In Australia telecommunications networks have utilised solar power to support critical loads since the early 1970’s. These systems have formed the back bone for the powering of major and minor networks ever since. As major long haul, inter capital radio systems were replaced with optical fibre systems, these solar systems continued to support the load requirements without issue.
The renewable energy industry will normally showcase the impressive vista of large scale solar installation and compete for attention, but rarely do they show how the panels look after twenty to thirty years once they have been subjected to the elements.
This paper is not intended to be a scientific address on manufacturing techniques but a pragmatic review of panel performance.
This paper will examine some typical examples of the degradation of two brands of solar panels over their lifetime. The paper provides an insight into when asset owners and operators should consider replacing the panels and if in fact the degradation in output warrants the replacement. Manufacturers will often provide warranties to 80% of rated output at 20 years; so how do operators determine ‘end of life’ and rate efficiency of solar panels getting to 20 years operation – are they able to safely support critical infrastructure?
Consideration is given to utilising the legacy solar array systems with “on grid” or “off grid” applications.
Data and samples are taken from recovered solar panels that have been used in the harsh Australian outback with temperatures ranging from -2 degrees to + 50 degrees Celsius with relative humidity up to 98%. The seasonal transitions in the regions where solar systems are typically deployed are extreme and violent, with areas of the Australian coastline
subjected to severe tropical cyclones to Category 5 with wind speeds to 300 kph .
Figure 1. Cyclone Damaged Solar Array.
The other extremes include snow capped alpines of the south east of Australia. Samples used have been subjected to the high extreme temperatures with humidity levels up to 98% from the north west of Western Australia with two samples recovered from the Nullarbor Plain in the south with low winter temperatures (- 2 degree Celsius) and high summer temperatures to 45 degree Celsius.
The particular samples used in this assessment were manufactured in France by Phillips  and in Australia by Solarex, the latter having since ceased production in Australia. The panel year of manufacture ranged from 1979 through to 1986 with six of each brand used for the initial assessment, Refer Table 1.
Table 1 DETAILS OF SAMPLE PANELS USED IN THIS REVIEW.
The fundamental differences between the two panels relate to the construction methods:
The Phillips BPX 47C panels were constructed with the single crystal silicon wafers sandwiched between two layers of clear glass. These are held in place with a polymer resin to seal the cells from moisture ingress and provide strength to resist thermal and mechanical shock.
Figure 2. Phillips BPX47C panels.
Solarex LX81 BGT series panels utilised a silicon rubber backing sheet that was heat vacuum sealed and white in colour. This method is still often utilised by the majority of panel manufacturers but with the use of cheaper ethylene vinyl acetate (EVA) products and Tedlar.
Figure 3. Solarex X81BGT panel. (The Good)
These panels were deployed in their thousands across Australia  and whilst end of life (EOL) replacement
programs have recovered a large number, many are still in use today.
Many of the EOL replacement criteria are subjective and can lead to either replacing the panels too early or simply too late.
The criteria included:
II. Cracks in the glass. III. Shattered glass. IV. Corroded inter-cell tracks. V. Dirt build up in the substrate between the glass
surface and backing. VI. Junction terminal failures.
VII. Discolouration of the wafers. VIII. High resistance interconnect joints from corroded
solder joints: and IX. Perceived reduction in output.
The point, at which degradation should be acted upon, is highlighted in the following sections.
The output characteristics of each panel were measured and compared against the original manufacturer’s data to determine the level of degradation and assess if the degradation is directly attributed to any one of the above causes.
2 CRACKS IN THE PANEL GLASS - BPX47C SERIES.
This is most likely caused by the prevailing weather conditions such as severe cyclonic storm events with objects or material being lifted from the ground and/or hail which has struck the front or rear of the glass laminated panels. This has allowed moisture to enter the laminated glass as it heats and cools and tracks along the interconnecting wiring and onto the actual wafers. Given the high costs of solar panels (in the 1980’s and 1900’s), attempts were made to seal the cracks in the glass. This proved futile as the cracks extended to the inaccessible parts of the support frame allowing moisture to continue entering at those points or by the time the cracks were found the moisture had already entered the laminate. With surface temperatures on the wafers as high as 60 degrees Celsius the cracked laminated glass allowed atmosphere to enter the panel. The process of delaminating continued to a point where the resin was allowed to heat up and run to lower parts of the void consequently holding the glass apart by as much as 6mm. This void simply became a well for moisture and contaminants to promote decay and corrosion. This is clearly evident in Figure 4.
