Modeliranje aerozagađenja od drumskog saobraćaja - identifikacija uticajnih parametara i izbor modela

Abstract

Ova disertacija predstavlja rezultat višegodišnjeg istraživanja emisije vozila i njene disperzije unutar gradskih oblasti. Tokom pomenutog istraživanja je utvrđeno da vozni park Republike Srbije ima veći procenat vozila koja veoma zagađuju okolnu sredinu nego što je to slučaj sa voznim parkovima zemalja u okruženju. Dodatno, emisioni faktori dodeljeni zemljama zapadnog Balkana od strane Evropske agencije za zaštitu životne sredine, do danas nisu bili predmet istraživanja u našoj zemlji, već su prilikom proračuna nacionalnih emisija usvajani bez provere. Sve je to uslovilo realizaciju četiri nezavisna eksperimenta (merenje emisije, brojanje saobraćaja, merenje brzine vozila i merenje imisije) i testiranje više poluempirijskih modela disperzije kako bi se utvrdili uzroci povećane emisije, precizirali realni emisioni faktori voznog parka naše zemlje i dao odgovor na pojedina aktuelna pitanja naučne zajednice oko doprinosa i uticaja komponenti koje ključno utiču na kvalitet urbane atmosfere. Na početku disertacije je predstavljena osnovna hipoteza koja je za cilj imala proveru pitanja:,, Da li bolje ocenjeni model za procenu kvaliteta vazduha jedne specifične gradske strukture predstavlja kvalitetan i pouzdan simulacioni alat ?“. Za to je bilo neophodno u pregledu literature jasno predstaviti ekološke probleme jedne savremene gradske sredine, sa naglaskom na štetnu emisiju vozila. Materijal i metod teorijskog istraživanja emisije vozila i njene disperzije unutar ulica oblika kanjona čine važnu okosnicu trećeg poglavlja, pri čemu su objašnjene i statističke metode radi kasnijeg dokazivanja navoda iz osnovne i dve radne hipoteze. Takođe su u okviru ovog poglavlja objašnjeni uslovi eksperimentalnih postavki, kao i specifikacija merne opreme. Zatim su predstavljeni rezultati merenja emisije, a primenom binomne logističke regresije na mernom skupu podataka, utvrđeno je da starost vozila jeste najuticajnija promenljiva na povećanje emisije, da su najveći zagađivači domaća vozila, a u opštoj oceni to su vozila male snage motora. Veoma je važno da je štetna emisija pod kontrolom tokom čitavog veka eksploatacije vozila i u svetu često korišćen pristup za rešavanje ovog problema jeste program inspekcije i održavanja koji može u relativno kratkom periodu eliminisati emisiono neispravna i/ili vozila koja veoma zagađuju okolnu sredinu. S obzirom da se on ne primenjuje u Republici Srbiji, priprema i uvođenje jednog takvog programa bili bi mnogo jednostavniji ukoliko bi se za njegovu implementaciju iskoristili zaključci ovog istraživanja. S obzirom da do danas nema jasnih dokaza koji pokazuju da složeniji modeli sistematski bolje operišu u pogledu prognoze greške od manje složenih modela, testirana je prva radna hipoteza koja je dala odgovor na pitanje o boljoj proceni imisije između jednog složenog i drugog, manje složenog, modela disperzije. Pri tome je njihova preciznost procene ograničena tačnošću ulaznih podataka, a to su emisioni faktori, meteorološki podaci, brzina vozila i procenat hladno startovanih motora. Analiza je pokazala bolje procene složenijeg modela, međutim, pošto je najbolji rezultat koeficijenta determinacije procenjenih i realnih imisija 0,65, možemo reći da je još uvek prisutno 35% nepoznatih uticaja. Takođe, analiza je pokazala da složeniji model ne uzima dovoljno dobro u obzir režime kretanja vozila na količinu emisije u oblasti signalisanih raskrsnica. Ukoliko se izdvoje rezultati dobijeni pri uslovima delovanja vetra iz okolne sredine pod uglom manjim od 30° u odnosu na simetralu ulice, prostiji model disperzije ima bolje procene. Nakon izbora i adaptacije složenijeg modela, odnosno dokazivanja druge radne i osnovne hipoteze, usvojeni su odgovarajući emisioni faktori voznog parka Republike Srbije postupkom dedukcije. Ovakavim pristupom izbegavaju se skupa i komplikovana istraživanja koja zahtevaju velika materijalna sredstva, dosta vremena, kao i složenu mernu opremu. S obzirom da se količina emisije često procenjuje za nacionalne teritorije, to bi greške neprecizno usvojenih emisionih faktora bile još izraženije usled nemogućnosti sagledavanja svih promenljivih prilikom eksperimentalnog merenja. Na kraju ovog istraživanja je simulacijom imisije izvršeno rangiranje procentualnog uticaja njenih komponenti (pojedinačnih i združenih), pri čemu je bilo potrebno formirati potpune višefaktorske planove i primeniti metodologiju odziva površine. Time je pružen deo odgovora u okviru aktuelne polemike naučne zajednice oko uticaja i doprinosa komponenti imisije.

