Healthy Buildings 2017 Europe July 2-5, 2017, Lublin, Poland

Transcription

1 Healthy Buildings 2017 Europe July 2-5, 2017, Lublin, Poland Paper ID 0224 ISBN: Prediction of airborne particulate matter concentration in underground stations using a two size class conservation model Edouard Walther 1, Mateusz Bogdan 1,*, Rudy Cohen 2 1 AREP, Paris, France 2 École Normale Supérieure - Paris Saclay (LMT Cachan), Cachan, France * Corresponding mateusz.bogdan@arep.fr SUMMARY This study presents an attempt of predictive modelling of airborne Particulate Matter (PM) in underground railway stations. The phenomenon is represented by a set of simple, ordinary differential equations that describe the apparent emission of particles in relation with train traffic, ventilation and deposition phenomena. The parameters of the equations are identified numerically by a genetic algorithm and comply with their respective expected order of magnitude. A quantitative comparison of numerical results with experimental data measured in one of Paris underground station exhibits a good accordance. A qualitative comparison with results from the literature also supports the model. KEYWORDS Box-model, identification, railway, PM 10, PM 2,5 1 INTRODUCTION Underground air quality is a growing concern as hundred millions people over the globe use subterranean public transportation. The World Health Organisation proposes a threshold of no more than 50 µg.m -3 PM10 for 24h, not exceeding 35 days a year. In railway stations, the majority of the aerosol is composed of metallic elements; however the uncertainty about adverse health effects remains, as summarized in the literature review by Gustafsson et al. (2012). This work hence addresses the problem of modelling the particulate matter concentration evolution in relation with train traffic. The underground context is complex, due to tortuous air flows and to emission rates that are not systematically characterized. In addition, the dynamics of PM differ by several orders of magnitude depending on their aerodynamic diameter. Simulating the behaviour of a heterogeneous aerosol hence requires considering several classes of particles (Riley et al., 2002). We therefore present here an extension of a previous work by the authors (Walther E. and Bogdan M., 2016), with two particle size classes. To date, few studies have coupled the train traffic to air quality. However a box-model approach was used successfully by Song et al. (2014) to quantify the evolution of CO2 concentrations in underground platform and concourse in Korea. Using a similar box-model approach, this work

2 is an attempt of predictive modelling of the PM concentrations in underground stations based on mass conservation of the PM0-2.5 and PM size classes. A B Figure 1. Weekly PM pattern; a) Outdoor PM2,5 and PM10 concentration (Paris Centre 2005 data), b) Underground PM10 and Train frequency at Paris Gare du Nord Station, EXPERIMENTAL DATA Outdoor air The outdoor air data was retrieved from Airparif (2005) online open database and averaged over a year so as to provide a typical week profile such as Fig. 1-a. Interestingly, the concentration minima exhibit a bell-shape profile (see dotted line). Underground PM concentrations The French National Railway Company (SNCF) made PM normalized data available to the authors. Measures have been taken by a gravimetric method at the Paris Gare du Nord Station in The data displayed on Fig. 1-b shows the normalized PM concentration and the theoretical train frequency, averaged over a year. A strong dependency between train movement and PM concentration can be observed, which lead to this modelling attempt. 3 PHYSICAL MODEL Considering the aerosol as mono-disperse is certainly insufficient to capture the dynamics of the real, multi-component particulate matter. Solving the conservation equation for each size class as in (Riley et al., 2002) was not possible due to the lack of available data. In the spirit of Nazaroff (2004) work, two size classes are considered in this model, representing fine and coarse particles. Let "a" be the subscript for particles with an aerodynamic diameter below 2.5 µm (PM2.5) and "b" the subscript for larger ones (PM2.5-10). The differential equation ruling the evolution of PM then reads as the sum of an apparent emission term, a ventilation term and a deposition term: = + = (1) + where N is the train traffic. For both size classes, the term stands for the apparent emission rate, for apparent ventilation and for deposition.

