Professional Documents
Culture Documents
Journal
Cansu Dagsuyu
To cite this article: Cansu Dagsuyu (2020) Process capability and risk assessment for air quality:
an integrated approach, Human and Ecological Risk Assessment: An International Journal, 26:2,
394-405, DOI: 10.1080/10807039.2018.1511368
Introduction
Air pollution is caused by heating, transportation, and industrial activities. Human
health and the ecosystem are damaged by pollution on different rate. Differences
generally induced by many criteria such as transportation, industrialization rate, type of
fuel, quantity and quality used in industry, building construction and heat insulation,
chimney cleaning, population density of the region. These differences lead to increase in
the rate of carbon monoxide, ozone gas, particulate matter etc. in the air. Besides, air
pollution rates are determined by many criteria. Ozone, carbon monoxide, sulfur
dioxide, particulate matter (PM), nitrogen dioxide, and lead are referred as pollutant
criteria which have been set by U.S. Environmental Protection Agency (USEPA) as U.S.
national air quality standards (Kaya and Kahraman 2011). Additively, it has been found
that PM has a greater impaction human health outcome than other pollutants such as
sulfur dioxide, nitrous oxides and ozone (Dockery et al., 1993).
PM is one of the significant part of air pollution which is formed by small pieces and
liquid droplets including acids, organic chemicals, metals, and soil or dust particles
(Anderson et al. 2012). These contents lead PM to have a huge impaction on human
health. Particles which are less than 10 lm in diameter pose cause serious health prob-
lems since they can settle into lungs and bloodstream of humans (https://www.epa.gov/
pm-pollution/particulate-matter-pm-basics#effects).
In literature, there are many studies which examines the affects of PM to environ-
ment and human health. Grazia et al. (2001) have analyzed PM2.5 and PM10 samplings
in central urban area of Milan. It has been stated that samplings of PM2.5 and PM10
lead PM2.5 is a great part of PM10 with a good correlation. Chang and Lee (2007)
collected data by observing air quality in Taipei between January 2001 and May 2001.
The obtained data were integrated with artificial neural network by factor analysis and
the yellow sand period and PM10 increase in this period were determined. Romanazzi
et al. (2014) estimated a healthy risk associated with PM10 metals in the Italian city of
Torino. In the study, Metals PM10 size distribution analysis was performed and the
Target Hazard Quotient (THQ) and the Cancer Risk (CR) were used. Yang et al. (2014)
have evaluated TSP, PM10 and heavy metals (Cr, Ni, Cu, Zn, Cd, and Pb) in the
ambience of the workshop. Their study have stated that mechanical–physical process for
e-waste recycling do exist the pollutant mission according to the results of health risk
assessment since the total hazard index has exceeded the danger threshold. Ozel and
Cakmakyapan (2015) performed air quality evaluation according to PM10 using
gamma-Poisson process. Considering the Gamma-Poisson process, estimates are made
in the Central Anatolia Region for daily, monthly, and yearly time periods. Habeebullah
et al. (2015) have aimed to focus risk assessment of PM10 in four different areas in
Suudi Arabia. Number of hospital admissions has been estimated in the study. As a
result of exposure to PM10 concentrations have caused a risk of lung cancer according
to risk assessment which has been conducted in the study. Li et al. (2016) have analyzed
the samples of PM10 and PM2.5 particles from an industrial city of China. Assessment
of atmospheric heavy metal exposure related with public health risks have been
conducted in the study. It has been inferred that; results of the risk assessment have
indicated as the elements of Cr, Cd, and Co pose long cancer to residents. Alghamdi
(2016) has investigated seasonal distribution of PM10 and its contents in Jeddah in his
study. Risk assessment for heavy metals has been conducted in the study for residents.
According to the health risk assessment of the study, approximately 11% of the
population of Jeddah might have gotten cancer. Megido et al. (2017) have investigated
12 potentially toxic elements through PM10 in suburban sites of Spain. Study has been
aimed to evaluate the human health hazards related with those toxic elements for
children and adults separately. Cheng et al. (2018) have investigated trace element
pollution of PM10 in a city of China to assess human health risks. They have concluded
that Cr element is carcinogenic to children and adults. Addition to other elements
which are evaluated to be below legislation limit values.
