and Meteorological Instrumentation 4/28/214 OPEN HARDWARE, LOW COST, AIR QUALIT STATIONS FOR MONITORING OZONE IN COASTAL AREA Marco Lima Co-authors: Davide Donzella, Fabio Pintus, Adriano Fedi, Daniele Ferrari and Marco Massabò

2 Outline Official monitoring networks Why Ozone Air Quality assessment in the digital era Low Cost Gas sensors: Electrochemical and metal oxide technologies ACRONET paradigm: Infrastructure Sensor board prototype Data Analysis and Field calibration Temperature error correction Forward calibration test Conclusion and future trends 2/21

3 Air Quality Monitoring Networks Air pollution is monitored by networks of high reliability, static measurement stations referring both to standard air quality monitoring techniques, recommended by EU Commission and U.S. Environmental protection Agency, or to other equivalent measuring methods. These monitoring stations are bulky shelters with very high costs, both. in terms of purchasing and maintenance. All this led governments implementation of low monitoring networks. to the density 3/21

4 Why Ozone It is a good test-bench It has low spatial variability There are a lot of official monitoring stations measuring O3 O3 level is one of the parameters that should be monitored in AQ assessment scenarios Ground-level ozone is a toxic gas linked to lung diseases and is formed by reactions of precursor gases, such as carbon monoxide. While emissions of precursors fell by between 15 and 32%, during , ozone concentrations decreased only slightly, and between 14 and 65% (this range partly reflects variations caused by atmospheric conditions which differ between years) of Europe s urban population were exposed to levels above EU targets. The researchers also note that ozone is causing widespread damage to natural ecosystems and reductions in crop yields. EEA (213). Air quality in Europe. EEA Report No. 9/213: It affects other gas measurements 4/21

5 Sensor market analysis Principle of operation Certified Range Error Estimation Sensitivity to other Gas or T or HU Documentation of the sensor characteristic cost availability Calibration available Alphasense NO 4-electrode sensors -2 ppm <5 ppb H2S, NO2, Cl2, SO2, H2, CO, NH3, CO2, O3, Halothane 77 Performance data available Alphasense O3 4-electrode sensors >2 ppm <3 ppb H2S, NO2, Cl2, NO, SO2, H2, CO, NH3, CO2, 84 Performance data available Alphasense NO2 4-electrode sensors -2 ppm <3 ppb H2S, NO, Cl2, SO2, C2H4, CO, NH3, CO2, O3, Halothane 77 Performance data available Alphasense CO 4-electrode sensors -5 ppm <1ppb H2S, NO2, Cl2, NO, SO2, H2, NH3, C2 H4 85 Performance data available SGX EC4-2-NO Electrochemical -2 ppm.5 ppm TBA 95 $ Not in stock - SGX EC4-25-NO Electrochemical -25 ppm.5 ppm SO2, NO2, CO, H2 93 $ 4 - SGX MICS 2614 O3 Metal Oxide Semiconductors 1 1 ppb 7$ 11 SGX EC4-2-NO2 Electrochemical -2 ppm 95 $ Not in stock - SGX MICS 2714 NO2 Metal Oxide Semiconductors,5-1 ppm 7$ 11 SGX MICS 4514 CO Dual Metal Oxide Semiconductors 1-1 ppm 6$ 747 SGX MICS 5524 CO Metal Oxide Semiconductors 1-1 ppm 5$ Not in stock Shinyei Model PPD42NS Particle Sensor Optical >1µm 4 (not sufficient) Qbit PM1 station laser 1-1 µg/m3 35 (?).1 ppm H2S, NO, Cl2, SO2, CO 5/21

6 Low Cost AQ Sensors: Metal Oxide vs Electrochemical METAL OXIDE ELECTROCHEMICAL Electronics Heating circuit Potentiostat Transimpedence amplifier Power usage Low Very low Stabilization time 1h 1h to 1day Sensitivity to wind exposure es (may cause cooling effect) No Sensitivity to sun exposure es es Weather proof es es Temporal resolution Fine enough Fine enough Sensitivity to T and RH es es 6/21

7 MiCS-2611 Raw Data MiCS O3 mean - Temperature MiCS [ ADC count ] Temp [ C ] - O3 [ ug / m3 ] 1 MiCS2611 Temperature [ C ] O3 [ ug/m3 ] / 3 / 214 No calibration curve provided ADC output analysis 7/21

8 Low Cost AQ Sensors: Electrochemical Low Power Robust Low Cost 4 electrodes electrochemical gas sensor Alphasense O3 B4 vs Temperature 375 Alphasense O3-B4 offset : 6: 12: 3 / 9 / : Temperature [ C] V work : voltage on working electrode V aux :voltage on auxiliary electrode V workzero :V work at ppb ozone concentration V zuxzero :V aux at ppb ozone concentration S : sensitivity 35 O3-B4 [mv] (V V workzero ) (V aux V auxzero) C O3= work [ ppb ] S 5 Vaux Vwork Temp 1 : 8/21