3 A stringent closure of the model would require including the quantity of particles deposited and resuspended from surfaces (Qian et al., 2008). As the particle mass deposited on surfaces is difficult to evaluate, this approach was put aside for the context of underground stations. Source term Direct emission and resuspension are considered to be the main phenomena at the origin of airborne PM. A separate identification of both terms is challenging: direct emission by wearing (abrasion of train parts such as catenary, braking pads, rail, etc.) is concomitant to resuspension linked to the energy generated by train movement. Train movements are essentially discrete events. However, we consider in the box-model that they lead to an average velocity within the enclosure. The first layers of deposited particles are subject to resuspension by kinetic energy, which is hence proportional to the square of train movement:. Ventilation The ventilation is split into two parts: base natural ventilation rate and train-driven ventilation! : = +! [vol/h], (2) where! =* +,-./0+ /* -,12 [-] is the ratio between the volume of air displaced when the train arrives/departs and the station one. An order of magnitude of! can be obtained by a Bernoulli approach as in (Wieghardt, 1962) or experimentally as in (Kim et al., 2004). For the geometry of the considered station it was estimated as! <1. Deposition 5 In still environments, the order of magnitude of the deposition rate,, is dependent on the particle diameter, which defines the underlying physical phenomena (Brownian diffusion, turbulent diffusion or gravitational deposition). Analytical derivations of this parameter can be found in (Crump et al., 1981; Park, 2000). The deposition rates in the domestic context were determined experimentally in (Thatcher et al., 1995). As underground stations are subject to important air velocities, a correction for the increase of deposition related to turbulent flows should hence be implemented in the spirit of (Lai and Nazaroff, 2000). This increase is mostly noticeable for very fine particles, and does not affect large ones significantly. In this first approach 6, 8 will hence be bounded between and 7 [1/h] as per (Nazaroff, 2004). Other physical phenomena occurring in polydisperse aerosol are neglected, e.g. particle size changes by condensation and coagulation as in (Park, 2002; Park, 2000) or thermal effects. 4 IDENTIFICATION PROCEDURE The numerical integration of Eq. (1) is performed by a Crank-Nicolson semi-implicit scheme. This method is less straightforward than a simple Euler integration scheme but has the advantage of both unconditional stability and a second-order temporal integration error. A sensitivity study over the time step showed that 60 seconds is a conservative value for integration. The initial conditions are given by the available PM10 measurements for the coarse particles concentration =0. As no measurements were available for the PM2.5, the initial concentration =0 was estimated after the PM2.5/PM10 ratio from (Fortain, 2004; Querol

4 et al., 2012). The influence of this initial condition is however reduced to the first hours simulated and does not propagate significantly afterwards. Using a genetic algorithm to minimise the average difference between simulated and measured results proved to be the most robust numerical procedure for the determination of the optimal parameters of the model. Prior to the global identification of parameters, an order of magnitude of natural ventilation and deposition is required. As train traffic stops overnight, the source terms of Eq. (1) are equal to zero and the concentration decrease is driven only by ventilation and deposition and, which simplifies the identification. Once the order of magnitude of the different parameters is known, the global identification procedure can be launched. The parameters are given bounds within ± 50\% variation around the value computed beforehand. The bounds of the parameters are following (Wieghardt, 1962), (Nazaroff, 2004): Piston effect:! <1, Deposition rates of the PM0-2.5: 0.01 : :1 h -1, Deposition rates of the PM2.5-10: 1 : : 7 h RESULTS Simulation of the PM2.5 and PM10 evolution The results of the identification are shown versus measurements on Fig. 2-a. The PM10 dynamics and amplitude are captured properly and show a good accordance around the concentration minima as well. Although no PM2.5 measurement data was available, the behaviour is very similar to the one observed by Querol et al. (2012) in two of Barcelona's underground stations, one of which is reproduced in Fig. 2-b. The identified parameters are summarized in Tab. 1, and comply with their respective bounds. A B Figure 2. a) Simulation results of PM2.5 and PM10 evolution versus measured PM10 for Gare du Nord station. b) Measurements of PM2.5 (black line) and PM10 (grey line) evolution for Barcelona Sagrera station (Querol et al., 2012). Evolution of the PM2.5/PM10 ratio Fig. 3-a shows the behaviour of the PM2.5/PM10 ratio obtained numerically. One can observe that this ratio drops to ~0.35 during train operation, which implies the coarse particles dominate. At night, ventilation and deposition of the coarse particles make the ratio reach approximately 1, signifying that mainly fine particles (PM2.5) remains. A B

