APR-11-004

7/18/2019 APR-11-004

http://slidepdf.com/reader/full/apr-11-004 1/10

Atmospheric Pollution Research 2 (2011) 24‐33

Atmospheric Pollution Researchwww.atmospolres.com

Atmospheric

dispersion

modelling

of

the

fugitive

particulate

matter

from

overburden

dumps

with

numerical

and

integral

models

Konstantinos E. Kakosimos 1, Marc J. Assael

1, John S. Lioumbas

2, Anthimos S. Spiridis

2

1 Thermophysical Properties Laboratory, Department of Chemical Engineering, Aristotle University of Thessaloniki, Thessaloniki GR‐54124, Greece2 Spiridis A. ‐ Koutalou B. LLP – ΗΥΕΤΟS, 3 & 6, Navarinou Square, Thessaloniki GR‐54622, Greece

ABSTRACT

The present work addresses the problem of atmospheric dispersion of particulate matter (PM) from the overburden

dumps of a mine, using a steady–state Lagrangian numerical model (commercial CFD software) and the integral model AERMOD (U.S. EPA). The analysis includes most of the complex physical phenomena in atmospheric diffusion.

In the vicinity of the city of Amyndaion in Northern Greece, there is a large mine that provides lignite for the thermal

power stations of the Hellenic Public Power Corporation. The excavated land is dumped in nearby open pits, which

are planned to be extended towards South. These pits are sources of air–suspended particulate matter that can

affect the nearby residential areas. The numerical model was applied for a number of specific meteorological

scenarios, while the integral model for five years period. The results from both models showed an increase on the

concentration levels, near the residential area, of two orders of magnitude if the dumps expand Southern. The

models inter–comparison was based solely on the average yearly concentration and showed a fair agreement.

Nevertheless, the numerical model tends to underestimate the concentration levels mainly because of the “limited”

number of particles employed in the Langrangian model and the lack of wind–meandering.

Keywords:

Dispersion Particulate matter

Mine

Modelling

Pollution

AERMOD

Article

History:

Received: 03 May 2010

Revised: 15 July 2010

Accepted: 01 August 2010

Corresponding Author:

Konstantinos E. Kakosimos

Tel: +30‐2310996170

Fax: +30‐2310996170 E ‐mail: [email protected]

© Author(s) 2011. This work is distributed under the Creative Commons Attribution 3.0 License. doi: 10.5094/APR.2011.004

1. Introduction

Lignite (brown coal) is the most important energy mineral raw

material of Greece. Lignite exploitation has a highly significant contribution to the development of the energy sector in Greece during the last 50 years, and will have, according to estimations, continue to be the primary source of energy for another 40 years, as Greece is very rich in lignite resources. The main basins – from

where lignite is extracted by opencast mining – is in northwestern

Greece (Eordea mountain basin), where 70% of the electricity of the country is generated. Despite the fact that the European Union puts strong emphasis on the use of renewable energy resources, Greece is still investing on lignite by building new power plants and mines (Markakis et al., 2010). Unfortunately, although many measures are taken for a cleaner energy production during lignite burning, very few are taken in the mining of lignite. Mining operations have always generated substantial quantities of airborne respirable dust, which led to the development of lung diseases in mine workers and inhabitants of the surrounding area.

Many computer models have been developed for predicting pollutant dispersion or dust generation. However, despite the evolution of computer processors and complex numerical algo‐rithms, most of the models are based on empirical formulae, which

take as input the wind flow or perform some coarse calculations and then use statistical correlations for predicting dust dispersion. One of the main reasons is that the application of complex numerical models, based on Navier–Stokes equations, requires

many input parameters as well as the knowledge of simulating the atmospheric boundary layer in a realistic mesoscale environment. Present work demonstrates an integrated methodology, that employs a well–known commercial CFD software, which is based on the solution of Navier–Stokes equations, to model the atmospheric phenomena of wind flow and dust diffusion. The target area is an overburden dump of one of the lignite mines in the Eordea valley near the city of Amyndaion in Northern Greece.

Today the modelling of the pollutant’s dispersion is achieved by several basic mathematical algorithms: the box model, Gaussian model, Eulerian model, and Lagrangian model. The box model is the simplest of the modelling algorithms (Lettau, 1970). It assumes that the air shed is in the shape of a box of homogeneous concentration. Although useful, this model has limitations. The Gaussian models (Pasquill, 1971), being the most common mathematical models used for atmospheric dispersion, assume that the pollutant will disperse according to the normal statistical distribution. Eulerian models solve the conservation of continuity momentum and mass equation for a given pollutant. The wind field vector, which is normally used, is considered turbulent and it also affects the pollutant concentration. The direct solution of the governing equation is demanding and for this reason various approximations of the turbulent characteristics of the flow are

incorporated like, k–ε, k–ω, RNG, LES, etc. Lagrangian models predict pollutant dispersion based on a shifting reference grid, generally based on the prevailing wind direction, or vector, or the general direction of the dust plume movement. These models are

7/18/2019 APR-11-004


Kakosimos et al. – Atmospheric Pollution Research 2 (2011) 24‐33 25

usually appropriate for the simulation of dust dispersion having the discreet form of the pollutant (a number of particles with specific size distribution).

