Troubleshooting efforts in PV plants: follow up Q&A from the recent pv magazine webinar

Soiling, insulation errors, broken bypass diodes and failed module strings are only some of possible causes for yield loss that operators of solar plants have to take into account. pv magazine has reported on cases where causes were researched and faults were detected using today’s standard  technology. Ingmar Kruse, CEO of SunSniffer – who partnered the pv magazine webinar on November 20 – is developing new monitoring technology that analyzes module-resolved data for a more automized O&M service. Kruse, alongside John Davies, Operations Director – Solar & Energy Storage of Cobalt Energy Limited, and analyst Götz Fischbeck of Smart Solar Consulting, discussed troubleshooting four case scenarios presented in the webinar.

Davies pointed out why IV-measurements often are not very effective at troubleshooting and, as such, are usually not something investors should ask fort. In the second half of the webinar, Kruse and Fischbeck shared their views on module resolved-monitoring technology.

What follows are the most insightful and pertinent questions and answers from the discussion.

(Click here to listen to the webinar and view the slideshow presentations discussed).

In two separate cases we have seen costs for thermography range between €500 and €1,000/MW for a multi-MW-plant. What is your experience?
John Davies (JD): Certainly in the U.K., costs range from £250/MW to £450/MW for the aerial thermography service on a site-wide basis. However, the price and the selection of the contractor varies significantly in terms of quality of service, and quality of data output and presentation. As is normally the case, you get what you pay for.

When you carry out thermography, how do you quantify economic losses? What are the alternative and the latest techniques for detection of (partial) failures?
JD: As an O&M service provider our contacts have annual aerial thermography as part of the scope. Before the high irradiation months of the summer, we generally plan to carry out the first spring season grounds maintenance in late spring, followed by a module clean, and then carry out the IR inspection. This way you get the best results from you survey.

IR is a great tool, which enables you to assess entire sites at a reasonable cost. However, the IR method is limited in its effectiveness of providing conclusive detail on module defects or deterioration phenomena. IR will find ‘low hanging fruit’, such as strings out, modules out, activated diodes, and then pretty much everything else is an interpretation of hot-spots. In some cases aerial IR surveys have diagnosed PID on sites, where the owners then requested our services to carry out Electroluminescence (EL) testing to verify the PID. Unfortunately, the patterning identified was simply the natural variance in the cells on these particular modules, the delta in temperatures was only 1^oC, which is simply too small to be anything of concern. The EL images, carried out at high and low currents, categorically proved that there was no PID in these modules. So our advice is: don’t rely solely on thermal IR when it comes to warranty-based claims or situations where large levels of money are concerned. Invest in high quality EL/flash testing, and you will at least be certain what the issues are.

Is thermography the best way to spot bad bypass diodes, as seen applied in the first troubleshooting case presented in the webinar?
JD: In Cobalt Energy’s opinion, yes. If you are searching for activated, or faulty bypass diodes on a site that does not have any module level monitoring, then this is the cheapest and quickest method. Certainly a lot cheaper and quicker than string testing your entire site. However, it would rarely justify the cost of the drone survey just to send it up and carry out a full site survey, just for the identification of diode issues. Drone IR enables you to see strings out, modules out and indicate where you have issues causing hot-spots. So collectively in many cases it is justified. Of course, if you had a module-based digitalization system like Sunsniffer, you wouldn’t need drone IR or to string test.

In the presented case study, failed bypass diodes were localized using thermography: If one would have used Sunsniffer instead, how could you know that a voltage drop measured at a module output was due to a damaged diode and not due to hot spot?
Ingmar Kruse (IK): We recognize diode failures with a certainty of 99.5%. Hot spots are symptoms to a variety of causes. Defective diodes themselves can cause hot spots on the module. But defective diodes – like other failure causes – show distinctive characteristics. Diode failures, for example, result in a sudden drop of voltage, which is around 33% for each burnt diode.