Figure 4. Delaminated BPX47 C panel.
3 SHATTERED GLASS – LX81BGT.
These panels suffered the same fate as the glass laminated BPX 47C panels when subjected to severe weather. In this case with the tempered glass, EVA and tedlar backing sheets the seals were rarely broken, with little or no signs of moisture ingress from the rear. Moisture had entered the panel through crack lines in the glass and has commenced discolouring (brown) the silicon wafer and interconnection tracks. This is highlighted in Figure 5 where the discolouration follows the cracks.
Figure 5. Moisture ingress on X81BGT shattered panel, (The Bad)
4 CORRODED INTER-CELL TRACKS.
Corrosion is highly visible on the damaged BPX 47C laminated glass modules. Once the moisture had entered the panel it crept along the tracks causing corrosion.
5 DIRT BUILD UP IN THE SUBSTRATE BETWEEN THE GLASS SURFACE AND BACKING.
This issue was not as evident as the moisture damage.
6 JUNCTION BOX FAILURES.
The junction boxes fitted to the Phillips BPX 47CF clear glass panels were directly exposed to the sun and have become very brittle and disintegrated exposing the connections terminals and diodes. The earlier model BPX47C panels were fitted with diecast aluminium junction boxes that were in very good condition internally and externally. It should be noted that the cable entry conduits had to be sealed with silicon compounds to prevent moisture from travelling up the conduits and collecting in the junction boxes causing terminal corrosion and high resistance joints.
The junction boxes fitted to the Solarex LX81BGT panels are protected from direct sunlight by the silicon rubber backing to the panels. As a result the junction box plastics were in good condition and were not brittle. These particular junction boxes do not provide a positive seal and as such allowed any collected moisture to evaporate or drain off without adversely affecting the connection terminals.
All junction boxes are mechanically fixed to the panel support frame. It was found on later model panels that the junction box is secured to the tedlar with common silicon which “falls off” after a few years compromising the connection cables and the seal where they enter the panel face. Strong winds then blow the junction box around causing further damage and stress to the module cabling.
7 DISCOLOURATION OF WAFERS.
Both module types experienced discolouration.
Panels operating in the Australian tropics (to approximately16 degrees south of the equator) with high levels of humidity suffer from severe mould growth and discolouration whilst the same panel operating in the drier southern regions of Australia appear to be in near perfect condition.
This mould growth condition is exaggerated if the glass has been damaged. Field inspections have found this to be an ongoing issue even with more modern panels manufactured to this day.
On a number of the BPX47C panels the resin between the layers of glass surrounding each wafer has been badly disturbed and would appear to be caused by hot spots as a result of shading. As noted in Intelec papers from 1984 “Solar Power For Telecommunications – The Last Decade” . The reverse currents can excessively heat the wafer melting and bubbling the surrounding resin. The damage and disfigurement has allowed moisture to enter these crevices and voids.
Figure 6. BPX 47 C Deteriorated Laminated Panel (The Ugly)
On at least one sample panel (LX81BGT) some wafers displayed a symmetrical brown/orange discolouration in the centre of the wafers covering about 70% of the surface of each wafer. These particular panels had cracks in the rear silicon rubber backing sheets. Whilst not aligned with the larger areas of discolouration it is probable that it will be in the future. The likely point of entry is the rear of the junction box were the conductors exit/enter the panel. The soil in many remote areas of the Australian outback is red in colour. This has also been found on much younger panels and likely to be the result of the early stages of moisture ingress.
Figure 7. Discoloured wafers.
8 SUPPORT FRAMES.
All support frames were in very good condition with no sign of corrosion. The panels had been installed in accordance with the manufacturer’s recommendations to separate the anodised aluminium surfaces from dissimilar metals, typically hot dipped galvanised structures.