In most air quality applications, one is concerned with dispersion in the atmospheric boundary layer (ABL), the turbulent air layer next to the earth's surface that is controlled by the surface

heating and friction and the overlying stratification. The ABL typically ranges from a few hundred meters in depth at night to 1 – 2 km during the day. So the problem of atmospheric dispersion involves mainly two phenomena, the flow of the wind and the dispersion of the pollutant. One can simulate the ABL and calculate the velocity components or use empirical equations that approxi‐mate the phenomena in the ABL. The last approach constitutes the main advantage of the empirical models against the numerical ones (Eulerian, Lagrangian). For these reasons, the empirical models are the most commonly used ones (box or Gaussian) and as they can be applied with a minimum set of required data and in a short time. The list of empirical models (Meroney, 2004) is very large and covers a wide area of applications (ISC3, CALPUFF, AERMOD, SLAB, DEGADIS, FDM, etc). In addition, most of them are accompanied by many validation tests and are recommended by

national environmental organizations (e.g., U.S. Environmental Protection Agency). Despite the above advantages of the empirical models, if a more detailed analysis is required or the terrain is quite complex, their application is risky and restricted and thus a more sophisticated CFD model certainly has to be employed.

CFD is a numerical analysis method employed to solve fluid flow problems with a computer. This method generally follows an Eulerian approach applied to the airflow modelling (Anderson, 1995). However, it can also incorporate a Lagrangian algorithm for the modelling of particles subjected to forces of gravity and airflow. In such cases, the forces (gravity and airflow) are based on an Eulerian reference frame, whereas the Lagrangian algorithm is used to characterize the advection and diffusion processes that occur among the individual particles and influence each particle's

trajectory. CFD algorithms (Ansys CFX, Fluent, etc) have been extensively used and have been validated on simulations in the lower ABL (Prospathopoulos and Voutsinas, 2006; Blocken et al., 2007), usually near the ground and over obstacles (i.e., buildings). According to the authors' best knowledge, the application of a CFD algorithm (like CFX, Fluent) in a large area (15x15 km), like the valley of the city of Amyndaion, rarely appears in the literature (Moussafir et al., 2010).

The main objectives of the present work are: (i) to present a methodology to estimate the atmospheric dispersion of the fugitive particulate matter (PM) from overburden dumps with a steady state Lagrangian numerical model, (ii) to estimate the influence of the expansion of the dumps to the local air quality, and (iii) to evaluate the results of the numerical model with an

integral one.

First, the methodology is described, in more detail: the topographical and meteorological data, the numerical model of the commercial CFD software (ANSYS CFX) used, the calculation procedure of the emission rate and the integral model AERMOD (U.S. EPA). Then the results are presented concerning the objectives of the paper and finally, the conclusions of this work are explained in the same context.

2. Methodology

The Eordea Mountain Basin is a highly active industrial area in the northwestern part of Greece, which suffers from major air quality problems. The problems are mainly caused by lignite–fired

power stations and lignite mining operations in the basin. The most significant pollutants emitted in large quantities are suspended particles and SO2. The influence of the presence of power stations to the regional air quality has been studied (Triantafyllou et al.,

2001; Triantafyllou and Kassomenos, 2002; Triantafyllou et al., 2002), while a recent survey (Sichletidis et al., 2005) in the area has directly related the diseases of the respiratory system, especially those of children, to the high levels of air suspended particulate matter. However, there is no study available to the public, regarding the influence of the surrounding mines and especially the dump sites on the local air quality. Furthermore, it is a

common belief that the spreading of the mines and the dump sites is unrestrained and takes place without the necessary environ‐mental studies. For this reason, this work studies the present and future influence of a specific overburden dump near the city of Amyndaion (40.69°Ν 21.68°Ε).

2.1. Topography

The valley of Amyndaion (part of the Eordea mountain basin) is characterized as a broad, relatively flat bottomed basin surrounded by high mountains with heights ranging from 800 to more than 1 500 m above mean sea level (MSL) (Figure 1). The sides and the floor of the valley are partly wooded, covered by isolated trees, small bushes and rock outcrops, while there are also dispersed cultivated areas. The mean height of the valley above

MSL is approximately 650 m. The valley is 40 km long, while its width varies between 10 and 25 km. Its main axis is from NW to SE, with a steep slope from SE to NW. On the E and NE sides there are two lakes, whose water volume is constantly decreasing. The topographic complexity and the variety of physical–geographical characteristics in the valley are expected to induce local circu‐lation, such as anabatic/katabatic flows. There are a large number of villages spreading all over the valley. Figure 2 illustrates the digital map of the area that was used (see also Figure 1); the circular region represents the outer limits of the boundary divided into eight sectors, one for each wind direction. Based on this map a structured mesh of hexahedral elements was generated. The mesh density (∼2.5 million cells, 30 layers at the vertical direction, height of first cell, ∼0.5 m, average expansion ratio, 1.2) was selected following a series of tests with various mesh densities. This is

enough to describe the terrain in detail and stabilize the mathe‐matical solution of the problem with the minimum computing power.

Figure

1. A satellite image of the valley of the city of Amyndaion. The area

inside the black frame is the one that was used at the calculations (see

Figure 2).

7/18/2019 APR-11-004


26 Kakosimos et al. – Atmospheric Pollution Research 2 (2011) 24‐33

Figure

2.

