You are on page 1of 9

Renewable Energy 130 (2019) 613e621

Contents lists available at ScienceDirect

Renewable Energy
journal homepage:

Observation and modeling of dust deposition on glass tube of

evacuated solar thermal collectors in Mongolia
Purevdalai Erdenedavaa a, *, Atsushi Akisawa a, Amarbayar Adiyabat b,
Erdenesuvd Otgonjanchiv b
Graduate School of Bio-Applications and Systems Engineering, Tokyo University of Agriculture and Technology, Japan
School of Engineering and Applied Science, National University of Mongolia, Mongolia

a r t i c l e i n f o a b s t r a c t

Article history: The present study investigated the effects of dust deposition on transmittance of glass tubes of a solar
Received 1 February 2018 thermal collector in Ulaanbaatar during the cold period and presents the results based on real exposure
Received in revised form tests conducted between October 2015 and May 2016. In addition, the dust deposition on glass tubes was
29 May 2018
empirically modeled using the test results and the environmental data such as wind speed and direction,
Accepted 19 June 2018
Available online 20 June 2018
daily average airborne dust rate, snow, and rain. Based on observation of long- and short-term tests,
snow was deduced to be able to clean dust accumulation on the glass tubes even if the ambient tem-
perature is below zero. Also, the snow was found to be more effective in decreasing dust accumulation
Dust deposition
than rain according to the estimation.
Dry and cold region © 2018 Elsevier Ltd. All rights reserved.
Dust model
Solar-evacuated glass tube

1. Introduction Ulaanbaatar Mongolia is one of the five cities with worst air
quality and the coldest capital city in the world. In order to decrease
The cold climate and poor infrastructure in cities in Central and air pollution, the Mongolian government has been implementing
East Asia lead to air pollution from coal firing by detached houses, some measures of promoting and encouraging the use of eco-
which is one of the major issues during the winter season. Since the friendly technology, and established several monitoring stations
detached houses that have been spread out throughout the vast for measuring air pollution with the assistance of international
area cannot connect to the heating network of districts, raw coal organizations such as The World Bank, Asian Development Bank,
remains the dominant heating source for householders as it is cost- and Japan International Corporation Agency, since 2001. These
effective. stations continuously measure and report the data on air pollution
Solar thermal devices are suitable for domestic applications. and weather during the cold season [2e6]. The air pollution of the
Although solar thermal collectors are well applied in warm and city is composed of smoke from thousands of coal stoves, waste,
coastal regioins, the application in cold regions is relatively low due smoke from factories and plants, and fuel combustion of vehicles.
to some challenges and lack of sufficient studies. One of the chal- The city is divided into two sectors: city center and Ger areas. The
lenging issues is a prediction of the energy production during the pollution of small particulate matter (PM) emitted by coal stoves
winter season. The snow piles up behind and under evacuated tube accounts for >50% during the cold period in the city center, while it
solar collectors such that the reflected solar irradiance from the accounts for the vast majority of the Ger area as the factories and
piled snow hits the collectors from the back side as an additional basic lines of traffic are distant from the Ger districts [2,3].
energy source. However, direct and diffused radiations might not In recent years, the installation of evacuated tube collectors ETC
be able to fully penetrate up to the inside of the glass tubes because is increasing in Ger districts in Ulaanbaatar Mongolia; a few of them
of dust deposition from air pollution [1]. have been investigated since 2013 [1]. Owing to abundant dust
deposition on the glass tubes of solar collectors, we conducted a
pre-experiment on a real evacuated glass tube and typical flat glass.
They were tilted at 60 and exposed from December 2014 to March
* Corresponding author.
E-mail address: (P. Erdenedavaa).
2015 without cleaning under the actual ambient conditions.
0960-1481/© 2018 Elsevier Ltd. All rights reserved.
614 P. Erdenedavaa et al. / Renewable Energy 130 (2019) 613e621