You calculated that without Sunsniffer’s interception the faulty bypass diodes would cause damage of €252,000 Euro. This would be the case, if the failure had existed for six years. So why do you assume that the failures could have existed for such a long time? It showed up in unclear results in the string monitoring, and 4% reduced performance ratio, which everybody should have noticed.
IK: A reduction of 4% in the performance ratio will occur over a period of time and not instantaneously. The existing monitoring system obviously was not able to detect failures in the beginning. Only when losses reach a certain threshold will they be noticed, as was the case here. But until this point where the need for action has been acknowledged, losses were highly accumulated. With SunSniffer, each defective diode would have been detected the moment it failed. Action could have been taken instantly – or bundled – and losses would have been minimized. In addition: without constant module monitoring and analysis it is more than likely that the same problem continues to occur over the next six years, causing gradual but increasing losses until the next IR inspection.

Which reduction in performance ratio normally triggers a troubleshooting action? What is the uncertainty of the performance ratio determination?
JD: Of course, the PR figure as a key performance indicator is something that we keep an eye on, but this is typically calculated on a monthly basis in retrospect. Troubleshooting actions are generally triggered by monitoring system alerts and alarms, which can be set specifically on a site-by-site basis. If a string wasn’t performing correctly, if an inverter wasn’t performing correctly, or if we encountered an area of the plant out of operation, then this would trigger an intervention.

For us as an O&M, PR is purely an indicator of how a plant has been effectively designed, what irradiance levels there have been that particular year, and how much ‘uptime’ we have been able to provide. The uptime is the only thing we can control as an O&M service provider.

How effective will the performance ratio be in judging plant output? Is there any new and better parameter available?
IK: Performance ratio is a good indication of a plant’s performance. But there are lots of variables that should be taken into consideration in order for it to be reliable. SunSniffer has a different approach: we have the quality data of each module – this means that we can determine the quality of a module plus all string and inverter data. The result is a very precise judgement of the performance of the plant.

JD: Following the SolarPower Europe O&M Best Practice Guidelines, O&M contracts are turning away from PR as standard in modern O&M contracts. The emphasis is now turning to availability (typically 99%), and response & rectification times. Both of these KPIs require a strong work management system and skilled and organized resources in close proximities to your sites. An interesting logistical challenge that continuously changes as you onboard new sites in different locations.

John Davies presented in the webinar how many owners request of their O&M service providers string IV curve measurements, and an estimation of costs. How often do you find issues with these measurements that trigger action?
JD: IV Curve measurement is an indicative method that provides you with clues that something isn’t right, but it is not intelligent enough to tell you which module is faulty and what’s wrong with it from a single string measurement. The tool needs to be used by a skilled and experienced solar technician who knows how to interpret the curves and data output, to point in the direction of chasing that defect or fault. As well as this complexity, it is also irradiance dependent, so you have a limited window of opportunity to carry out the tests, especially with the U.K. weather. We estimate that we find irregularities in 10-20% of IV curves that we take. However, these could vary in production impact.

How often plant owners request these IV curve measurements and why?
JD: Typically, in the majority of our O&M contracts IV curve testing is part of the standard O&M scope of works. The quantity varies from 10% of a plant annually, to 100% of a plant annually. The reason that IV curves are required is that technical advisors have probably recommend that IV curve tracing is best industry practice and they need to be in the scope to compare and contrast string performance. The reality that we have experienced, however, is that the huge horde of paperwork that this produces simply does not get looked at by most owners. Our advice would be to be more specific in the scope as to what owners are trying to achieve. If an O&M was tasked with plotting the performance of the strings on a site, in a type of priority list, then this would be a useful tool for both owner and O&M service provider. Whereas at the moment, I believe that it’s being done as it is in the contract.

What is the current state-of-the-art process in remote monitoring of PV farms?
IK: SunSniffer relies on the digitalization of plants and on artificial intelligence to analyze performances of your fleet and make sure they produce what they should. With setting power loss-thresholds you can bundle modules that need to be exchanged and then can set up regular exchange-services in order to reduce costs and ensure plant performance on the highest level. Or let the system determine automatically the right timing for exchanges to keep costs low. In addition, the quality assessment of each module of SunSniffer leads to a precise plant assessment at the push of a button. So far, monitoring has concerned the power collection point. But SunSniffer starts with the power production point: by this we know exactly where failures occur and where not. The module is the essential part of a plant, and we analyze it.