Part of the digitized valley and the calculation mesh (see also

Figure 1). The circular region represents the outer limits of the boundary

divided into eight sectors one for each wind direction.

2.2. Meteorology

The necessary meteorological data originate from the municipal meteorological station (Class A – City of Amyndaion).

The data studied covers the years from 1963 to 2007. The local climate is characterised by the low temperatures and frequent snow falls during winter, the extended rain falls during autumn and a short dry period during summer. In more detail, the mean annual precipitation is around 416 mm with two discrete rain periods one during late autumn and one during late spring. The driest month is August with mean precipitation around 22.4 mm. The mean temperature of the area is 12.3 °C while the hottest month is July (22.3 °C) and the coldest one is January (2.4 °C).

A detailed statistical analysis was performed for the years from 2003 to 2007 to identify the most frequent meteorological conditions concerning wind speed and direction. This analysis revealed that the most frequent wind blows from North (∼18%) while the next three from Southeast, Northwest and West (5–10%) (Figure 3). Moreover, the strongest winds blow from North and Northwest, while there are often calm periods (<1 m s

–1 around

15%). The results from the statistical analysis were used to form

nine meteorological cases/scenarios to represent some of the most frequent winds which are also transferring PM from the dumps to the nearby city. The formed scenarios are presented in Section 3 (Table 1). Furthermore, to strengthen the creditability of the data, they were cross checked with simulated data from the National Oceanic and Atmospheric Administration of U.S. (GDAS, 2008). These data are on a three–hour base and available at a resolution of 100x100 km. Both sets are in full agreement, and thus the GDAS data were also used in cases where no data were available at the meteorological station. Specially, simulated vertical soundings were used to estimate the characteristics of the boundary layer at

present work’s numerical model. The mixing height and the stability class also originate from the GDAS data. The boundary layer depth ranged from a few hundred meters up to 1 500 meters while the stability class was mainly stable or neutral. It should be mentioned that the numerical model employed to simulate the dispersion only under neutral conditions. Therefore, a mixing height around 500 m was selected. On the other hand the integral model is able to take into consideration all stability classes,

therefore detailed meteorological data of five years (2003–2007) were used for the hourly calculations.

2.3. Numerical Model

Theoretically, the Navier–Stokes equations, in their native form, have the ability to describe either laminar or turbulent flow.

However, the solution of these equations for a real problem is quite demanding and requires enormous computing power. For this reason, turbulent models were developed that describe indirectly the turbulent effects mainly with a statistical approach. The turbulent model employed in the present work and used in the commercial CFD software (ANSYS CFX) was the k–ε model. This model was chosen among others because it is considered to estimate the atmospheric boundary layer more accurately (Hargreaves and Wright, 2007).

Figure

3. A wind rose diagram for the year 2006 (wind directions are

“blowing from”).

k–ε approach. The k–ε model introduces two new variables into the system of equations the turbulent kinetic energy, k (kg m

–2 s

–2),

and the turbulent dissipation rate, ε (m2 s

–3). A brief description of

the equations of the model follows. Thus the governing equations are the continuity equation:

(1)

and the momentum equation:

( )

( ) ( ) ( )

T

eff eff

U U U U p U B

t

ρ ρ μ μ

∂+ ∇ • ⊗ − ∇ • ∇ = −∇ + ∇ • ∇ +

∂

(2)

eff t μ μ μ = + ,

2

0.09=t

k μ ρ

ε (3)

Table

1. Meteorological scenarios and average calculated PM10 concentrations (at Amyndaion)

Concentration

(μg m‐3

)

Wind Speed (m s‐1

)

1.5 2.5 3.5

Sources 0 1 2 3 0 1 2 3 0 1 2 3

Wind Direction

N 0.37 0.40 0.40 0.40 0.15 0.19 0.21 0.3 0.15 0.12 0.07 0.11

NNW 0.39 0.53 0.60 0.77 0.05 0.13 0.32 0.59 0.04 0.07 0.15 0.32

NW 0.00 0.10 0.46 1.38 0.0 0.01 0.76 1.71 0.0 0.01 0.21 0.36

( ) 0U t

ρ ρ

∂+ ∇ • =

∂

7/18/2019 APR-11-004



where ρ (kg m –3

) is the fluid density, U (m s –1

) is the velocity field, t (s) is time, B is the sum of body forces, μeff (kg m

–1 s

–1) is the

effective viscosity accounting for turbulence, p (Pa) is the modified pressure and μt (kg m

–1 s

–1) is the turbulence viscosity. The values

of k and ε come directly from the differential transport equations:

( ) ( )1.0

t

k k Uk k P

t ρ μ ρ μ ρε ∂ ⎡ ⎤⎛ ⎞+ ∇ • = ∇ • + ∇ + −⎢ ⎥⎜ ⎟∂ ⎝ ⎠⎣ ⎦

(4)

( )( ) ( )1.44 1.92

1.3

t

k U P

t k

ρε μ ε ρ ε μ ε ρε

∂ ⎡ ⎤⎛ ⎞+ ∇ • = ∇ • + ∇ + −⎢ ⎥⎜ ⎟∂ ⎝ ⎠⎣ ⎦

(5)

( ) ( )2

33

T

k t t kbP U U U U k Pμ μ ρ = ∇ • ∇ + ∇ − ∇ • ∇ • + + (6)

where Pk is the turbulence production due to viscous and buoyancy forces, while Pkb is the buoyancy production rate (usually with the Boussinesq model; Cotton, 1997). Finally the energy equation is:

( )

( ) ( ) ( )tot

tot M E

h p

Uh T U U S S t t

ρ

ρ λ τ

∂ ∂

− + ∇ • = ∇ • ∇ +∇ • • + • +∂ ∂

(7)

where SE is a source of energy, htot (kg m2 s

–2), is the total enthalpy

while the term U•SM accounts for the external sources of momentum and it is usually neglected. More details about the model equations can be obtained in the cited literature (e.g. Ansys, 2006).