Transmittance of the glass tube and flat glass was measured via two the nonlinearity. When the particles settle beside each other on a
pyranometers under global solar irradiance after allowing surface, the distance between them is not uniform. This phenom-
contamination for 3 months. Consequently, when the solar irradi- enon might be attributed to the formation of the clusters, following
ance was >650 W/m2, the reduction in transmittance of the front which, the airborne particles start to overlap on the clusters,
side of the glass tube was 35%, while that of the flat glass was 29% causing the particles to stick to a free area. Hegazy [20] studied the
due to dust deposition. As such drops can significantly affect the transmittance reduction as functions of tilt angles, dust deposition
energy production, the study of dust deposition is important. weigth and number of exposure days in Egypt with hot climate
Reviews by Travis et al. [7] and Mani et al. [8] are valuable while condition, and compared the obtained results with that of neigh-
addressing the dust problems. The deposition behavior and the boring regions. He recommended weekly regular cleaning for the
accumulation rates can dramatically vary depending on the two collectors in the desert belt. Generally researchers provide regular
primary factors: local environment and dust properties. In the local cleaning schedules for certain areas based on long-term studies
environmental factor, the following parameters could be involved: considering weigth, density and tickness of dust deposition, and
the amount of airborne dust, amount of rain and snow, air pollutant transmittance of sample glass. However, the regular schedule is not
sources (vehicle movement, fuel combustion, soil erosion, and ur- appropriate solution for users of solar devices because of weather
banization), wind velocity, and dominant wind direction. In the condition. Since the maintenance cost is high due to dificulty of
dust properties factor, the following characteristics could be cleaning in cold regions, it is very important when to clean.
involved: chemical composition, size, shape, stickiness, and elec- Recently, Wasim Javed et al. published a paper related to soiling loss
trostatic property. The characteristics of the dust factor depend of photovoltaic [21]. Their work was one of a few research works
strongly on the source of air pollution. In addition, the surface modeling the relationship between environmental variables and
factor is also critical. The tilt angle and smoothness of the surface daily change in performance loss. An experimental field was in hot
that is polluted also cause slow or fast dust accumulation. and dry region in Doha, Qatar. They used artificial neural network
Travis et al. [7] consolidated the studies related to dust depo- aproach (ANN) as a modeling method and environmental param-
sition since 1942, emphasized some of the progressive results, and eters such as PM10, wind and moisture were selected as predictor
mentioned that several types of mitigation techniques have been variables.
investigated in recent years. The study also mentioned that To summarize above, the following issues can be emphasized: I.
although several groups investigated the dust cleaning methods, There is no study investigating dust deposition behavior on the
water cleaning was a superior option. However, during the winter evacuated glass tubes in the cold and dry area with much air
season, in harsh cold regions such as Ulaanbaatar, the water pollution, II. It is not clear whether snow cleans the glass tubes, III. A
cleaning method is not suitable to clean the glass tubes due to method able to predict dust deposition rate and the change in
freezing. On the other hand, cleaning the glass tubes on detached transmittance, using weather data, is lacking and needs to be
houses’ roofs is quite difficult. These issues make maintenance cost developed.
of collectors high. Therefore, it is valuable to conduct a long-term dust deposition
The review by Mani et al. [8] categorized the climatic zones for test in that region for understanding the reduction in transmittance
mitigation measures against the impact of dust accumulation and of the glass tube and dust deposition behavior, and to develop a
showed the geographic distribution of studies addressing the prediction method employing weather data for the purpose of
impact of the dust on the performance of solar collectors. Also, the modeling the change in transmittance. It will provide fundamental
studies related to dust deposition in the cold and dry regions were information so that the ETC users can manage when they clean
found to be scarce, emphasizing that clearing of snow accumulation their solar collectors using daily weather news.
needs to be addressed. Currently, there is no study in Mongolia on Thus, the present study aimed to measure transmittance of real
dust deposition, although many mega solar projects have been glass tubes that were contaminated for various periods in a cold
planned for implementation. and dry site in Ulaanbaatar and model the change in transmittance
Several investigators conducted studies on the performance with respect to the factors such as airborne dust, wind direction,
degradation of photovoltaic PV and flat plate collector FPC due to wind speed, rain, and snow.
dust deposition [9e12]. However, studies on ETC are scarce. El- In order to model the dust deposition on glass tubes, long-and
Nashar [13e15] published three articles about the effect of dust short-term tests were conducted simultaneously. The role of the
deposition on the performance of ETC. The study was conducted on short-term dust depositon test SDDT was expected to indicate the
a large field of evacuated tube collectors associated with a multiple- influence of the environmental factors distinctly with a constant
effect distillation plant in Abu Dabi, UAE. The result showed that interval. Conversely, the long-term dust deposition test LDDT
when transmittance of the glass tubes decreased from the initial directly observed the effect of dust deposition and accumulation on
value of 0.98 to a low value of 0.6, the production of the distillation the glass tubes.
plant declined from 100% to 40%. In order to measure the trans-
mittance, a sample tube (un-evacuated) was used instead of real 2. Measurement of transmittance of glass tubes
glass tubes. He offered an empirical expression for the reduction in
light transmittance as a function depending on constant co- 2.1. Process to expose glass tubes to the environment
efficients for each month as well.
There are several studies that modeled dust depositon [16e21]. To evaluate the behavior of the dust, dust deposition tests were
The study by Ahmad and Al-Hassan was among the initial designs conducted between October 2015 and May 2016 under the real
[16]. The study mathematically derived a linear equation of altered environmental condition in Ulaanbaatar. A total of 30 glass tubes
transmittance. When transmittance is plotted against the number with heat pipes manufactured by 1st Sunflower Renewable Energy
of particles, the calculated straight line is in agreement with the Co., Ltd (Model: SFB305818) were exposed on a collector frame.
measured values of up to 50% decrease in light transmittance. Each glass tube is composed of a cover glass tube with diameter of
However, the model failed to account for this process due to the 58 mm, an inner glass tube with diameter of 47 mm and a metal
nonlinearity. Furthermore, based on this theory, Neil et al. [17] heat pipe component. The 30 glass tubes were equally divided into
modeled the free fractional area of a glass slide not filled with dust both LDDT and SDDT. The tubes for LDDT were exposed for various
particles. The study presumed that the clusters of particles cause periods extending from 4 to 32 weeks (for example, 4, 8, 12 … 32
P. Erdenedavaa et al. / Renewable Energy 130 (2019) 613e621 615