What monitoring hardware technology is used by SunSniffer, and must it be installed at the time of plant installation or can it be retrofitted?
IK: The SunSniffer system consists of sensors that are installed in junction boxes for greenfields and retrofitted for brownfields: string readers for each string, and gateways to gather data from the field. Either greenfield or brownfield: using SunSniffer, plant owners and operators can ask themselves how much autonomous plant driving they wish/need. It can start at the inverter level with the extreme precise PR calculations, or string monitoring with the dual functionality as “power-meter” or measuring device for the individual modules. Both are possible with SunSniffer, so plant owners can start with highly precise string monitoring and later on retrofit the whole, or parts, of the plant to have module measurements.

Sunsniffer allows module level monitoring. Can it be implemented in MW-scale plants using central inverters?
IK: Yes SunSniffer is a cost-effective and feasible solution for all plant sizes. As for central inverters, we will introduce in 2018 a solution that deals with the high noise levels generated from most central inverters.

Can you give an example of a solar PV power company that has embraced the new Sunsniffer technology?
IK: Several module manufacturers have installed SunSniffer in their modules. These include companies like AEG, CSun, Chinaland, Suntech, Linuo, and others.

Can Sunsniffer distinguish the causes of reduced performance, such as shadowing, hot spot, improper earthing, damaged bypass diodes or PID? If so, how?
IK: SunSniffer can distinguish between losses due to faults or due to shading using our Artificial Intelligence, which looks for specific patterns for each type of loss.

Will equipment providers (module manufacturers, inverter manufacturers and others) accept claims based on SunSniffer findings? Or are further tests necessary anyway, after module faults have been detected by Sunsniffer?
IK: We do not see a reason why a manufacturer will not act upon a proper claim from its customers. Modules are being measured precisely with SunSniffer. Manufacturers are worried about false claims mainly by misinterpreted IR pictures mixing up hot spots with power losses, but they have no problems with the reliable measurement of SunSniffer.

Is the SunSniffer system listed for use in the U.S.?  Do you have any installations in North America?
IK: SunSniffer has a distributor in the United States and yes, the system is sold in the country.

What are the CAPEX costs per MWp for the system? Does it cover material and installation costs?
IK: In general, SunSniffer starts from as low as 1.2 dollar cents per Wp depending on the modules used and sizes of plants.

What is the OPEX cost of the Sunsniffer, including any issues related to communications of the many data you collect?
IK: This varies depending on the plant size. Annual costs range from $0.6 to $2.5 per kWp. Complete fleets will get additional discounts.

How robust is the protection of the system against lightning?
IK: We have not seen unusual problems with lightning. The ports are well protected and our first installations are now almost eight years old and are all still working fine.

If you calculate a Sunsniffer monitoring for a smaller, up-to-date-plant built in 2017, how do savings compare, give that PID and other failures are not as likely nowadays as they were for modules installed five years ago?
IK: We provide an excel sheet where each case can be calculated. We have not seen one case, where the ROI of the invest in SunSniffer has not beaten the ROI of the plant itself.

Are there any failures that can be found with thermal imaging diagnosis that cannot be seen using SunSniffer?
IK: Thermal imaging relies on detection of temperature differences and is a single snapshot performed once under certain weather conditions. SunSniffer relies on accurate electrical measurements done every 30 seconds on each module. So SunSniffer sees deviation from under 1% and thermal imaging is, according to the ZAE, limited to power losses of 4% and higher.

How is Sunsniffer’s solution different from monitoring using optimizers from SolarEdge or Tigo Energy?
IK: Optimization solutions influence the natural performance of the module and therefore analyzing errors will be difficult as module faults are concealed. SunSniffer focuses on obtaining raw data from the modules. Analyzing them with artificial intelligence will detect module faults.