According to the literature search which are related with PM10, its harmful affects
are generally mentioned, but its superiority and impacts are not studied.
According to directive 2008/50/EC for PM10 is determined to be 50 lg/m3 not to be
exceeded by more than 35 times per calendar year. According to this evaluation
method, measurement stations exceeding the limit value are evaluated 35 times as dirty
as in terms of air quality. However, no comparison is made between two measured
396 C. DAGSUYU
stations, which exceed the limit value by more than 35 times. The PM10 measurement
values are categorized by the National Air Quality Index Cutting Points. However,
when examining the air quality of the stations, these cutoff points are generally not
taken into consideration except for considering the measurement value is greater than
50 lg/m3. This does not have a different effect when the PM10 measurement value is
51 or 150 lg/m3 in the evaluations. In the literature review, there is no study that took
this difference into consideration.
Limit values for PM10 are taken into account in process capability in literature. There
are many studies which take advantage of process capability ðCp Þ considering the limit
values allowed by laws and legislations in environmental assessments. Process capability
calculations are made and measurement value dispersions are examined and interpreted
in these studies. Hence, suitability of long term researches are provided for environmen-
tal risk assessments (Corbett and Pan 2002). Corbett and Pan (2002) conducted environ-
mental assessments with statistical process control techniques (CUSUM chart, IX-MR
chart) and process capability analysis. In particular, the importance of process capability
analysis in risk management has been conducted in some studies. Kaya and Kahraman
(2009) utilized traditional and fuzzy process capability indices methods to determine air
pollution in their works. The six-sigma approach has been used when creating member-
ship functions for process capability indices. Kaya and Kahraman (2011) utilized fuzzy
process capability analysis in evaluating SO2 and PM10 from their important indicators
_
of air pollution. The study applied in Istanbul-Turkey and stated that there would be
more accurate measurement of air pollution with developed method.
It is difficult to make long-term evaluations on environmental evaluations based on
the number of cross-overs of the measured values of the stations. There may also be
some confusion in making environmental assessments based solely on Cp values.
Process specification limits for PM10 are determined under standards as a constant
value. The standard deviation value can vary on station. However, standard deviation
value showing variability does not indicate whether the measured values in the process
are within the PM10 specification limits (0–50 lg/m3). For instance, if we assume that
the 5-day of PM10 measurement values (10, 12, 14, 16, and 20) and (50, 52, 54, 56, and
60) are for two different measurement stations, the Cp values will be the same because
of the standard deviation values for both stations. But the fact that some Cp values will
not give any idea about the air quality of the processes.
Integration the amount of measurement values exceeding the specification limits and
Cp value will provide more accurate evaluation of air quality of measurement stations in
terms of PM10. Two new approaches are developed in this study. In the first developed
approach, severity scale according to measurement values of stations are developed
with the help of risk assessment method. PM10 measured values are matched with
considered methods’ scale. Aim of developing the first approach is to evaluate air
quality with the consideration of severity values and probability of PM10 values. In the
second developed approach, severity and probability values are taken into consideration
with stations Cp values in order to evaluate air quality. Thus, the size of each measured
value received from the measuring stations will be reflected in the air quality. The air
quality evaluation based on the risk analysis method and the integration of Cp values
constitutes the originality of this study.
HUMAN AND ECOLOGICAL RISK ASSESSMENT 397
The rest of the article is organized as follows. The developed approach method is
described in Section 2. The numerical examples and results are presented in Section 3.
Conclusions are outlined in Section 4.
Table 1. National Air Quality Index Cut-Off Points and Fine Kinney Scale.
National Air Quality Index Cut-Off Points Fine Kinney Scale
(http://www.havaizleme.gov.tr/Default.ltr.aspx) (Kinney and Wiruth, 1976)
i PM 10 value (mg/m3) 24 hours average values Interpretation Possible consequences Value (Si )
1 0–50 Good Noticeable 1
2 51–100 Middle Important 3
3 101–260 Sensitive Serious 7
4 261–400 Unhealthy Very serious 15
5 401–520 Bad Disaster 40
6 >521 Dangerous Catastrophe 100
398 C. DAGSUYU
the determination of the environmental impact index as “good” for the PM10 value of
0–50 lg/m3 corresponds to the noticeable group according to the Fine Kinney severity
scale and the value is expressed as “1.” Likewise, the PM10 value of 521þ lg/m3 is deter-
mined as “dangerous” according to the index and corresponds to the Catastrophe group
according to the Fine Kinney severity scale and the value is determined as “100.” Thus,
in the case of PM10 values measured at two different measurement stations of 350 and
450 lg/m3, together with the “unhealthy” and “bad” evaluations, the measurement value
of these measured values is 450 lg/m3, 350 lg/m3 will be about 2.6 (40/15) times higher
than the measurement.
