MEASURING LOSSES IN POWER PRODUC TION DUE TO SOILING OF SOLAR PANELS

A jay S ingh (as ingh@cam p b e l lsc i .com ) a n d Ma tth e w Pe r r y ( m p e r r y@ca m p b e l lsc i .com )Ca m p b e l l S c ie n t i�c, In c. Log a n , UT USA

INTRODUC TION EXPERIMENTAL PROCEDURE

As photovoltaic ar rays have increasingly generated more electr ical energy in the past decade, they have establ ished themselves as v iable sources of renewable energy. At present , the e�ciency of solar panels is at best 15-23%, meaning only 15-23% of incident energy in the form of l ight get conver ted into electr ical energy. There are fur ther per formance losses in the l i fe t ime of a solar ar ray. Loss due to dust accumulat ion on power generat ion is becoming increasingly impor tant as the total instal led s ize increases. System owners have to �nd a way to assess losses in energy production due to var ious factors. Incidental ly, there are large PV instal lat ions in dr y deser t regions where natural ra in events are scarce and loss in energy production is encountered due to dust accumulat ions on the solar panels result ing in soi l ing losses. The decis ion to c lean these panels can be cost ly and the owners need a monitor ing system to make this deci-s ion in a cost e�ect ive manner. Several studies have been made in the past to quanti fy these losses. R yan et a l .

1 compared output of a c lean panel

versus an identical test panel accumulat ing dust natural ly. Several other researchers2,

3 est imated the loss in energy

production f rom the plant per formance data . This paper descr ibes a �eld deployable soi l ing loss monitor ing system that is based on the method adopted by R yan et a l .

1, and is easy to instal l . I t provides a dai ly averaged soi l ing loss index (SLI ) that can be used to make cleaning de-

cis ions. The SLI is calculated from e�ect ive i r radiance reaching the cel ls calculated form the shor t c i rcuit current of the panels. This method is s imple and avoids the complicat ion and cost of a ful l I -V cur ve tracer. We also present some results f rom test ing at our s i te in Logan, UT.

Two identical ful l s ize PV panels are mounted s ide by s ide. The shor t c i rcuit current is measured on both panels with a precis ion current sensing shunt . Back of-module temperatures are a lso measured using a thermistor. Measurements are per formed dur ing sunset and sunr ise. The two modules are shor t- c i rcuited using a sol id state relay for about 5 second ever y 30 seconds. The shor t- c i rcuit current is measured using a current shunt . Back- of-module temperatures on both modules are a lso monitored.

An e�ect ive i r radiance reaching the cel ls is calculated in accordance with the IEC 609045, according to the equation 2 above. An init ia l o�set between the two panels is est imated in the beginning and per iodical ly af ter wards each t ime both panels are c leaned. This o�set is appl ied to the subsequent measurements. In accordance with IEC 60904, we l imit the SLI calculat ion within one hour of local solar noon and for e�ect ive i r radi-ance > 800 W/m2. This avoids any di�erences in soi l ing loss due to zenith angle of sun, module current dependence on i r radiance level or spectral di�erences. We also use a �lter ing scheme to �lter out any unstable data due to c louds etc. The stabi l i ty cr i ter ia adopted is based on the recommendations made in IEC 60904.

EXPERIMENTAL PROCEDURE

KEY DEFINITIONSWe begin by a de�nit ion of the soi l ing loss index in terms of quantit ies that can be measured direct ly f rom a PV module. We de�ne Soi l ing Loss Index (SLI ) as loss in the i r radiance reaching the solar cel ls . I f a l l other factors are same then this loss is pr imar i ly due to the loss in transmiss ion proper t ies of the glass as a result of the soi l accumulat ion. The i r radiance is calculated from shor t c i rcuit current

6 as

𝐺𝑒𝑓𝑓=𝐼𝑆𝐶 [1−∝(𝑇−𝑇𝑜)] 𝐼𝑆𝐶,𝑆𝑇𝐶  Where, 𝐺𝑒𝑓𝑓 : E�ect ive i r radiance reaching the solar cel ls𝐼𝑆𝐶 : measured shor t c i rcuit current of the module𝐼𝑆𝐶,𝑆𝑇𝐶 : shor t c i rcuit current at STC𝑇 : back of module temperature𝑇𝑜 : back of module temperature at STC, typical ly 25° Cα: Temperature coe�cient of shor t c i rcuit current .