9 Air Quality assessment in the digital era 9/21

10 The ACRONET Paradigm: The infrastructure 1/21

11 The ACRONET Paradigm: AQ sensor board ACROSTATION r1a Acroboard Station V1 r8a Solar Panel Lead Acid Battery Acroboard Power V1 r4a Antenna Enclosure TOTAL ~ 3 SENSOR BOARD Sensors: Alphasense O3-B4 (O3) SGX MiCS-2611 (O3) Sensirion SHT25 (T and RH) Enclosure TOTAL ~ /21

12 Sensor Network Field calibration Idea: Real time estimation of parameters for linear calibration. INTEGRATION WITH OFFICIAL MONITORING NETWORKS Why? Official AQ Station Low cost AQ Station Official and low Cost stations coupled AQ data Simple and Low cost Follows seasonality Compensates sensor degradation AQ and calibration data But... How frequently? What is the maximum distance 12/21

13 Test Description and Data processing Experimental measurement test: Date: 214/3/4 214/3/26 Measures: O3 (x2) Temperature Raw electric data Relative Humidity Official AQ monitoring: ADC Average on 2 sequential measures f = 8 khz [2 min] Send data to server Photometric technology Sample rate: 2 min Aggregation rate: 1 h Human data post-validation (24h) Estimated O3 concentration Calibration [1 min] Moving average 1 hour 13/21

14 O3-B4 Linear Calibration 1/2 (V V workzero ) (V aux V auxzero) C O3= work [ ppb ] S O3=Δ V α S +βo Factory Calibration: estimated O3 concentration O3: Δ V :V work V aux α S :(O3 B4 sensitivity) 1 βo :O3 B4 offset αs = [ ppb / mv ] βo = [ ppb ] ΔV vs. O3 Linear calibration f(x) =.74x R² =.252 Factory linear calibration f(x) =.239x R² = O3 [ppb] 4 O3 [ ppb ] O3-B4 [ ppb ] Vwork - Vaux [mv] /21

15 O3-B4 Linear Calibration 2/2 Linear calibration CO3 vs. ΔV trend line: Factory linear calibration vs. New linear calibration f(x) =.74x R² = New Linear Calibration: 1 αs =.79 [ ppb / mv ] O3 [ ppb ] 75 βo = [ ppb ] 5 Approximation Error ( O3-B4 - O3 ) Error / O3 Approximation O3-B4 [ ppb ].5 O3-B4 new param. [ ppb ]. -.5 Linear (O3-B4 new param. [ ppb ]) -1. O3-B4 alphasense [ ppb ] O3 [ ppb ] Linear (O3-B4 alphasense [ ppb ]) f(x) = 1.x -. R² =.252 f(x) =.239x R² = /21

16 O3-B4 Temperature correction O3 B4 is a reducing sensor for which the ISB (Individual Sensor Board) is reversing the output, thus with positive temperature changes you are seeing negative value changes. Linear Calibration (new parameters) R2 =.252 Linear temperature correction Linear Temperature Correction R2 = Approximation Error vs. Temperature Mean SV-Q (O3meanSVQ Approximation - O3-B4)/O3meanSVQ Error 6 f(x) = -.2x +.34 R² = O3-B4 Lin Temp correction O3 [ ppb ] new param Temperature [ C ] 16/21

17 Calibration Results Official O3 vs. O3-B4 New Linear Calibration with temperature error correction O3 [ ppb ] Timeline /3/214 O3 [ ppb ] alphasense Official O3 [ppb] Lin Temp correction - 17/21

18 Forward Calibration test Official O3 Data DATASET Training Set: 3-14/3/214 Test Set: Test Set Raw Data ( O3 + T ) Training Set Raw Data ( O3 + T ) 14-25/3/214 Calibration line with temperature correction Calibration line with temperature correction Estimated O3 R2.515 O3 [ppb] 5 4 mean [ppb] Lin Temp correction Test Set Timeline 18/21

19 AQ Research Project CIMA Research Foundation & La Spezia Port Authority started a new experimental campaign near a trusted AQ station (~ 5m): Objectives of the research are the implementation and evaluation of a lower cost AQ sensor node. Sensors: Alphasense O3-B4 (Ozone) QbitPM1 (PM1) DS18B2 (Temperature) x2 19/21

20 Conclusions and future trends CONCLUSIONS Two O3 low cost sensors were tested New field linear calibration with linear temperature error correction were implemented FUTURE TRENDS Test experimental field calibration for a longer period Test also nonlinear calibration Look for peaks and low level different calibration curve Test other lower cost Air Quality sensors (NOx, CO, PM1, PM2.5) 2/21

21 Thanks for your attention Working on the Shoulders of Giants /21