QUARRYING ACTIVITIES AND DUST
EMISSIONS: A GEOSTATISTICAL METHOD
IN RISK ANALYSIS
Degan G. A., Lippiello
D., Pinzari M.
Dipartimento di Ingegneria Meccanica e Industriale,Università degli studi Roma TRE, Roma, Italy
Abctract: Quarrying activities, as for their characteristics, imply a deep environmental impact
and dust emissions represent one of the most important pollutant factors. The aim of the present study, is to identify, according to a geostatistical approach, a simple procedure to forecast spatial variability of dust emissions in order to evaluate occupational risks connected with airborne dust exposure.
1. Introduction
Dust emissions represent one of the main environmental factors in quarrying activities. The present study, developed in an Italian quarry, identifies a method to evaluate risks connected with airborne dust exposure (PM 10). The first step is based on a procedure for the identification of dust sources in the work area. Then each dust source is characterized using a sampling station in order to define the PM 10 concentration (mg/Nm3) and the
most important environmental parameters such as pressure, temperature, atmospheric humidity, wind direction and its speed are recorded.
This monitoring system has allowed the creation of an irregular data grid that represents the first point in order to analyze the spatial variability of the phenomenon. Then a variographic analysis of C variable (mg/Nm3) is realized followed by a suitable model of
the experimental variogram. Moreover ordinary Kriging is used for creating a continuous map of the variable itself and finally a cross validation procedure is realized in order to test the obtained results.
In the last part of the study a map of the action space of the hazard itself is defined and then it is used according to the FAST technique (Functional space analysis technique) to
evaluate occupational risks connected with airborne dust exposure These results have been compared with values obtained using a personal sampler to test the exposed procedure.
1.1. Dust sources and their classification
As for the exposed aim, the first step is represented by dust source individuation and classification. In particular a procedure has been developed in order to locate and classify sources themselves. The first step is represented by the analysis of the quarry plan characteristics with particular attention to:
individuation and classification of functional areas in the working zone, individuation of critical activities (as for dust emission) in each functional area, individuation of every dust source in each functional area and its classification taking
care of some features as explained below.
In particular dust sources can be classified as ordinary or extraordinary ones, depending on the nature of activity that implies dust emission. They can be located in a specific position or distributed in a more or less large area. Moreover dust emission could lasts for the entire eight hour work day or just for a short period. This classification is important to find, according to a not specific observation site, all the process activities connected with airborne dust emission, independently from its intensity that will be sampled afterwards. Finally it is possible to obtain the following classification:
localized and continuous sources, localized and discontinuous sources, distributed and continuous sources, distributed and discontinuous sources.
The site has been divided into three functional areas. In those areas we are able to locate dust sources according to the process described above and in particular it is possible to focus our analysis on: digging area, sifting and crushing area, stocking area. All the activities taking place into each area have been listed as showed in the following scheme:
Then in each critical activity dust sources have been localized and characterized as showed:
Functional area Critical activities Source Classification
Sifting and crushing area
Crushing Crushers Localized continuous
Sifting Riddles Distributed continuous
Transportation Conveyor belts Distributed discontinuous
Functional Area Critical activities
Digging area Drilling, blasting, truck loading
Sifting and crushing area Crushing, sifting, gravel transport
Stocking Heaps Localized discontinuous
1.2. Source characterization
The sources listed below have been characterized with a sampling station that surveyed the values showed in the following tables (only some samples are showed ).
Fig.1-3. Some dust sources in a limestone quarry Table 1 – 3. Some dust samples
Source Crusher
1 Crusher 2 Crusher 3 Sieve 1 Sieve 2
C(mg/Nm3) 2,24 2,61 2,87 1,75 1,85
Wind (m/s) 0.7 1.4 1.2 0.7 1.0
Temp. (°C) 16 15.5 14.5 16 16
Source Heap 1 Heap 2 Heap 3 Heap 4 Heap 5
C(mg/Nm3) 2,9 3,15 3,0 2,72 2,38
Wind (m/s) 1.7 1.1 1.4 0.6 1.3
Temp (°C) 15 14.5 16 15 15.5
Source Conveyor belt 1 Conveyor belt 2
Sample number 1 (7mt) 2 (14mt) 1 ( 5 mt) 2 (10 mt) 3 (15 mt)
C(mg/Nm3) 2,4 1,9 2,6 2,3 1,7
Wind (m/s) 1.8 1.4 1.2 1.0 0.9
Temperature(°C) 16 14 14 15 16
2. Geostatistical application
Geostatistics might be viewed as a methodology for interpolating data on an irregular pattern. Each data value is associated with a location in space and there is at least an
implied connection between the location and the data value. It may be useful to introduce some geostatistical definitions for an exact interpretation of topics.
2.1. Variogram
It is a geostatistical function used in characterizing the spatial variability of a local phonomenon. Let S be a domain of definition of FA z(x) and given x0 and x0+h a couple
of S points h distant.
z
x0 h
z
x0
is defined as a new variable called accretion,its half variance is for definition the variogram:
x0,h
12Var
z
x0 h
z
x0
(1)
The variogram estimation is realized according to experimental samples: it is shown the sampling points map on an irregular grid with different concentration values.
As for stationary and semi-stationary functions, the accretion z(x+h)- z(x) does not change, so the value to be calculated will be:
NR i i r i z x x z h 1 2
(2)The experimental variogram will be calculated on arithmetic mean of the square increase with h distance. R represents the grid side and xi+r and xi the position of two points whose
distance is r in X direction. N represents the number of couples whose distance is r and the direction is X.