Particulate matter dispersion. For the simulation of the particle dispersion, the Lagrangian approach was preferred instead of the Eulerian one, because it calculates concentration distribution more accurately (Loomans and Lemaire, 2002) and takes into account effect of obstacles more efficiently (Riddle et al., 2004). Unfortu‐nately when dealing with real problems a very large number of virtual particles is needed. The particle displacement, x (m), is calculated using forward Euler integration of the particle velocity, over a time–step δt as:

n o o

i i pi x x v t δ = + (8)

where the superscripts o and n refer to the old and new values respectively, and v (m s

–1) is the particle velocity. In forward

integration, the particle velocity calculated at the start of the time–step is assumed to prevail over the entire step. At the end of the time–step, the new particle velocity is calculated using the analytical solution:

( )exp 1 expo

p f p f all

t t v v v v F

δ δ τ

τ τ

⎛ ⎞⎛ ⎞ ⎛ ⎞= + − − + − −⎜ ⎟⎜ ⎟ ⎜ ⎟

⎝ ⎠ ⎝ ⎠⎝ ⎠

(9)

where, superscripts f and p refer to the fluid and the particles respectively, and τ is the particles’ life time (in our study something larger than the necessary travelling time from the source to the receptor). The forces, F (N), acting on the particle which affect the particle acceleration, are due to the difference in velocity between the particle and fluid, as well as to the displacement of the fluid by the particle. The equation of motion for such a particle is

p

p D B R VM p

dvm F F F F F

dt = + + + + (10)

where mp (kg) is the particle’s mass, F D (N) is the drag force, F B (N)

is the buoyancy force, F R (N) is due to domain rotation, F VM (N) is the virtual mass force and F P (N) is the force due to pressure.

Atmospheric boundary layer. The most important part of the numerical model is the approximation of the atmospheric boundary layer (ABL). The characteristics of the boundary layer define the boundary conditions used by the model. The boundaries are the inlet, outlet, upper boundary and the required variables are velocity, turbulent kinetic energy and turbulent dissipation rate.

The following paragraphs describe the calculation of the boundary conditions according to the most common empirical formulae.

The numerical criterion that distinguishes the states of the ABL is the sensible heat, H (Van Ulden and Holtslag, 1985) which is derived from the simplified form of the energy balance for ABL:

n H G Rλ + Ε + = (11)

where, λ is the latent heat of evaporation, E is the evaporation rate, G is the energy flux from ground, and Rn is the net radiation. An empirical approximation follows:

0,unstable0.9

0,neutral110,stable

<⎧⎪

= =⎨+ ⎪ >⎩

n

o

R

H B

, where p a

o

w v

pC M

B M L e

ΔΤ

= Δ (12)

Βο is the Bowen ratio (Bowen, 1926; Kakane and Agyei, 2006), p (Pa) is the pressure, C p (J kg

–1 K

–1), is the specific heat of air, Lv (J),

is the evaporation enthalpy of water, Mw and Μa, the molecular weights of water and air respectively, and ΔT (K) and Δe (Pa) is the temperature and pressure difference at two different heights. The calculation of the net radiation depends on whether there are available measurements of solar radiation or not. In this case (AERMIC, 2006) it can be calculated by the theoretical solar radiation and the total cloud cover, n, as:

6 4

1 2

3

(1 )

1

ref SB ref

n

a R c T T c n

R c

σ − + − +

= + (13)

Here, c1 = 5.31x10 –13

Wm –2

K –6

, c2 = 60 W m –2

, c3=0.12, σ SB = 5.67x10

–8 W m

–2 K

–4, and a is the ground reflection

coefficient. The calculation of the velocity vertical distribution is based on the following equation:

( ) ( )* ln0.4

z m m o

o

u zu z z

z

⎛ ⎞⎛ ⎞= − Ψ + Ψ⎜ ⎟⎜ ⎟⎜ ⎟

⎝ ⎠⎝ ⎠

(14)

( ) ( )*

0.4

ln

=⎛ ⎞

− Ψ + Ψ⎜ ⎟

⎝ ⎠

ref

ref

m ref m o

o

uu

z z z

z

(15)

where, u* (m s –1

) is the friction velocity, zo (m) is the surface roughness, while the function Ψm refers to the atmospheric stability (zero for neutral conditions). The above mentioned characteristics differ along with the atmospheric state. For the calculations the Monin–Obukhov Length, L (m) , (Venkatram, 1980) is necessary. For an unstable atmosphere:

( ) ( ) ( )

( )( )

( ) ( )