Environmental effect
According to the National Air Quality Index Cut-Off Points, the environmental effects
of the PM10 measurement values on a yearly basis were calculated for the measuring
stations. The environmental effect was obtained by multiplying the probability value of
observations in each category (Pi ) of PM10 and the severity value obtained from the
Fine Kinney scale ðSi Þ and is given in Equation (2).
Environmental Effecti ðEEi Þ ¼ Pi Si i ¼ 1:6 (2)
With Equation (2), environmental affects of measured PM10 values will be quantita-
tive for each class of measurements station. For PM10 measurement values which are
0–50 lg/m3 range is acceptable according to the standards and regulations. Thus, number
of measurements in this range is supposed to be as high as possible for the requirements
of high air quality. Environmental Quality Ratio (EOR) for each PM10 measurement
station is calculated by dividing each station’s environmental affect within interval of
0–50 mg/m3 ðEE1 Þ to total environmental affect EOR calculation is given in Equation (3).
EE1
Environmental Quality Ratio ðEORÞ ¼ (3)
P
6
EEi
i¼1
Air quality ratio originated from PM10 is determined with EOR. EOR values are
obtained by proportioning the value of the PM10 value in the desired range (EE1 ) to
P
environmental impact value of annual PM10 values ( 6i¼1 EEi Þ from the stations which
will provide comparisons of stations. The EOR value is 1 if all PM10 measurements
taken during the year within the range of 0–50 lg/m3 and the EOR value is 0, if no
HUMAN AND ECOLOGICAL RISK ASSESSMENT 399
observations are between 0 and 50 lg/m3. Thus, change interval for the EOR value is
0–1. As the PM10 measurement values move away from 0 to 50 lg/m3, the EOR value
will converge to zero and the unwanted EOR will be obtained. As value of EOR is
approached to 1, it can be said that the station has the desired PM10 values and the
desired air quality ratio will be obtained.
For the comparison of stations, besides to EOR; Cp value is also taken into account.
Process capability
The air quality ratio from the PM10 on station basis is analyzed by EOR. However, this
value does not provide information on the air quality capability of the considered
measurement station by itself. Process capability index is the number that summarizes
behavior of a product or process characteristic relative to specifications (Kaya
and Kahraman 2009). These indices help us to decide how well the process meets the
specification limits (Montgomery 2005).
Cp is calculated by dividing specification width to six sigma. The specification width
represents process requirements. If the process variation is very large, the Cp value
is small and it represents a low process capability (Kaya and Kahraman 2009). Cp
calculation is presented in Equation (4) (Montgomery 2005).
Spec: Width Allowable Process Spread USLLSL
Cp ¼ ¼ ¼ (4)
Process Spread Actual Process Spread 6r
where r is the standard deviation of the process. And USL is upper specification limits
and LSL is lower specification limits.
The relationship between quality conditions and Cp value are identified by Tsai and
Chen (2006) and is given in Table 2. A value of Cp greater than 1.33 is considered
acceptable for the process.
The limit values of PM10 and standard deviation of stations are taking into account
that Cp can be calculated according to Equation (4). Cp results are evaluated according
to Table 2 in this study.
EOR value can be used to compare the measurement stations among each other, but
if the evaluation of the EOR considers to value together with Cp ; the value of the
measurement stations will give more reliable and dependable results. The process
capability analysis helps decision-makers to evaluate even for a large proportion of time
effectively since it has a capability of conforming with environmental legislation
(Corbett and Pan 2002). The air capability value (ACV) is obtained by multiplying EOR
and Cp is expressed by Equation (5). The spread within the desired tolerance limits of
environmental quality ratio of stations have been examined with ACV. It is desirable
that the ACV value is as great as possible, so that the spread of the PM10 measurement
values are in the range of 0–50 lg/m3 and at higher Cp values.