weeks). These tubes were left together under a real environmental and 120 cm from closed ends of the tubes.
condition at the beginning of October 2015, taken out of the frame
one by one every 4 weeks, and maintained in a safe wooden box in 2.3. Index of transmittance
order to measure the transmittances of all the glass tubes under the
same global solar irradiance. Each glass tube for SDDT was exposed In order to calculate the reduction, transmittance of each clean
for 4 weeks in every month between October and June. The first and dirty glass tube was measured simultaneously under global
tube was set on the frame at the beginning of October 2015, similar solar irradiance using three pyranometers.
to those for LDDT. After 4 weeks, the tube was taken out of the Transmittance of the pristine glass tube was calculated as
frame and kept in the box. Instead of the removed tube, the next below:
clean tube was mounted and exposed for 4 weeks. Such process
continued until the middle of May 2016. Fig. 1 shows the collector Gr1
t1 ¼ (1)
frame with the tubes for the dust deposition tests DDT; the frame K1 Go
was installed being directed to the solar noon at the tilt angle of 60
Transmittance of the polluted glass tube was calculated as
next to the operational ETC system in Tawan-buudal Ger district,
Ulaanbaatar (106º540 E and 47º570 N), where the operation and
performance of ETC system were continually observed and Gr2
measured for several years. In order to recognize front sides of all t2 ¼ (2)
K1 Go
exposed tubes, narrow adhesive tapes were put on those of each
tube. After finishing the exposure process, measurement of trans-
mittance was started in June 2016.
Fig. 2 shows the daily average relative humidity and ambient air 2.4. Uncertainty analysis
temperature measured with a portable station (CR-10x-Set) during
the exposure duration, which is next to the collector frame installed As dust distribution on the glass tubes is not uniform, trans-
by the house as shown in Fig. 1. mittance of the exposed tubes was measured several times under
The time chart below (Table 1) shows the days on which the the measurement criteria above. Each tube of LDDT and SDDT was
glass tubes for LDDT and SDDT were placed and taken out of the measured 5 times on the front side, while back and left sides of the
frame. Each tube was signed with different marks: I, …, VIII, or I0 , …, tubes exposed for LDDT were measured twice and once, respec-
VIII’. tively. Standard error of the mean was calculated as below [22].
u n
2.2. Measuring transmittance u
u tj  t 2
To measure transmittance of the polluted glass tubes, the device
Sn ¼ (3)
nðn  1Þ
shown in Fig. 3 was made. It includes 2 silicon pyranometers
(ML020) and a typical pyranometer (MS-802 Eiko Seiki), and can
hold 2 glass tubes simultaneously. One of the two must be a pristine
glass tube and it was not changed during that measurement, while 3. Modeling dust deposition
another one is either one of the dirty tubes and that dirty tube was
replaced with one another after measuring for 3e5 min under 3.1. Model formulation
global solar irradiance. In such way, each tube of LDDT and SDDT
was measured 5 times between June and October in 2016. Before Important factors that affect the change of dust deposition are
the measurement was started, open ends of all the exposed glass the local environment, dust properties and collector surface. As a
tubes were broken in order to take the inner glass tubes with se- dominant source of air pollution in the Ger district is smoke from
lective coating away from the cover glass tubes, so that the drop of thousands of coal stoves, it is assumed that dust properties are
transmittance of the dirty tubes can be measured with the silicon unchanged. The condition of collector surface is supposed to vary
pyranometers inserted into the tubes. Figs. 4 and 5 show the glass when the tubes are exposed for long period, which is discussed in
tubes without the inner glass tubes that were polluted by the the comparison between the results of SDDT and LDDT. Hence, the
exposure to SDDT and LDDT, respectively. main factor that can significantly affect the change of dust depo-
While measuring transmittance, the following criteria were sition is the local environment such as rain, snow, wind and
followed: sunny day with no clouds, global solar irradiance airborne particles (PM10).
>850 W/m2, the pyranometers normalized to the direct solar irra- In order to simplify the model, the function expressing trans-
diance, and the silicon pyranometers placed inside between 100 mittance was empirically formulated. It is considered reasonable