2

1

31 4 *

1 12 ln ln 2 tan 2 ,

2 2

1 16 ,0.4

−⎛ ⎞+ +⎛ ⎞⎜ ⎟Ψ = + − +⎜ ⎟ ⎜ ⎟⎝ ⎠ ⎝ ⎠

= − = −

m

a p ref

b z b z z b z

C T ub z z L L

gH

π

ρ

(16)

For a stable atmosphere:

7/18/2019 APR-11-004



( ) ( ) ( )2

*0.29 2

*

*

17 1 , , 0.09 1 0.50.4

−Ψ = − − = − = −ref z L

m

T u z e L T n

gT

(17)

where T ref (K) is the temperature and g (m s –2

) is acceleration of gravity. The height of the ABL depends mainly on the atmospheric stability class and the wind velocity. It can be calculated, as everything else, empirically (Steeneveld et al., 2007). In our case

the results of the GDAS database where used.

Another important characteristic of the ABL is the vertical variation of the wind direction. This variation is introduced because of the Coriolis force and the round shape of the earth. The criterion to check whether the influence of the Coriolis force is significant or not is the Rossby number, Ro, defined as:

5 -1, 2 sin 7.29210 rad s

D

u Ro f

L f ϕ −∞= = Ω Ω = (18)

where u (m s –1

) and LD (m) are the characteristic velocity and length scale of the phenomenon and φ (rad) is the latitude.

If the number is greater than one, then the influence is not important (Luketa–Hanlin, 2006; Prospathopoulos and Voutsinas, 2006). In our case the Rosby number is around 100, thus the Coriolis force is neglected. However, there is still the vertical variation, Dz (m), of the wind direction by the distance from the ground, z (m), owing to the round shape of earth. This variation is noted up to the mixing height and is calculated according to:

1.58 1 z h

z h D D e−⎡ ⎤= −⎣ ⎦ (19)

where, Dh is the wind direction at specific height h. Indicative values of Dh can be found in literature.

Finally, the turbulent kinetic energy and the turbulent

dissipation rate for the inlet and the upper boundary are derived by the following set of equations (Prospathopoulos and Voutsinas, 2006; Blocken et al., 2007):

2 3

* *, ( )0.40.09

= =u u

k z z

ε (20)

In addition at the upper boundary a constant velocity field (or shear stress) is applied, as it is recommended when dealing with simulations in an ABL.

2.4. Emission rate

The source under consideration is an overburden dump near

the mines of Amyndaion, Greece operated mainly by the Greek Public Power Corporation. For the present study literature resources (U.S. EPA, 1995; Chaulya et al., 2003) were used to estimate the various emission rates of processes concerning the overburden dumps. The overall calculated emission rates showed that the main source of particulate matter is the bare land of the overburden dumps, because of the vast exposed area (each virtual source bounds more than 400 000 m

2, Figure 1) and the wind drift.

That is, the emission rate from one of the virtual sources – overburden dumps – is at least two orders of magnitude larger than any other source (e.g. conveying and unloading). For this reason, the emissions from the exposed overburden dumps were employed solely and all other sources were considered to be of no significant strength.

The particles emission rate from an overburden dump depends on the wind velocity (friction velocity) and the size of the particles. The emission rate, E (g m

–2 s

–1), can be calculated by the

following equation (Chaulya et al., 2003):

0.2 0.1100

100 2.6 120 0.2 276.5

m s u a E

m s u a

−⎛ ⎞ ⎛ ⎞= ⎜ ⎟ ⎜ ⎟

− + +⎝ ⎠ ⎝ ⎠ (21)

where, m (%) is the soil humidity, s (%) is the silt content, u (m s –1

) is the wind velocity and a (km

2) is area of the source. The average

emission rate for total solid particles in our tests was equal to

0.0002 g m

–2

s

–1

. The size distribution of the particles was the one recommended for coal mining overburden soil by EPA (U.S. EPA, 1995) that is particles smaller than 30 μm diameter 48%, smaller than 15 μm 28%, smaller than 10 μm 23% and smaller than 2.5 μm

3%.

2.5. Air quality model

Local field measurements of PM are missing, so for the evaluation of the results of the numerical model a well–known air quality model was implemented. The AMS/EPA Regulatory Model (AERMOD) described here is applicable to rural and urban areas, flat and complex terrain, surface and elevated releases, and multiple sources (including, point, area and volume sources). The AERMOD modelling system consists of two pre–processors and the

dispersion model. The meteorological pre–processor (AERMET) provides AERMOD with the meteorological information that needs to characterize the ABL. The terrain pre–processor (AERMAP) both characterizes the terrain, and generates receptor grids for the dispersion model (AERMOD).

General description. AERMOD is a steady–state plume model. In the stable boundary layer (SBL), it assumes the concentration distribution to be Gaussian in both the vertical and horizontal direction. In the convective boundary layer (CBL), the horizontal distribution is also assumed to be Gaussian, but the vertical distribution is described with a bi–Gaussian probability density function (pdf). This behaviour of the concentration distributions in the CBL was demonstrated in literature (Willis and Deardorff, 1981; Briggs, 1993). Additionally, in the CBL, AERMOD treats “plume

lofting,” whereby a portion of plume mass, released from a buoyant source, rises to and remains near the top of the boundary layer before becoming mixed into the CBL. AERMOD also tracks any plume mass that penetrates into the elevated stable layer, and then allows it to re–enter the boundary layer when and if appropriate. For sources in both the CBL and the SBL, AERMOD treats the enhancement of lateral dispersion resulting from plume meander. Using a relatively straight forward approach, AERMOD incorporates current concepts about flow and dispersion in complex terrain. Where appropriate the plume is modelled as either impacting and/or following the terrain. This approach has been designed to be physically realistic and simple to implement while avoiding the need to distinguish among simple, intermediate and complex terrain, as required by other regulatory models. As a result, AERMOD removes the need for defining complex terrain

regimes. All terrain is handled in a consistent and continuous manner while considering the dividing streamline concept in stably–stratified conditions (Snyder et al., 1985).