Air Capability Value ðACVÞ ¼ Cp EOR (5)
Case study
According to the air evaluation report of the Ministry of Environment and
Urbanization (2017); Ankara, which is taken as polluted city, and Artvin, Bitlis,
Eskişehir, Kırıkkale, Kırşehir, and Yozgat, which are taken clean cities, are taken into
consideration in this study. Clean and dirty cities are evaluated by developed approach
in this study. The PM10 measurement is carried out in Ankara at eight stations and the
evaluations are carried out on a station basis. In city of Artvin, measurements are taken
from two stations while Bitlis, Eskişehir, Kırıkkale, Kırşehir, and Yozgat have one
measurement station. This study is conducted from those measurement stations. Data
taken into account in the study are taken from the weather monitoring network on a
day-by-day basis, which is 24-hour averages of PM10 values from 1 January 2017 to 31
December 2017 on a station basis (http://www.havaizleme.gov.tr/Default.ltr.aspx).
Summarization of PM10 measurement data of eight stations belong to Ankara city is
given in Table 3. Considering the quality conditions in Table 2, it is seen in Table 3
that the values of all the stations in Ankara are “poor.” Although the minimum–
maximum intervals of the stations are close to each other according to Table 3, the
Cp values of the stations are different because the standard deviation value indicating the
spread of the data is different.
Although the interpretation of the environmental competence between the stations
can be made with the value Cp , it is not possible to determine which station has more
critical air quality since measurement stations have the same value. The values of
Dikmen and Sincan measurement stations are approximately the same in Ankara city,
indicating that the station variability is close to each other. Low Cp values indicate that
stations are not competent. However, in order to be able to determine which station
area has a higher air quality risk value, the PM10 measurement values should be
examined. Stations of Dikmen and Sincan have different min–max value intervals even
though they have almost same Cp values since they have almost same standard
deviation. If evaluation is made according to only Cp value, these two stations will have
same competence. However, since the stations have different min-max intervals and the
number of data in each class differs according to the PM10 classification, air quality
values will be different. As in Table 3, there are 217 observations at 0–50 lg/m3, 108
observations at 51–100 lg/m3 and 38 observations at 101–260 lg/m3 in the Bahçelievler
station. Since there are days in the stations where PM10 values are not measured, the
total number of measurements is less than 365, which has 1 year observation values.
The calculation results of Equations 2, 3, and 5 are given in Table 4. According to
Table 4, PM10 quality in Bahçelievler region with 0.2689 EOR is the best in the Ankara
city compared to other stations. ACV value obtained by multiplying Cp and EOR values
has also the highest value in Bahçelievler station.
As the Cp , EOR, and ACV values increase, the PM10 air quality of the zone to which
the station belongs increases. As seen in Table 3, despite the fact that the Dikmen area
is in the 2nd place according to the Cp process capability, it is 5th and 4th rank respect-
ively according to the EOR and ACV values in Table 4. As in Table 3, although the first
six stations in the measurement stations in Ankara are in the order of Bahçelievler,
Dikmen, Sincan, Cebeci, Keçi€ oren, and Demetevler according to Cp values for PM10,
according to EOR that the severity and probability values are taken into consideration,
the stations are in the order of Bahçelievler, Demetevler, Sincan, Cebeci, Dikmen, and
Keçi€
oren in Table 4. When multi-criteria evaluation is made with Cp and EOR, the
order of the stations is Bahçelievler, Sincan, Cebeci, Dikmen, Demetevler, and Keçi€oren.
According to the results of Cp , EOR, and ACV, Bahçelievler has the best air quality and
Kayas and Sıhhiye have the worst air quality in Ankara and it seen in Table 4.
In Table 5, a confusion matrix is created for stations in Ankara based on EOR and
ACV values. According to Table 5, when air quality evaluation is done by ACV instead
of EOR, it is determined that the ranking of 4 stations did not change, the air quality of
the demetevler station was worsened, and the air quality of Sincan, Cebeci, and Dikmen
stations was better.
Ministry of Environment and Urbanization (2017) were taken from the weather
monitoring network. The data was taken day-by-day with 24-hour averages. The mean,
standard deviation and Cp values are calculated according to data which is given
in Table 6.