Fig. 1. Site of tube exposure tests in Ulaanbaatar.

616 P. Erdenedavaa et al. / Renewable Energy 130 (2019) 613e621

90 Humidity Ambient temperature 15

Daily average humidity, %

Daily average ambient T, C

80 10
70 5
60 0
50 -5
40 -10
30 -15
20 -20
10 -25
0 -30












Fig. 2. Daily average relative humidity and ambient temperature during the exposure period.

Table 1
Time chart of the exposure procedure.

Fig. 3. Device for measuring transmittance of dirty and clean tubes simultaneously.

that reduction in transmittance was proportional to airborne dust where A, B, C, and D are coefficients to be estimated; x, y, z1, and z2
rate and inversely dependent on rain. In contrast, the effects of are input variables which represent daily average PM10, daily
snow and wind on dust deposition on the glass tubes are not yet average wind factor derived from minutely measured wind data,
clarified. Also, it can be said that the input parameters shown in daily rain and daily snow, respectively.
Figs. 6 and 7 are uncorrelated to each other. The following As shown in Eq. (4), t1 is the upper limit of transmittance
expression was adopted to predict the change of transmittance of because it is detected on the clean tube. It should be noted that the
the glass tubes: model allows transmittance to recover the clean condition, which is
recognized as non-linear behavior.
< t2;i ¼ t2;i1  A xi þ B yi þ C z1;i þ D z2;i In Eq. (4), all variables are direct measurement values except for
t ; if t2;i  t1 (4) the wind factor y. The study by Goossens et al. [23] mentioned that
: 2;i
t1 ; if t2;i > t1 high wind speed increases the dust deposition on the surface,
which implied that the wind speed was primarily supposed to
P. Erdenedavaa et al. / Renewable Energy 130 (2019) 613e621 617

function represented as below:

yi ¼ 1  eeLðS t Q Þ wt (5)
T t¼1

1; if wind comes from the south; southeast or southwest

0; if wind comes from the north; east or west

where L and Q are coefficients to be estimated, S is the measured

data of wind speed, and w is the measured wind direction. If the
wind blows from the front side of the glass tubes, w ¼ 1, and in the
reverse case, w ¼ 0. The coefficient L expresses the effect of wind
speed on the reduction in transmittance, while Q indicates the
point where the effect of wind speed changes from positive to
The model was formulated as an optimization method in which
the objective function is to minimize the sum of squared errors
Fig. 4. Tubes exposed for 4 weeks in every month in SDDT. between estimated and measured values. As there wasn’t any data
of PM10 since April 30, 2016, the measument results of VIII and VIII0
weren’t involved into the analysis. Thus, measurement results of 7
tubes exposed between Oct 1, 2015 and April 16, 2016 were used to
estimate the coefficients of the model.

7 h
X i
Objective function ¼ tm
2;i ðkÞ  t2;i ðkÞ
/Min (6)

where tm2;i ðkÞ

represents the measurement result from the dust
deposition tests, t2;i ðkÞ e the estimated transmittance.

3.2. Input variables

The following data were measured for the model: airborne dust
<10 mm in diameter (PM10), daily rain and snow rate (precipitation),
and wind speed and direction. The data of PM10 and precipitation
were obtained from the nearest station (100-Ail station) that
measures the daily air pollution rate and some weather parameters.
Fig. 5. Tubes exposed for various periods from 4 to 32 weeks in LDDT. The station was about 1 km distal from the exposed site, which is
also in the Ger district. The speed and direction of the wind were
measured with the portable weater station next to the house. The
wind was measured at the height similar to that of the top of the
frame, and the data were logged every 2 min. Fig. 6 shows the daily
average PM10 and daily precipitation rate measured from October 1,
2015 to May 1, 2016. Fig. 7 shows the daily average wind speed and
the daily fraction of the wind that blew from the south, southeast
and southwest. The fraction is calculated as below:

1X T
W¼ wt (7)
Fig. 6. Daily average airborne dust (PM10), daily total rain and snow rate. T t¼1

As shown in Fig. 6, the rains were prevalent until the end of

100 4 October 2015, followed by snow from the beginning of November
Fraction of southern wind, W Daily average wind speed, S
until the beginning of March 2016. After that, the rain started again
Fraction of southern wind,

Wind speed, m/s

in March. The total precipitation rate in January and February was
2 2.8 mm, which was only about 7% of the total precipitation between
the measurement durations. Also, the figure shows that December,

20 January, and February had much air pollution, especially on January
0 0 12, 2016, the daily average PM10 was 1946 mg/m3, which was about














20-fold higher than the Mongolian air quality standard [2].

As shown in Fig. 7, the daily average wind speed was high in
Fig. 7. Daily average wind speed and daily fraction of the southern wind. spring and autumn and low in winter. This phenomenon was
similar to the data measured by stations for several years. However,
influence the reduction in transmittance. Therefore, the effect of the evaluation of wind direction differed from that of the statistical
the wind factor would be expressed as a non-linear monotonous data, which was related to the location of the exposure site and the
618 P. Erdenedavaa et al. / Renewable Energy 130 (2019) 613e621

surrounding objects. The exposure site was on the left front pied- deposition. According to the estimation, snow had a 5-fold large
mont and leeward side of Chingeltei Mountain. The dominant wind coefficient than rain regarding magnitude.
direction of Ulaanbaatar is from the west and north-west. The estimated coefficients suggested a critical value regarding
the effect of wind. When wind speed was >1.8 m/s, the effect was
4. Results negative, which was in contrast to that postulated by Goossens
et al. It implies, in Ulaanbaatar, the wind with slow velocity
4.1. Measurement result of SDDT and modeling enhanced the dust deposition, while the wind with high speed
decreased the deposition.
4.1.1. Measurement results.  
Transmittance of each tube, exposed for SDDT, was measured t2;i1  t2;i ¼ 2:28x þ 311 1  e0:05ðS1:80Þ w  8:87z1
from the front side under global solar irradiance >850 W/m2  
(Fig. 8). Transmittance of the clean tube was measured to be 0.96. A  47:4z2 R2 ¼ 0:963 (8)
statistical test was employed to examine the difference of trasn-
mittance between the exposed tubes and a clean tube. According to Using the equation above and the input variables, the change of
t-test, all measurements are significant less than 5% of probability. transmittance is estimated as shown in Fig. 9. The filled circles
For all measured dirty tubes, standard error of the mean was indicate the measurement values of the exposed glass tubes for
<0.013. SDDT. The dotted lines indicate the daily behavior of transmittance.
The high level of airborne dust led to high dust deposition. The graph shows that the estimation gives good agreement with
However, Fig. 8 indicated that when the airborne dust was the the high R2 value of 0.963.
highest in January, transmittance of the IV0 glass was higher than
that of the III0 and V0 glasses. This transmittance could be explained
by snowfall that occurred 1 day before the IV0 glass was taken out of 4.2. Measurement result of LDDT and modeling
the frame. Therefore, the snow was able to clean the glass tubes
similar to rain even if the ambient temperature was below zero. 4.2.1. Measurement results
Transmittance of each tube exposed for LDDT was measured
4.1.2. Modeling results from the front side, back side, and left side (at 45 from the front
Using the measurement result of SDDT and the input variables, side) under global solar irradiance >850 W/m2 (Fig. 10). The graph
Eq. (8) was derived. The equation indicated that the airborne dust shows that transmittance of the front side decreased significantly
rate was positively proportional to the reduction in transmittance, to 0.5 during winter, while that of the back side was at least 0.84.
and rain exhibited an inverse dependence as expected. Notably, According to t-test, they are all significant except that the mea-
snow was more effective than rain in decreasing the dust surement of back side of the glass tube exposed in October is found
to be insignificant with the criteria of 5% of probability. Standard
error of the mean was <0.017 for all back and front measurements
1.00 0.91 0.91 0.94 0.94 except the front side measurement of VI tube, which was 0.029. It
0.84 0.83
0.77 was considered to be caused by smears and flecks due to snow and
0.80 rain in March.