Obligatory input data. AERMOD is designed to run with a minimum of observed meteorological parameters. As a replacement for the ISC3 (Industrial Source Complex) model, AERMOD can operate using data of a type that is readily available from weather service stations. AERMOD requires only a single surface measurement of wind speed, wind direction and ambient temperature. It also needs observed cloud cover. However, if cloud cover is not available (e.g. from an on–site monitoring program) two vertical measurements of temperature (typically at 2 and 10 meters), and a measurement of solar radiation can be substituted. A full morning upper air sounding (RAWINSONDE) is required in

order to calculate the convective mixing height throughout the day. Surface characteristics (surface roughness, Bowen ratio, and albedo) are also needed in order to construct similarity profiles of the relevant ABL parameters. AERMOD has become EPA’s

7/18/2019 APR-11-004



preferred regulatory model for both simple and complex terrain. A description of the fully revised model is presented in literature (Cimorelli et al., 2005). Performance of the final version of AERMOD is documented in Perry et al. (2005).

3. Results and Discussion

This section presents the results of the numerical model (Ansys CFX) for the selected meteorological scenarios. Moreover, the results of the numerical model are compared with the ones from the air–quality model (AERMOD). However, the air–quality model used the detailed hourly–based meteorological data for the full length of the study (from 2003 to 2007).

3.1. Selection of scenarios

An extensive statistical analysis on the meteorological data of the target area for the years from 2003 to 2007 was undertaken to form a specific number of scenarios (Table 1) concerning the wind direction and speed (see Section 2.2). These scenarios consist of three wind directions (North, North–Northwest and Northwest) and three wind speeds. Moreover, these scenarios appear to be

the most frequent ones– statistically speaking– (∼40%) and at the same time are the ones that transfer the pollution from the dump site to the nearby residential area.

3.2. Simulations of scenarios

For each scenario, the wind’s velocity field is calculated in steady state; a value of 10

–4 of the residuals was required for a

successful convergence for all the variables. Figure 4 illustrates the resulting velocity field in two of the scenarios. Consequently the particle tracking algorithm (presented earlier) is employed for each source. For each scenario the average ground concentration of PM

was calculated for the city of Amyndaion. The results are summarized on Table 1 and representative contour plots are illustrated in Figure 5, where one can observe the PM dispersion

from each one of the virtual sources for the same wind direction (North–Northwest) and speed (2.5 m s

–1). In addition, from this

figure it is clear that the concentration levels are increasing as the dump site is expanding Southern and closer to the city. This increase is also demonstrated in Figure 6, where the average concentration for each source at each scenario is compared (as a ratio) with the average concentration for source 0 (current state – see Figure 1) at the same scenario. For example for a North–Northwest direction the concentration levels increase up to two orders of magnitude when the dump site is located at the virtual source 3.

This high increase at the concentration levels could not be appointed to the fact that the Southern sources are closer to the city. A thorough study of the topography of the sources reveals

that the two Northern virtual sources (source0 and source1) are located in a small basin and the surrounding hills prohibit a large portion of the dust cloud to escape the dump site. This is clearer in Figure 4b and Figure 5a, where the continuity of the dust cloud is disrupted by the surrounding hills. On the other hand, the two Southern sources are located outside this basin and the dust cloud can travel directly to the city.

3.3. Results evaluation

As it was described earlier, to evaluate the results of the numerical model the well known integral model AERMOD (U.S. EPA) was employed. The detailed hourly meteorological data for the years from 2003 to 2007 were used and the average concen‐tration at the city of Amyndaion was calculated. The calculated annual average concentration levels for each source are illustrated in Figure 7. A quantitative analysis of the contour plots presents similar quantitative results as in the case where the numerical model was used. In Figure 7, one can also observe the PM

atmospheric dispersion at all wind directions. The highest impact of the dump sites appears to be North–eastern of them (frequent South–western winds) and Southern (North–North western winds).

Figure 8 shows a comparison of the calculated average concentration values between the numerical model and AERMOD. The predicted concentration levels are of the same order of magnitude and the agreement between the two models could be considered fair. Especially, the two predictions come closer if one compares the results of AERMOD with the addition of the numerical model’s results for the Northwest and North–Northwest wind direction. This last modification is probably necessary because the influence of these two wind directions in the AERMOD simulations is not discrete. The overall differences of the two models could be attributed to a series of causes, a list of the most important ones follows:

i) AERMOD takes in consideration the elevated terrain, but not

its influence on the wind field therefore the impact of the surrounding hills appears to be lower than in the results of the numerical model,

ii) the characteristics/parameters of the atmospheric boundary layer are explicitly defined in the numerical model while in the case of AERMOD they are internally calculated,

iii) the numerical model is applied only on the most frequent meteorological scenarios while AERMOD used the detailed meteorological data,

iv) the Langrangian numerical model calculates the PM

atmospheric dispersion based on a finite number of particles (around a few millions) which are significantly lower than the real case.