According to Table 2; if the measured Cp value is more than “1,” quality condition is
interpreted as “capable,” But even Turkey’s cleanest cities have the Cp value of between
0.537996–0.894737 which is illustrated in Table 6. This shows that, air quality decisions
cannot be made only interpreting Cp value.
Cp , EOR, and ACV calculations and rankings are given Table 7 to make more reliable
interpretations. Since there are no measurement above 260 lg/m3 in PM10 values taken
from clean cities, values higher than 260 lg/m3 are not included in Table 7.
Although the Cp value is not within the desired range according to Table 2, it is seen
in Table 7 that the EOR value at Artvin Hopa station is close to 1. If the EOR value is
close to 1, it shows that PM10 values of the station being measured are within the
desired limit range. When Artvin Hopa and Yozgat are evaluated in terms of air quality,
these stations have the same ranking according to Cp , EOR, and ACV values. Eskişehir
is in the 5th place according to the EOR and ACV value even though it is in the 3rd
place according to the Cp value which shows Cp does not mean anything by itself.
HUMAN AND ECOLOGICAL RISK ASSESSMENT 403
The confusion matrix based on the EOR and ACV values are obtained for clean cities
in terms of PM10 in 2017 is given in Table 8. As it is illustrated in Table 8, ACV order
did not change the order of Artvin Hopa, Yozgat, and Eskişehir measuring stations
according to EOR order. While the order of the Bitlis and Kırşehir measurement stations
deteriorates, the order of the measurement stations of Artvin and Kırıkkale is improved.
Conclusion
Since air quality is directly related with human health, high quality of air is desired in
cities. Air quality is affected by many factors such as transportation, heating and energy.
Each factor causes the formation of harmful gases in the air and pollution of the air. In
this study, the air quality is evaluated based on the PM10 measurement values on
stations basis. The number of days exceeding the PM10 limit values of the measurement
stations is generally taken into consideration in the PM10 evaluations.
In this study, in addition to the limit values defined for PM10 for the stations on the
measurement days, the amounts exceeding the limit values and their severity were taken
into consideration. Fine Kinney risk analysis method was used to determine the severity
value according to the measured values. EOR values were obtained on a station basis by
performing a probabilistic calculation of the measured values and comparisons were
made. The ACV value is obtained by taking into account together the Cp value and
the EOR value used in the air quality evaluation and stations were compared. As a
result of the calculations, the EOR and ACV values were compared with the air quality
orders of the measuring stations. It can be seen that the order of the measurement
stations changes with the consideration of the indicator Cp of the competence of the
measurement values. This situation stated that taking Cp value into consideration is
important in evaluating the quality of measurement stations.
According to Ministry of Environment and Urbanization (2017), Ankara is considered
as dirty while Artvin, Bitlis, Eskişehir, Kırıkkale, Kırşehir ve Yozgat are considered as
clean cities in terms of PM10. Air qualities of those cities’ stations are compared with the
values of EOR and ACV. Results indicate that station ranks are determined with EOR as
Bahçelievler, Demetevler, Sincan, Cebeci, Dikmen, Keçi€ oren, Sıhhıye and Kayas. While
the ranks of stations are as Bahçelievler, Sincan, Cebeci, Dikmen, Demetevler, Keçi€ oren,
Sıhhıye and Kayas according to the value of ACV which is calculated with Cp value.
Clean cities ranking with EOR value is determined as Artvin Hopa, Yozgat, Bitlis,
Artvin, Eskişehir, Kırşehir ve Kırıkkale. When those cities are evaluated with ACV, 57%
of the station classes are changed and 43% of them remained in same class.
404 C. DAGSUYU
Besides of evaluating only PM10 value, which is exceeding specification limits or not,
If severity, probability and Cp values of PM10 are taken into consideration, assessments
are going to become more sensitive for the areas of measurement station of cities.
At some measurement stations, there are days when measurements are not taken. If
the number of unmeasured days is too high, this will reduce the accuracy of the
assessments which generates a limitation of this study. In future studies, the risk
analysis method considered in the study can be changed or the fuzzy approach can be
integrated into the study.
References
Kaya I, and Kahraman C. 2009. Air pollution control using fuzzy process capability indices in
the six-sigma approach. Hum Ecol Risk Assess 15(4):689–713.