In terms of the front side of the glass tubes, transmittance of the

0.60 V glass was minimally 50%, which was exposed for 20 weeks from
October 3, 2015 to February 20, 2016. Fig. 10 also indicates that
0.40 transmittance increased gradually from March. Finally, the tubes
were recovered to the initial level in May, which can be explained
0.20 by high precipitation and decreased airborne dust concentration.
Therefore, transmittance of the VI glass, polluted for 24 weeks, was
0.00 higher than that of the V glass. With respect to the left and back
sides of the glass tubes, transmittance of the VI glass was the
Oct Nov Dec Jan Feb Mar Apr May
lowest. Supposedly, the snow and rain moved the dust from the
Fig. 8. Transmittance of glass tubes exposed for 4 weeks in every month (SDDT). front side to the left and back sides of the glass tubes (Fig. 11).

Fig. 9. Estimated daily behavior of transmittance of the glass tubes by SDDT model.
P. Erdenedavaa et al. / Renewable Energy 130 (2019) 613e621 619

1.20 Front Left Back 5. Discussion





The measured values of LDDT and SDDT were used for deriving





the coefficients such as A, B, C, D, Q, and L independently for



determining the long- and short-term dust deposition behavior.


Quantitatively, the coefficients of A, Q, and L were similar for both

0.60 LDDT and SDDT, while B, C, and D were different.
The coefficients of B, Q, and L described the effect of wind on the
reduction in transmittance. The coefficient of L was estimated to be
negative for both tests, which implied that transmittance was
decreased when the wind speed was less. Conversely, the trans-
0.00 mittance was increased when the wind speed was sufficiently high.
The strong wind blows away the dust particles on the glass tubes.
Oct Nov Dec Jan Feb Mar Apr May
On the other hand, dust accumulates on the glass tubes when the
Fig. 10. Transmittance of glass tubes exposed for various periods from 4 to 32 weeks. wind speed is slight or zero.
The coefficient B of Eq. (9) for LDDT was more than 2-fold than
that of Eq. (8) for SDDT. The coefficient B implicitly expressed the
4.2.2. Modeling results dust settlement on the glass tubes. The larger the coefficient value,
When the measurement results of the front side of LDDT were more the dust settlement while the wind speed is small. Suppos-
used instead of SDDT, the following equation was derived. The signs edly, the glass tubes that have been exposed for long-term are
of the coefficients obtained from Eq. (9) were identical to those of rougher than the glass tubes exposed for short-term such that the
Eq. (8). However, the quantitative values of the coefficients are particles settle easily and accumulate more on the glass tubes in
different. The result of LDDT also indicates that snow was more LDDT than in SDDT.
effective in reducing the dust deposition than rain. The critical The coefficient D is the effect of snow on the reduction in
value of the wind speed was also estimated to be similar to that of transmittance. Both Eq. (8) and Eq. (9) indicated that the cleaning
SDDT. ability of snow is higher than that of rain. However, the values of
the D coefficient are different. Also, the cleaning ability of snow in
SDDT is better than that in LDDT.
  In addition, the cleaning ability of rain in SDDT is much better
t2;i1  t2;i ¼ 2:57x þ 662 1  e0:08ðS1:75Þ w  0:56z1 than that in LDDT, which could be explained by Cuddihy [24e26].
  The field observations in the study described that the dust accu-
 32:9z2 R2 ¼ 0:993 (9) mulation on the surface consists of three soil layers termed as A, B,
and C indicating primary, secondary, and top surface layer,
As a result of the equation above, the change of transmittance of respectively. The layers A and B are formed within 30e60 days due
the glass tubes exposed for long-term was estimated, which is to dew, humidity, and light rain and are extremely sticky such that
shown in Fig. 12. The filled circles indicate the measurement values rain and soft hand washing could not clean them completely. The
of the exposed glass tubes for LDDT. The measurement and the layer C is formed after A and B with weak stickiness that could be
estimation are in good agreement with the R2 value of 0.993.

Fig. 11. Images obtained on March 4, 2016: A. After heavy snowfall in the morning, B. Snow on the tubes starting to melt, C. Dust on the front side was cleaned by the melting snow,
D. Tube without inner glass, which was brought in laboratory after the heavy snow.
620 P. Erdenedavaa et al. / Renewable Energy 130 (2019) 613e621


Transmittance 0.80



0.20 Estimated Measured




















Fig. 12. Estimated daily behavior of transmittance of the glass tubes by LDDT model.