Figure 4. Velocity field for N‐1.5 (a) and NW ‐1.5 (b) at height 10 m above ground.

7/18/2019 APR-11-004



Figure

5. Concentration contours for scenario NNW ‐2.5 for each one of the sources at the ground level ( a‐Source 0, b‐Source 1, c ‐Source 2 and d ‐Source 3).

Figure

6.

Ratio of PM average concentration for each source to PM average concentration for source 0 (current state – see Figure 1) at the city of Amyndaion.

7/18/2019 APR-11-004



Figure

7. Average concentration ( μg m‐3 ) at the area of the city of Amyndaion as it was calculated for three different meteorological

scenarios (wind direction) by CFX and for the whole year (2004) by AERMOD ( a‐Source 0, b‐Source 1, c ‐Source 2 and d ‐Source 3).

Figure

8. Average daily concentration ( μg m‐3

) of PM10 for each source at ground level for the year 2004, by AERMOD.

7/18/2019 APR-11-004



4. Conclusions

In conclusion, a numerical model and an integral model were employed to simulate the present and future state of an overburden dump of lignite mine near the city of Amyndaion. Previous studies have proven that air suspended PM impacts the

health of inhabitants, and thus such a study was considered of great importance. A series of assumptions, mainly related to the definition of the boundary conditions of the numerical model, were undertaken in order to simulate the atmospheric dispersion of dust in the atmospheric boundary layer.

The results with both models showed that the yearly average concentration induced solely by the overburden dump (current and future state) is approximately 2 μg m

‐3. This concentration is

quite lower than the European limit for PM10 which is 40 μg m‐3

(EC, 2008). However, it should be remembered that the present work studies the impact from a single overburden site and takes into consideration only its emissions without the addition of background concentration and the emissions from the neighbouring sites. If all of them are summed then the total

emission rate increases drastically and so does the induced PM concentration. Furthermore, a southern expansion of the over‐burden dumps showed that will increase the concentration levels at the city of Amyndaion. For this reason, a more detailed study should be conducted by the local authorities in order to indentify the most appropriate areas to expand the dump sites and to take into advantage the local topography.

The overall application of the complex numerical model in an environmental study was considered successful. The comparison of the two models showed that both predictions agree fairly well, though the numerical model tends to underestimate the concen‐tration levels. This could be appointed mainly on the use of a few

meteorological scenarios and the detailed data and the finite number of particles used in the Lagrangian model. However,

caution must be shown at the description of the ABL into the numerical model, because it requires experience on dealing both with numerical models and atmospheric phenomena. A successful application of a CFD model in one case cannot guarantee a similar success in all cases. Therefore, to our opinion, both integral and numerical models should be applied in order to get the full picture.

Acknowledgements

The research study was financed by the Municipality of Amyndaion, Greece (Research Committee of Aristotle University, Project No 83314) and part of it was presented at the 6

th

International Congress on Computational Mechanics, Thessaloniki Greece 2008.

References

AERMIC, 2006. AERMOD (Version 07026): EPA‐US, http://www.epa.gov/

ttn/scram/dispersion_prefrec.htm#aermod, Accessed 12/2010.

Anderson, J.D., 1995. Computational Fluid Dynamics: The Basics with

Applications. New York: McGraw‐Hill.

ANSYS, 2006. ANSYS CFX version 5.7 Manual. ANSYS Europe Ltd.

Blocken, B., Stathopoulos, T., Carmeliet, J., 2007. CFD simulation of the

atmospheric boundary layer: wall function problems. Atmospheric

Environment 41, 238‐252.

Bowen, I.S., 1926. The ratio of heat losses by conduction and by

evaporation from any water surface. Physical Review 27 , 779‐787.

Briggs, G.A., 1993. Plume dispersion in the convective boundary layer. Part

II: analyses of CONDORS field experiment data. Journal of Applied Meteorology 32, 1388‐1425.

Chaulya, S.K., Ahmad, M., Singh, R.S., Bandopadhyay, L.K., Bondyopadhay,

C., Mondal, G.C., 2003. Validation of two air quality models for Indian

mining conditions. Environmental Monitoring and Assessment 82, 23‐

43.

Cimorelli, A.J., Perry, S.G., Venkatram, A., Weil, J.C., Paine, R.J., Wilson, R.B.,

Lee, R.F., Peters, W.D., Brode, R.W., 2005. AERMOD: a dispersion

model for industrial source applications. Part I: general model

formulation and boundary layer characterization. Journal of Applied

Meteorology 44, 682‐693.

Cotton, W.R., 1997. Explicit Simulation and Parameterization of Mesoscale

Convective Systems. U.S. Department of Energy, Atmospheric

Radiation Measurement Program, Report DOE/ER/61749‐3.

EC, 2008. Council Directive 2008/50/EC on ambient air quality and cleaner

air for Europe. Official Journal of the European Communities L(152), 1‐

44.

GDAS, 2008. The Real‐time Environmental Applications and Display System:

Air Resources Laboratory, National Oceanic and Atmospheric

Administration's.