This study resulted from the years-long research on vehicle emission and its dispersion within streets canyons. During the mentioned research it was found that a fleet of Republic of Serbia has a higher percentage of high polluting vehicles than is the case with the fleets of neighboring countries. In addition, emission factors assigned to the Western Balkan countries by the European Agency for Environmental Protection, have not been the subject of research in our country and that produce that calculated national emissions were adopted without checking. All that caused the realization of four independent experiments (emission measurement, traffic counting, vehicle speed measurement and immission measurement) and test more semi-empirical dispersion model to determine the causes of increased emissions, refine real emission factors of the Republic of Serbia fleet and gave a response to current questions of the scientific community about the contribution and impact of the key components that affect the quality of urban atmosphere. In the general outlines of the dissertation, the Introduction presents the null hypothesis which aimed at reviewing questions referring to whether better estimated model for air quality assessment in a specific urban structure represents a high quality and reliable simulation tool. Thus it was necessary that the literature review clearly presents environmental problems of a modern urban area, with special emphasis on harmful vehicle emissions. Materials and methods of theoretical study on vehicle emission and its dispersion within street canyons make an important backbone of the third chapter, where statistical methods are also explained in order to prove the assumptions from null hypothesis and two working hypotheses. Also this chapter explains the experimental setup conditions, as well as the specification of the used measuring equipment. The very beginning of the fourth chapter presents the results of emission measurements, and by applying the binomial logistic regression model to the measuring set of data, it was found that the vehicle age is the most influential variable for emissions increase, that domestic vehicles are the biggest polluters, and in general assessment they are vehicles with low-power engines. It is very important to minimize harmful emissions throughout entire vehicle lifetime and the approach for solving this problem worldwide is inspection and maintenance program because it can, in a relatively short period, eliminate highly polluting vehicles. This program is still not applicable in the Republic of Serbia, so the preparation and introduction of such a program would be much easier for its implementation if conclusions of this study were applied. So far there is no clear evidence that complex models systematically operate better in terms of error prediction compared to less complex models, so we tested the first working hypothesis which gave an answer to the question related to better immission assessment between the complex dispersion model and the less complex one. Also it must be noted that their estimated accuracy is limited by accuracy of input data, such as emission factors, meteorological data, vehicle speed and the percentage of cold-start engines. Analysis showed better concurrence of complex model with real immission, but since the best determination result of estimated and actual immissions is 0,65, we can say that 35% of variability is still unknown in the real immissions results. Therefore, if we take measurements position into account, we can conclude that complex model does not consider sufficiently the influence of vehicles regimes on the emissions quantity, especially in terms of light-controlled intersections area. If we distinguish the results obtained when background wind blows at an angle less than 30° relative to the street centerline, the less complex model, has a better estimation. After the selection and adaptation of complex model, that is after proving the null hypothesis and second working hypothesis, we adopted realistic emission factors of the Republic of Serbia’s vehicle fleet by deduction process. This approach avoids expensive and complicated studies that require significant financial resources, a lot of time as well as complex measuring equipment. As emission estimations are often determined for the national territory, errors of inaccurately adopted emission factors would be even bigger due to inability to comprehend all variables in experimental measurement. At the end of this research the immission simulation was carried out in order to rank the immission components by influence percentages (individual and corporate), and for that purpose it was necessary to form full multi-factor plans and implement the response surface methodology. This provided some of the answers within actual polemics of wider scientific community about the impact of the immissions components.