5 Figure 3. a) Ratio PM2.5/PM10 for Gare du Nord station b) Measurements of the PM2.5/PM10 ratio at Paris Magenta station (figure from (Fortain, 2004)) The results are very similar to the one obtained during the experimental campaign of Fortain (2004) in Paris Magenta underground station, shown on Fig. 3-b. A similar ratio of ~70% PM10 during train operation was found by Salma et al. (2007) and Querol et al. (2012), which we believe should support the model. However this ratio may vary a lot, as reported by Sioutas (2011), from 0.23 in Stockholm to 0.88 in Los Angeles, depending on the type of train, the stations site construction completion, and other parameters. Table 1. Identified parameters for Paris - Gare du Nord train station Parameter Paris Nord Unit 6 8! [µg.m 3 ] [µg.m 3 ] [-] [vol/h] [h -1 ] [h -1 ] ; 11.8 [%] 6 CONCLUSION The mass conservation model with two size classes of particles shows a good accordance with experimental data. The dynamics and amplitude of PM10 concentrations are respected at high and low concentrations. However, as no measurements of the PM2.5 concentrations were available for our test cases, we compared the results with experimental data for Barcelona underground (Querol et al., 2012) and another station in Paris (Fortain, 2007). This qualitative comparison showed that the model and the experiment exhibit similar PM2.5 trends. Although the results are promising, this modelling approach has two main drawbacks: even though the order of magnitude of the piston-effect can be estimated analytically or empirically, a campaign is necessary to secure the results; in the model, the air induced by the piston effect is supposed to be outdoor air, which does not comply with reality. A measurement campaign at Paris Saint-Michel station in early 2017 will allow for a better understanding of the flow patterns linked to train movement. The model will be subsequently amended. Eventually, this method allows for an a posteriori evaluation of corrective actions to be undertaken, such as filtration and ventilation, which prove to be of little efficiency compared to reducing the train velocity (this has been underlined by (Park and Park, 2015)), as the apparent emission term depends on the square of velocity.

6 7 ACKNOWLEDGMENTS The authors would like to express their gratitude to the SNCF for allowing the use of normalized concentration data. 8 REFERENCES AirParif, 2005, Air quality at Paris 1 st district measurement station. [Online, accessed June 2016] Crump J.G., Seinfield J.H., 1981, Turbulent Deposition and Gravitational Sedimentation of an Aerosol in a Vessel of Arbitrary Shape, J. Aerosol Sci. 12(5), Fortain A., 2007, Caractérisation des particules en gares souterraines, Ph.D. Thesis, Université de La Rochelle Gustafsson M., Blomqvist G., Swietlicky E., Dahl A., 2012, Inhalable railroad particles at ground level and subterranean stations - Physical and chemical properties and relation to train traffic, Transportation Research Part D, 17, Kim S.D.,Song J.H.,Lee H., 2004, Estimation of Train-Induced Wind Generated by Train Operation in Subway Tunnel, Korean Journal of Air-Conditioning and Refrigeration Engineering,16, Nazaroff W., 2004, Indoor Particles Dynamics, Int. J. of Indoor Env. and Health,17(7), Lai, A. C. K., Nazaroff, W. W. (2000). Modeling indoor particle deposition from turbulent flow onto smooth surfaces. Journal of Aerosol Science, 31(4): Park S.H., Lee K.W., 2000, Method for Aerosol Deposition, J. Aerosol Sci., 31(suppl. 1) Park S.H., Lee K.W., 2002, Analytical solution to change in size distribution of polydisperse particles in closed chamber due to diffusion and sedimentation, Atmospheric Env., 36, Park J.-H., Park D. 2015, Method of removing Fe particulate matters from subway environments. International Journal of Environmental Monitoring and Analysis, 3(1), 1-6 Qian J., Ferro A.R., Fowler K.R., 2008, Estimating the Resuspension Rate and Residence Time of Indoor Particles, J. of the Air & Waste Management Association,58(4), Querol X., Moreno T., Karanasiou A., Reche C., Alastuey A., Viana M., Font O., Gil J., DeMiguel, E., Capdevila M.,2012,Variability of levels and composition of PM 10 and PM 2.5 in the Barcelona metro system, Atm. Chemistry and Physics, 12, Riley W.J., McKone T.E., Lai A.C.K., Nazaroff W., 2002, Indoor particulate matter of outdoor origin: importance of size-dependent removal mechanisms, Env. Science & Technology, 36, Salma I.,Weidinger T.,Maenhaut W.,2007,Time-resolved mass concentration, composition and sources of aerosol particles in a metropolitan underground railway station, Atm. Env., 41, Sioutas C., 2011, Physical and chemical characterization of personal exposure to airborne PM in Los Angeles subways and light-rail trains. University of Southern California, METRANS final report. Song J., Pokhrel R., Heekwan L., Kim S.-D., 2014, Box Model Approach for Indoor Air Quality Management in a Subway Station Environment, Asian J. of Atmospheric Env., 8(4), Thatcher T.L., Layton D.W., 1995, Deposition, Resuspension and Penetration of Particles within a Residence, Atmospheric Env., 29(13), Wieghardt K., 1962, Belüftungsprobleme in U-Bahn-und Autotunnels, Schiffstechnik, 9(49), Walther E., Bogdan M., 2016, A novel approach for the modeling of air quality dynamics in underground railway stations, Air Pollution 2017 Conf. Proceedings