Kaya I, and Kahraman, C. 2011. A new tool for risk assessment of air pollution: fuzzy process
capability indices. Hum Ecol Risk Assess 17(3):613–30.
Corbett CJ, and Pan JN. 2002. Evaluating environmental performance using statistical process
control techniques. Eur J Oper Res 139(1):68–83.
Gerasopoulos E, Kouvarakis G, Babasakalis P, Vrekoussis M, Putaud JP, and Mihalopoulos N.
2006. Origin and variability of particulate matter (PM10) mass concentrations over the Eastern
Mediterranean. Atmos Environ 40(25):4679–90.
Directive 2008/50/EC of the European Parliament and of the Council of 21 May 2008 on ambient
air quality and cleaner air for Europe.
Anderson JO, Thundiyil JG, and Stolbach A. 2012. Clearing the air: a review of the effects of
particulate matter air pollution on human health. J Med Toxicol 8(2):166–75.
Alghamdi M. 2016. Characteristics and Risk Assessment of Heavy Metals in Airborne PM10
from a Residential Area of Northern Jeddah City, Saudi Arabia. Pol J Environ Stud
25(3):939–49. doi:10.15244/pjoes/61531
Cheng X, Huang Y, Zhang S-P, Ni S-J, and Long Z-J. 2018. Characteristics, sources, and health
risk assessment of trace elements in PM10 at an Urban Site in Chengdu, Southwest China.
Aerosol Air Qual Res 18(2):357–70. doi:10.4209/aaqr.2017.03.0112
Grazia MM, Stefano V, Gianluigi V, and Roberta V. 2001. Characterisation of PM10 and PM2.5
particulate matter in the ambient air of Milan (Italy). Atmos Environ 35:4639–50.
Li K, Liang T, and Wang L. 2016. Risk assessment of atmospheric heavy metals exposure in
Baotou, a typical industrial city in northern China. Environ Geochem Health 38(3):843–53.
doi:10.1007/s10653-015-9765-1
Megido L, Suarez-Pena B, Negral L, Castrillon L, and Fernandez-Nava Y. 2017. Suburban air
quality: human health hazard assessment of potentially toxic elements in PM10. Chemosphere
177:284–91. doi:10.1016/j.chemosphere.2017.03.009
Othman M, Latif MT, and Mohamed AF. 2015. The PM10 compositions, sources and health risks
assessment in mechanically ventilated office buildings in an urban environment. Air Qual
Atmos Health 9(6):597–612. doi:10.1007/s11869-015-0368-x
Yang Y, Fang W, Xue M, Xu Z, and Huang C. 2014. TSP, PM10 and health risk assessment for
heavy metals (Cr, Ni, Cu, Zn, Cd, Pb) in the ambience of the production line for waste cath-
ode ray tube recycling. J Mater Cycles Waste Manag 18(2)–296–302. doi:10.1007/s10163-014-
0331-1
Habeebullah TM, Mohammed AM, and Morsy EA. 2015. Risk Assessment of Particulate Matter
(PM10) in Makkah, Saudi Arabia. World Acad Sci Eng Technol Int J Environ Chem Ecol
Geolog Geophys Eng 9(8):947–54.
Romanazzi V, Casazza M, Malandrino M, Maurino V, Piano A, Schilir o T, and Gilli G. 2014.
PM10 size distribution of metals and environmental-sanitary risk analysis in the city of
Torino. Chemosphere 112:210–16.
HUMAN AND ECOLOGICAL RISK ASSESSMENT 405
Ozel G, and Cakmakyapan S. 2015. A new approach to the prediction of PM10 concentrations in
Central Anatolia Region, Turkey. Atmos Pollut Res 6(5):735–41.
Chang SC, and Lee CT. 2007. Assessment of PM10 enhancement by yellow sand on the air qual-
ity of Taipei, Taiwan in 2001. Environ Monit Assess 132(1–3):297–309.
Kinney GF, and Wiruth AD. 1976. Practical Risk Analysis for Safety Management, pp. 1–20.
Naval Weapons Center, California.
Montgomery DC. 2005. Introduction to Statistical Quality Control. Wiley, New York.
Tsai CC, and Chen CC. 2006. Making decision to evaluate process capability index Cp with fuzzy
numbers. Int J Adv Manufact Technol 30:334–39.