9 INSULATION PROPERTIES.
All panels were tested in accordance with IEC 61215 part 10.3 for insulation resistance from bridged connection cable to the aluminium support frame. The panels were dry when tested. The test instrument utilised was a HT Itialia Combi 419 with each module initially at 50V DC, 100 V DC, 250V DC, 500V DC and 1000V DC for 10 seconds. All of the LX81BGT panels passed this test even though four out of the six have shattered front glass. Testing the BPX47C panels revealed that all panels with damaged glass laminates, four out of the six, resulted in low insulation properties.
The tests were repeated with a 60 second testing time followed by a two minute test. All testing of the LX81BGT panels passed the longer prescribed test. Testing the BPX47C panels revealed that all panels with damaged glass laminates degraded even further, to a short circuit in one case and very low resistance in the remaining three. Refer to table 3 for the results of these tests.
The test was repeated for 120 seconds at each voltage range after exposure to prevailing rain showers. The final results are shown in Table 2.
Table 2 INSULATION TEST RESULTS FOR ALL PANELS.
Panel Number Insulation to frame
@ 500VDC for 120 seconds (M Ohms)
Insulation to frame @ 1000VDC for 120 seconds (M Ohms)
Each panel was placed at an angle of 60 degrees to the ground to capture the afternoon winter sun. The open circuit voltages and short circuit currents were taken for each module. These tests are typical of what a field service person would do with a standard multimeter to provide an indication of panel performance and integrity. Refer to Table 3.
Whilst these tests are subjective it does provide some indication of the panel’s ability to deliver power. These results were compared to the original manufacturer’s data. Table 2, summaries the results for each panel against this data.
The BPX47C panels with damaged glass and repair work were severely degraded and in one case could not support the short circuit test producing 0.0A DC, whilst the open circuit voltages ranged from 83% to 94% of OEM open circuit performance data, (with the exception of one at 64%). The short circuit currents varied greatly from 0 to 84% from the samples that appeared pristine.
Load profiling is required to determine the actual performance and will be covered in the next Silcar Energy Solutions paper – “Legacy PV Module Capacity”.
Panel integrity has not been of major concern for the traditional telecommunications Extra Low Voltage Direct Current (ELV DC) installations as they are operated at 24 and 48 volts DC. This allows you to ignore exposed terminals, cable damage, cracks in glass laminates and junction boxes deteriorating and falling apart. What was found in the sample
panels was that whilst operating at ELV DC levels, they still exposed the installation and the operator to potentially high fault currents at the terminals or tracks. Many switching regulators when switched on present the large station battery potentials at the array terminals and if not suitably insulated due to the failure modes identified above could result in short circuits and rupturing/tripping of protection devices if fitted or cable/module damage if not fitted.
If these failure modes emerge in the more modern panels used in higher voltage grid connect systems, people could be exposed to dangerous voltages. This also gives rise to concerns for tracking and the risk of fires on roof tops.
Careful inspection has found that on one of the sample panels LX81BGT, the silicon rubber backing sheet has three, full depth cracks in it which can eventually expose tracks of wafers to atmosphere. This has been evident on younger panels up to 15 years old, with severe cracks (shrink back) in the tedlar but not necessarily through the EVA at that time, see Figure 8.
Figure 8. Tedlar “Shrink Back” In Late Model Panels.
12 WHAT ARE THE FAILURE MODES?
Of the samples used for this demonstration 66% of the BPX47C panels failed to meet the warranted 80% of rated output at >20 years.
One LX81BGT panels had a faulty blocking diode which is field serviceable and restored output when replaced or bypassed. Only one panel (17%) was below the nominal 80% of the manufacturers rated output.
When segregating panels on the output performance based on regional environments in north of Western Australia i.e. high ambient temperatures, humidity and severe storm environments, 100% of the panels were below the manufacturers warranted 80% of rated output. The panels tested from the dry southern regions (low summer humidity) had a failure rate of 0%. Three additional panels were tested to confirm the initial findings. Although a small sample, these
observations are representative of panels that were used in this environment.
Electrical hardware used on the panels such as junction boxes failed on the BPX47CF modules due to UV exposure through the clear glass. The BPX47C utilised diecast aluminium junction boxes that were in perfect condition. The junction boxes on the LX81BGT were protected from UV exposure by the opaque silicon rubber backing sheets and remained in relatively good condition. These did not fail as there was only minor surface degradation and discolouration.