removed by light rain. The exposure period of SDDT was not pro- dust during the winter season and transmittance of the glass tubes
longed in the case of layers A and B; rain or snow could easily clean decreased significantly. However, transmittance could be recovered
the dust on the glass tubes. However, in the case of LDDT, the dust to the initial level due to snow and rain, thereby implying that the
on the glass tubes could not be removed easily after the layers A and time and amount of rain and snowfall are critical factors for pre-
B were grown on the surface, with a small magnitude of the co- dicting the dust deposition.
efficients C and D. Both long- and short-term behaviors of the tube transmittance
Furthermore, the authors reported the measured efficiency of were modeled using the results of LDDT and SDDT and the envi-
the solar collector installed at the site of the dust deposition test [1]. ronmental data. Consequently, PM10 had a negative effect on
The measurement and LDDT were conducted simultaneously. In transmittance, while rain and snow exerted a positive effect.
addition, transmittance of the solar collector tubes was affected by Conversely, the effect of wind speed was positive or negative
the dust deposition, which in turn, altered the performance of the depending on the speed. Thus, the magnitude of some coefficients
collector. Fig. 13 presents the relationship between the trans- of the equations was different between SDDT and LDDT, thereby
mittance measured in LDDT and the efficiency. The efficiency was suggesting that dust accumulation and natural cleaning mecha-
defined as the ratio of heat recovered by the collector to the input of nisms differed in both long- and short-term dust depositions.
solar energy. The plots represented the data of the days when the Thus, the cleaning ability of snow against the dust deposition is
tubes were taken out from the field for LDDT. Interestingly, a cor- more effective than rain, although the ambient temperature is
relation was established between the two factors in Fig. 13, which sufficiently cold in winter.
implies that the dust deposition on the collector tubes degraded the
performance of the solar heat recovery.

6. Conclusion This work was supported by the Mongolian-Japan Engineering

Education Development project/M-JEED/[JR code: J13A15].
The current work is the first study addressing the dust deposi-
tion behavior on the glass surface of evacuated tube collectors in
cold and dry region of Ulaanbaatar.
Herein, the long- and short-term dust deposition tests were
A Coefficient that expresses the effect of airborne particles
conducted under ambient condition, and transmittance of the glass
(PM10) on the reduction in transmittance
tubes exposed for the tests measured. To summarize the results
B Coefficient that expresses the effect of wind including
obtained from the tests, the glass tubes were easily polluted by the
wind direction and speed on the reduction in
65% C Coefficient that expresses the effect of rain on the
Solar collector efficiency

reduction in transmittance
60% D Coefficient that expresses the effect of snow on the
reduction in transmittance
55% G0 Global solar irradiance measured out of the glass tubes,
50% [W/m2]
Gr1 Global solar irradiance measured from a pristine glass
45% tube inside, [W/m2]
Gr2 Global solar irradiance measured from a dirty glass tube
40% inside, [W/m2]
35% i Index of days
j Index of measurements
30% k Index of tubes
30% 50% 70% 90% K1 Calibration coefficient
L Coefficient that expresses the dust settlement when the
Transmittance wind speed is 0
Fig. 13. Solar collector efficiency vs. transmittance of the glass tubes. n The number of measurements on each tube
P. Erdenedavaa et al. / Renewable Energy 130 (2019) 613e621 621

Q Wind speed, at which, airborne particles start to (2013) 698e733.