Hargreaves, D.M., Wright, N.G., 2007. On the use of the k‐ε{lunate} model

in commercial CFD software to model the neutral atmospheric

boundary layer. Journal of Wind Engineering and Industrial

Aerodynamic 95, 355‐369.

Kakane, V.C.K., Agyei, E.K., 2006. Determination of surface fluxes using a

bowen ratio system. West Africa Journal of Applied Ecology 9 , 1‐9.

Lettau, H.H., 1970. Physical and meteorological basis for mathematical

models of urban diffusion processes. Proceedings of Symposium of

Multiple‐Source Urban Diffusion Models, Chapel Hill, North Carolina,

Eds: Stern, A.C., USEPA AP‐86, pp. 2.1–2.26.

Loomans, M., Lemaire, T., 2002. Particle concentration calculations using

CFD—a comparison. Proceedings of Indoor Air , Monterey, California,

USA.

Luketa‐Hanlin, A., 2006. A review of large‐scale LNG spills: experiments and

modelling. Journal of Hazardous Materials 132, 119‐140.

Markakis, K., Poupkou, A., Melas, D., Zerefos, C., 2010. A GIS based

anthropogenic PM10 emission inventory for Greece. Atmospheric

Pollution Research 1, 71‐81.

Meroney, R.N., 2004. Wind tunnel and numerical simulation of pollution

dispersion: a hybrid approach, Invited Lecture: Croucher Advanced

Study Institute, Hong University of Science and Technology.

Moussafir, J., Olry, C., Perdriel, S., Castanier, P., Tinarelli, G., 2010.

Applications of the MSS (MICRO‐SWIFT‐SPRAY) model to long‐term

regulatory simulations of the impact of industrial plants. Proceedings of

Harmo 13, 1‐4 June Paris, paper No H13‐257.

Pasquill, F., 1971. Atmospheric dispersion of pollution. Quarterly Journal of

the Royal Meteorological Society 97, 369‐395.

Perry, S.G., Cimorelli, A.J., Paine, R.J., Brode, R.W., Weil, J.C., Venkatram, A.,

Wilson, R.B., Lee, R.F., Peters, W.D., 2005. AERMOD: a Dispersion

model for industrial source applications. Part II: model performance

against 17 field study databases. Journal of Applied Meteorology 44,

694‐708.

Prospathopoulos, J., Voutsinas, S.G., 2006. Implementation issues in 3D

wind flow predictions over complex terrain. Journal of Solar Energy

Engineering 128, 539‐553.

Riddle, A., Carruthers, D., Sharpe, A., McHugh, C., Stocker, J., 2004.

Comparisons between FLUENT and ADMS for atmospheric dispersion

modelling. Atmospheric Environment 38, 1029‐1038.

Sichletidis, L., Tsiotsios, I., Gavriilidis, A., Chloros, D., Gioulekas, D.,

Kottakis, I., Pataka, A., 2005. The effects of environmental pollution on

the respiratory system of children in Western Macedonia, Greece.

Journal of Investigational Allergology and Clinical Immunology 15, 117‐

123.

Snyder, W.H., Thompson, R.S., Eskridge, R.E., Lawson, R.E., Castro, I.P., Lee,

J.T., Hunt, J.C.R., Ogawa, Y., 1985. The structure of strongly stratified flow over hills ‐ dividing‐ streamline concept. Journal of Fluid

Mechanics 152, 249‐288.

7/18/2019 APR-11-004



Steeneveld, G.J., van de Wiel, B.J.H., Holtslag, A.A.M., 2007. Comments on

deriving the equilibrium height of the stable boundary layer. Quarterly

Journal of the Royal Meteorological Society 133, 261‐264.

Triantafyllou, A.G., Kassomenos, P.A., 2002. Aspects of atmospheric flow

and dispersion of air pollutants in a mountainous basin. Science of the

Total Environment 297, 85‐103.

Triantafyllou, A.G., Paliatsos, A.G., Voutsinas, M., Zuburtikudis, I., 2002. Air

quality assessment in an industrial area by using environmental

indicators. Fresenius Environmental Bulletin 11, 933‐939.

Triantafyllou, A.G., Patra, A., Zuburtikoudis, I., Pavlidis, A., Vosniakos, F.,

Zoumakis, N., 2001. An analysis of atmospheric pollutants

concentration in urban area in Greece. Fresenius Environmental

Bulletin 10, 615‐619.

U.S. EPA, 1995. Compilation of air pollutant emission factors, Volume I:

Stationary point and area sources. AP 42, Fifth edition, Research

Triangle Park, New York.

Van Ulden, A.P., Holtslag, A.A.M., 1985. Estimation of atmospheric

boundary layer parameters for diffusion applications. Journal of

Climate and Applied Meteorology 24, 1196‐1207.

Venkatram, A., 1980. Estimating the Monin‐Obukhov length in the stable

boundary layer for dispersion calculations. Boundary ‐Layer

Meteorology 19, 481‐485.

Willis, G.E., Deardorff, J.W., 1981. A laboratory study of dispersion from a

source in the middle of the convectively mixed layer. Atmospheric

Environment ‐ Part A General Topics 15, 109‐117.

APR-11-004

Documents

Transcript of APR-11-004