Figure 9. Badly deteriorated BPX47 C junction box.
Figure 10. 30 year old fully sealed junction box on BPX47 C panel
Both the LX81BGT and BPX47C front glass panes had failed due to impact from hail, debris or in some cases vandalism. Only the BPX47C utilising the glass laminate with resin between the layers failed or had significant reduction in output. The LX81BGT with the silicon rubber were shattered but did not actually fail although a reduced short circuit current was noted on two modules. The panel retained its general integrity and performance.
Figure 11 Melting of resin in the BPX 47C panels
13 HOW LONG WILL THEY WORK?
The panels have continued to work well past their 20 year warranty periods (up to 33 years) where the panel has not been physically damaged or severely affected by adverse weather conditions. When the glass is damaged it allows moisture and foreign matter to enter and corrode the wafers and tracks within the panels. If the bypass diodes do not fail, which allows hot spots to damage the resins and EVA’s and discolour the face of the wafers, then they will deliver power well past the manufacturers warranted output period.
The panels utilised in the 1970’s and 1980’s to support the remote telecommunications networks have continued to deliver power for up to 30 years, “The Good, The Bad and the Ugly”. If the glass laminated panels were damaged by impact then they begin to degrade quite rapidly and either fail or have a severely reduced output. The modules with the more resilient silicon rubber backing continue to deliver power for up to 30 years even with impact damage and shattering of the glass. In thes cases the output was reduced but did not actually fail.
Should these panels be included in an ‘End Of Life’ replacement program?
From my findings the panels that had the following characteristics should be included in an ‘End Of Life’ replacement plan;
• Damage to the glass structure; • Deteriorated junction box or cables; • Discoloured or disfigured wafers; • Corroded tracks; • Output below 80%; • Low insulation resistance to frame; and. • Cracking of tedlar and EVA backing sheets.
These types of failures can, in some cases, lead to excessive discharge of the station battery systems, burnt
backing sheets and increase chances of tracking to the earth/frame of the DC bus resulting is service disruptions.
Consideration should be given to retaining panels that displayed the following qualities irrespective of their age;
• >=80% of rated output; • Glass panels in good condition; and • High insulation resistance to frame.
The panels under test have a small foot print of approximately 0.455 square metres providing a low packaging density when compared to modern panels. This issue alone will invariably trigger end of life replacement to make way for larger higher density modules to support increases in telecommunication loads whilst retaining the same amount of real estate for the array structures. The original cost per watt for these legacy panels was as high as $100/W in 1974,  and rapidly declined to $9/W by 1984. In recent years, an increase in global panel production has dramatically reduced costs to around A$1.80/watt peak.
It is very clear from the samples under test that if the panels are not damaged they will continue to perform well past the 35 years of operation. As stated earlier in this paper, the output performance tests were made using measurement techniques used in the field being open circuit voltages and short circuit currents. This raises the question, is this a reasonable method to use? To confirm the actual performance characteristics of these panels a more scientific approach is needed which is covered by the subsequent Silcar Energy Solutions paper, “Legacy PV Module Capacity” . The actual performance results of the aged solar panels should then be utilised in the design aging factors as described in an earlier Silcar Energy Solutions paper “Hybrid Power System Model”, .
It is fair to say that modern PV systems will experience the same type of damage and subsequent degradation which will result in hazardous exposure to low voltage DC or tracking to earth of large scale systems.
 - Phillips data sheet “Terrestrial Solar Module” BPX47C/36, December 1979.
 - Intelec ’1984, “Solar Power For Telecommunications – The Last Decade”. Michael Mack, Telecom Australia Headquarters.
 - The Telecommunications Journal of Australia, Volume 38 Issue number 2, 1988, “Powering the Australian Optic Fibre Trunk Network”, G.C. Lee and N.K. Thaun.
 - Intelec 2012 “Legacy PV Module Capacity, How Far Can They Go”, Paul Murphy Silcar Energy Solutions.
 - Intelec 2012 “Hybrid Power System Model, How to get the most from your System”. Roberto Wust, Silcar Energy Solutions.