The soi l ing loss index (SLI ) uses the e�ect ive i r radiances of a c lean reference panel and a dir ty test panel , and is de-�ned as :

𝐺𝑒𝑓𝑓,𝑅𝑒𝑓 𝐺𝑒𝑓𝑓,𝑇𝑒𝑠𝑡

Where, 𝐺𝑒𝑓𝑓,𝑅𝑒𝑓 i s the e�ect ive i r radiance calculated from the clean reference panel and 𝐺𝑒𝑓𝑓,𝑇𝑒𝑠𝑡 i s the same quan-t i ty calculated from the test panel .

REFERENCES1. C . P. R yan, F. Vignola , and D. K . McDaniels , “Solar Cel l Arrays : Degradat ion Due to Dir t ,” in Proceedings of the A mer ican S e ct i on of t he Inte r na t i ona l S olar Energy S ociet y , Denver, CO, 1989, pp. 234–237.

2 . A . K imber, L . Mitchel l , S . Nogradi , and H. Wenger, “ The E�ect of Soi l ing on Large Gr id-Connected photovoltaic Systems in Cal i fornia and the Southwest Region of the United States,” in Conf erence R ecord of the 2006 IEEE 4th Wor ld Con fe re nce on Photovol t a i c Energy Convers ion , Waikoloa, HI , 2006.

3 . S . Canada, “Qual i ty Assurance: Impacts of Soi l ing on Uti l i ty-Scale PV System Per formance,” S olarPro Magazine , no. 6 .3 , pp. 14–20, May-2013.

4 . K . Emer y, “Radiometr y in in photovoltaics : cal ibrat ion of reference solar cel ls and evaluat ion of reference values”. In Handbook of Photovol t ai c S c ience and Engineer ing , 1st ed. , Wi ley, 2003, pp. 701–752.

5 . IEC 60904-1, Photovoltaic devices – Par t 1 : Measurement of photovoltaic current-voltage character ist ics, second edit ion, 2006-9

6. IEC 60904-10, Photovoltaic devices – Photovoltaic devices –Par t 10: Methods of l inear i ty measurement, second edit ion, 2009.

1−𝑆𝐿𝐼= ×100%

RESULTS

Figure 1 : E � ec t ive I r radiance as calc ulated f rom s hor t c i rc uit c ur rent of the PV m odules on a c lear s k y day. The green c ur ve s hows the s oi l ing index. The S LI depends on the t im e of the day.

Figure 3 : E � ec t ive I r radiance and S LI on a c loudy day. Figure 4 : G e� and SLI taken dur ing a c loudy day with G e�ref >800w/m 2 dur ing ±1 hour of solar noon.

Figure 5 : Change in S L I af ter a c leaning e vent . Both m odules were c leaned at a t im e cor res ponding to the record num ber 6 5 0 . The c leaning a� ec ts the tem -perature of the m odules and i t takes s om e t im e to recover f rom this tem pera-ture c hanges. The S LI s ett les bac k near zero af ter s om e t im e. H igher S L I before record num ber 6 5 0 is the indicat ion of d i r t y m odules with los s in e� ec t ive i r ra-diance ~ 2 %

I m age 1 : Photo of soi l ing test s i te at Campbel l S c ient i�c, Inc. , Headquar ters , Logan, UT, USA.

Figure 2 : E �ect ive i r radiance and SLI dur ing ±1 hour of solar noon and G e�ref > 800 W/m 2. The t ime of the day dependence is absent dur ing this t ime.

For more information on the CRSI2 Soi l ing Index Datalogger, please vis i t : https://www.campbel lsci .com/crsi2

28 August—1 September 2016