Fig. 5. Spatial representation Fig.6. Variogram representation
2.2. Model of regionalization
In order to create a variographic model, an experimental variogram of the C variable is realized. After we have checked an isotropy we have chosen for a one direction model On such experimental evidence it is developed a model of two nested structures like:
Nugget component, Sill 0,094
In order to fit the theoretical variogram with two nested structures it has been considered only one direction (0°) because of phenomenon’s isortopy.
Fig.7. Variogram fitting
2.3. Linear approach in geostatistical analysis: Ordinary Kriging
It is Z(x) an aleatory stationary function used to study the phenomenon in a probabilistic way, and given C(h) and γ(h) respectively as covariance and variogram function. N is the number of points required to estimate x0, xα and z(xα) are with α=1,n respectively their
position and their corresponding variable values. Z(x0) is the unknown value in x0, z*( x0)
is the linear estimator considered. It is showed in the following form:
X
NZ
X
Z
1 0
(3)Where λα are the coefficients of a linear combination
On this evaluation is associated an estimation error called , defined as the difference between the sampled value and the estimated one:
Z
X
0
NX
Z
1
(4)But it is also to be considered that
Z X X Z E 0 =0 0 1 1
N m So
1
1
N
(5)An important parameter to evaluate estimation quality is that of considering estimation variance defined as:
22
x
x
0x
x
.
s (6)In order to obtain the best estimation, it will be necessary to minimize the estimation variance so: 0 2
1
,
n
with the condition1
1
N
0 1 2 2 0 2
s x x x xn
,
1
And calculating the derivatives
2
2
2
0
0 2x
x
x
x
s
1
,
n
With the exactness condition the system becomes:
1
1
,
n
(7)The associate estimator is called Kriging and the system is called Kriging system. The coefficient’s matrix does not depend on entity to be estimated but they are exclusively dependent on the sampled points position and on variografic function. The known vector describes the connection between the estimating entity and sampled points.
2.4. Cross validation
The model has been tested using a cross validation method called “leave one out”. This methodology estimates the variable’s value in the known points and provides to delete from time to time the measured value.
The error’s report in shown below. In particular let zi(x) be the sampled value in the x
point and z*(x) the estimate value in the same point. Errors are defined as:
Mean error
z
x
z
x
n
i i *1
(8) Mean square error:
2 1 *1
n i i ix
z
x
z
n
(9) Mean standard error:
2 1 */
1
n cv i i ix
x
z
x
z
n
(10)Mean estimation variance:
n cv ix
n
11
(11)The geostatistical analysis is then completed according with the evaluation of scatterplot estimated/sampled values.
3. Risk analysis: the discrete approach in dust evaluation
In this method the production process is made discrete both as regards the work carried out (division in sub-activities), as well as the places where such work is carried out. To do this, FAST uses the PERT technique through which the productive process is subdivided into elementary activities connected to a network. Such activities can be chosen and identified using the most suitable time for evaluation, the precision being a lot better as each activity is short. The extreme case coincides obviously with the continuous approach even where the activity is infinite in its length.
Fig.9. Error report Fig. 10. Scatterplot
For every elementary activity, FAST foresees the definition of a Functional Space as well as the minimum space necessary for elementary activity to be done. During such an activity the worker is considered to be present in every point of the Functional Space for the entire duration of the activity.
The use of the PERT technique allows, above all, for availability of the organisational aspects particularities of the method, the levelling of resources, the use of postponement, etc.
For each type of hazard pointed out, the area of existence of the hazard (accident or professional illness) and its relative period of action, are related.
In this case, the area of existence of the hazard ( dust concentration) is characterised by a certain value that is represented in the exstimated map showed above.
3.1. Exposure evaluation
The exposure evaluation is particularly simple. It requires graphic control to see whether there is intersection between the functional space and the area of existence of the hazard. Once such intersection has been ascertained, the contemporaneity is checked; in other words, we say there is contemporaneity if there is a time span in common between the action time of the danger, and the carrying out of the activity.
3.2. Risk evaluation
In each functional space two values of the dust concentration C are considered: a maximum and a minimum one. During such an activity the worker is considered to be present in every point of the functional space for the entire duration of the activity. The differece between the minimum and the maximum level is function of the dimension of the functional space. Known these two values is possible to calculate the daily exposure using the following :
0
T
t
C
E
i i
(12)Where Ci (mg/Nm3) is dust concentration (minimum or maximum), Δt is the time of
presence in the functional space, T0 is the working day duration (480 min)
Minimum daily exposure: Emin= 2,329 mg/ Nm3
Maximum daily exposure Emax= 2,588 mg/ Nm3
Those values are compared with the value obtained with a personal sampler used by the worker. The comparison is positive, the dose measured from the personal sampling is equal to 2,56 mg/.Nm3.
References
1. Degan A.G., Lippiello D., Pinzari M.: A geostatistic approach to the Functional Analysis Space Technique: a case of study. ESREL, Gdansk, 2005, 27 Giugno, 2005. 2. Lega D. et al.: Exposure to airborne dusts in the quarrying industry in Lazio.
Prevenzione oggi. Quarterly review of studies and research on safety, ISPESL. April /June, 2005.
3. Degan A.G., Lippiello D., Pinzari M.: Particolato aerodisperso in una cava di granulati del Centro Italia. Atti del Convegno Nazionale A.N.E.L.P.A. - Aggregati per le costruzioni - 28, Bologna, Italia, Marzo 2003.
4. Degan A.G., Lippiello D., Pinzari M.: Dust propagation: a method in risk analysis. Atti del 3rd Safety and Reliability International Conference - KONBiN 2003, Gdynia 26-30 May 2003.
Functional
space Dimension Duration Cmin(mg Nm
3) Cmax (mg Nm3) SF 1 9 110 2,58 2,82 SF 2 6 60 1,97 2,12 SF 3 16 130 2,79 3,10 SF 4 16 120 2,34 2,6 SF 5 9 60 1,36 1,5