[8] M. Mani, R. Pillai, Impact of dust on solar photovoltaic (PV) performance:
accumulate on the glass tubes or remove from the glass
research status, challenges and recommendations, Renew. Sustain. Energy
tubes with regard to wind speed, [m/s] Rev. 14 (2010) 3124e3131.
Sn Standard deviation of the mean [9] Muhammed J. Adinoyi, Syed A.M. Said, Effect of dust accumulation on the
S Wind speed, [m/s] power outputs of solar photovoltaic modules, Renew. Energy 60 (2013)
T The number of daily measurement steps of wind, it [10] Motasem Saidan, Abdul Ghani Albaali, Emil Alasis, John K. Kaldellis, Experi-
equals 720 mental study on the effect of dust deposition on solar photovoltaic panels in
t Measurement step of wind is 2 min desert environment, Renew. Energy 92 (2016) 499e505.
[11] Jie Deng, Xudong Yang, Ming Yang, Zhifeng Wang, Experimental study of a
W Daily fraction of southern wind direction single-pass flat plate solar air collector with severe dust deposition on the
w Wind direction, [0e360⁰] transparent glass cover, Energy Procedia 70 (2015) 32e40.
x Daily average airborne particles PM10, [mg/m3] [12] Zhao Jing, Wang Zhiping, Wang Kezhen, Liu Jianbo, Dust effect on thermal
performance of flat plate solar collectors, J. Sol. Energy Eng. 137(1) 014502,
y Wind factor depending on the wind speed and direction doi: 10.1115/1.4028364.
z1 Total daily rain, [mm] [13] A.M. El-Nashar, Effect of dust accumulation on the performance of evacuated
z2 Total daily snow, [mm] tube collectors, Sol. Energy 53 (1994) 105e115.
[14] A.M. El-Nashar, Effect of dust deposition on the performance of a solar
t1 Transmittance of a pristine glass tube desalination plant operating in an arid desert area, Sol. Energy 75 (2003)
t2 Transmittance of a dirty glass tube 421e431.
t Arithmetic mean of n times measurement [15] A.M. El-Nashar, Seasonal effect of dust deposition on a field of evacuated tube
collectors on the performance of a solar desalination plant, Desalination 239
(2009) 66e81.
References [16] Ahmad Y. Al-Hasan, A new correlation for direct beam solar radiation received
by photovoltaic panel with sand dust accumulated on its surface, Sol. Energy
[1] Purevdalai Erdenedavaa, Amarbayar Adiyabat, Atsushi Akisawa, 63 (No. 5) (1998) 323e333.
Erdenesuvd Otgonjanchiv, Performance analysis of solar thermal system for [17] Neil S. Beattie, Robert S. Moir, Charlslee Chacko, Giorgio Buffoni, Simon
heating of a detached house in harsh cold region of Mongolia, Renew. Energy H. Roberts, Nicola M. Pearsall, Understanding the effects of sand and dust
117 (2018) 217e226. accumulation on photovoltaic Modules, Renew. Energy 48 (2012) 448e452.
[2] The World Bank, Air Quality Analysis of Ulaanbaatar; Improving Air Quality to [18] H.K. Elminir, A.E. Ghitas, R.H. Hamid, F. El-Hussainy, M.M. Beheary,
ReduceHealth Impacts, Washington, DC 20433 USA, 2011. https:// K.M. Abdel- Moneim, Effect of dust on the transparent cover of solar collec- (Accessed 5 October tors, Energy Conserv. Manag. 47 (2006) 3192e3203.
2017). [19] F. Mejia, J. Kleissl, J.L. Bosch, The effect of dust on solar photovoltaic systems,
[3] Metropolitan Air Quality Agency Fulfilled Works Regarding to Reduce Air Energy Procedia 49 (2014) 2370e2376.
Pollution on 2012e2013, Full Report for Mongolia Ulaanbaatar, 2013. http:// [20] Adel A. Hegazy, Effect of dust accumulation on solar transmittance through glass covers of plate-type collectors, Renew. Energy 22 (2001) 525e540.
(Accessed 5 October 2017). [21] Wasim Javed, Bing Guo, Benjamin Figgis, Modeling of photovoltaic soiling loss
[4] Final Report of Japan international cooperation agency, JIKA Capacity Devel- as a function of environmental variables, Sol. Energy 157 (2017) 397e407.
opment Project for Air Pollution Control in Ulaanbaatar City Mongolia, 2013, [22] Semyon G. Rabinovich, Evaluating Measurement Accuracy: a Practical (Accessed 5 October Approach, Springer, New York, 2010.
2017). [23] D. Goossens, E. Van Kerschaever, Aeolian dust deposition on photovoltaic
[5] K. Guttikunda, Sereeter Lodoysamba, Baldorj Bulgansaikhan, solar cells: the effects of wind velocity and airborne dust concentration on cell
Batdorj Dashdondog, Particulate pollution in ulaanbaatar Mongolia, Air Qual performance, Sol. Energy 66 (1999) 277e289.
Atmos Health 6 (2013) 589e601. [24] E.F. Cuddihy, Theoreticalconsiderations of soil retention, Sol. Energy Mater. 3
[6] Perry K. Davy, Gerelmaa Gunchin, Andreas Markwitz, William J. Trompetter, (1980) 21e33.
Bernard J. Barry, Dagva Shagjjamba, Sereeter Lodoysamba, Air particulate [25] E.F. Cuddihy, Surface soiling: theoretical mechanisms and evaluation of low-
matter pollution in Ulaanbaatar, Mongolia: determination of composition, soiling coatings, in: Proceedings of the Flat-plate Solar Array Project Research
source contributions, and source locations, Atmospheric Pollu. Res. 2 (2011) Forum on Quantifying Degradation, 1983, pp. 379e396.
126e137. [26] E.F. Cuddihy, P.B. Willis, Antisoiling Technology: Theories of Surface Soiling
[7] Travis Sarver, Ali Al-Qaraghuli, Lawrence L. Kazmerski, A comprehensive re- and Performance of Antisoiling Surface Coatings. Jet propulsion Laboratory
view of the impact of dust on the use of solar energy: history, investigations, Report, 1984. DOE/JPL 1012-102.
results, literature, and mitigation approaches, Renew. Sustain. Energy Rev. 22