Solar energy may be one of the fastest-growing renewable energy sources today, but it has its limitations. Critically, unlike many other energy sources, solar energy cannot run 24/7. Solar farms are constrained by requiring hours of sunshine during the day. Photovoltaic (PV) cells remain constrained by their technology, i.e. the need for sunlight to create electricity. That reality alone requires any viable means to reduce operating and maintenance (O&M) costs.
Enter predictive maintenance (PdM). Predictive maintenance is able to assess operational data continuously and in real time by harnessing the exponential capabilities of artificial intelligence (AI). It is maintenance that predicts outcomes. It therefore ensures that potential malfunctions or failures can be predicted in advance. This minimises the risk of unforeseen or sudden breakdowns, thereby reducing downtimes and net O&M costs. Autonomous robots are already playing a key role in the uptake of predictive maintenance at solar facilities.
Maintenance Issues At Solar Farms
In the absence of robots, maintenance-related hazards and risks are significant and can escalate at solar farms. Labour is one high-risk factor at these facilities. Solar farms have traditionally used teams of workers to inspect or maintain solar facilities. But this is a very labour-intensive, time-consuming endeavour, especially as solar farms typically occupy several acres of land.
There is the further risk of sending workers off to solar farms at usually remote locations for inspection and maintenance work. Roads en route can be treacherous and there is the ever-present risk of vehicular accidents. Working conditions at solar farms can be further hampered by inclement or extreme weather conditions, such as very hot conditions, rain or snowfalls. Tasks undertaken at solar farms can be hazardous, with potential risks such as electrocution. Furthermore, as with any industrial facility, tasks can be dull and repetitive, potentially resulting in costly errors.
Beyond labour, there are a host of hazards and risks pertaining to assets at solar facilities. Solar panels can get dirty quickly due to dust, particulate matter and other debris. Panels can also become obstructed by vegetation, either fallen or overgrown. The multiple solar cells inside panels can short, and can also suffer from browning and discolouration.
Other equipment-related hazards that can result in downtimes include defective trackers, transformer leaks, damaged inverters, broken conduits and damage to combiner boxes, to name but a few. Perimeter breaches can also occur due to intrusions by wildlife or illegal entry by people. These breaches can result in serious damage to or theft of equipment, including panel components and cabling.
Predictive Maintenance As Driver For Improved Performance At Solar Farms
The good news for an industry that has in-built periods of inactivity is that the technology has become steadily cheaper, particularly in terms of component/asset and installation costs. This was confirmed by the National Renewable Energy Laboratory (NREL) in its report issued in 2021. Predictive maintenance can help the solar energy industry capitalise even more on these encouraging cost savings with allied O&M cost savings due to more efficient, smarter maintenance and inspection regimes.
Installation-related costs are indeed the overriding expense in the solar industry today. However, as the industry matures and the majority of facilities are built, so O&M costs will prevail. Most of these costs will be asset-related. For example, it has been estimated that almost 50% of O&M costs in the coming decade will be for repairs, such as replacing inverters. AI-enhanced data analytics will become hugely valuable as ageing solar infrastructure increasingly requires maintenance to cover parts. Consider that solar installations in the United States alone will be expected to spend $3.5 billion on O&M costs by 2030, with the Asia Pacific region expected to account for $5.7 billion.
Autonomous Robot Use Cases For Assets
Autonomous robots can be an integral part of enabling a maintenance regime that is data-driven and truly predictive. PV panels are a case in point. Consider that a Finnish case study goes as far as to say that, “The ultimate goal of predictive maintenance is to identify malfunctioning PV panels”. This rationale is that the efficacy of energy generation by a solar panel is directly proportional to the solar irradiance received by the panel and its ability to convert said energy. Malfunctioning panels clearly undermine the essential attributes (and limitations) of solar technology. Autonomous ground robots can undertake inspections and condition monitoring beneath panels and arrays in order to allay this risk. Specific examples of this panel-related condition monitoring include detecting defective panels with the use of infra-red, as well as inspecting connecting panel bolts.
There are other instances in which autonomous robots can be used on solar farm assets as part of a comprehensive predictive maintenance regime. For example, they can check for corrosion on the outside of enclosures and the racking system, as well as detect cables that may be tied, hanging or disconnected. They can also be deployed on an ad-hoc basis for analysing any damage that may have arisen due to unforeseen external factors such as weather, vegetation, electrical short-circuits or intrusions by animals or people, to name just a few possible scenarios. Importantly, autonomous robots can monitor all solar panels and tables/arrays frequently, instead of only inspecting or monitoring a sample range, which would be the norm with non-robotic, manual inspections.
Furthermore, ground robots can do all of these inspections and monitoring far more efficiently, at less cost and with far less labour-related inputs than any inspection by even the most expert inspection team. Drones can also be very helpful in aerial inspections and monitoring. For example, they can be used for overhead inspections of solar panels, with the benefit of being able to detect defects and faults far more quickly and efficiently than manual inspection.
Other Autonomous Robot Use Cases At Solar Farms
Other maintenance tasks not directly aimed at asset can also be undertaken by autonomous robots. These tasks by ground robots can include checking for signs of animal infestation under the array. The monitoring for ground erosion near the footings of solar arrays is especially important, since ground erosion can destabilise solar arrays and mounts, thereby putting panels at risk.
Vegetation maintenance is another important maintenance factor at any terrestrial solar plant, and there is already a number of grass-cutting and vegetation-monitoring robots on the market. Panel-cleaning robots can be invaluable in their ability to autonomously undertake probably the most labour-intensive and tedious task in any solar farm. An exciting aspect of all this technology is that a combination of autonomous robots and drones can facilitate an ecosystem that fosters predictive maintenance at a solar installation.
The technology is speeding ahead. Today, asset owners are embracing a site-centric model of autonomous inspection which enables the quick and effortless creation of digital twins, whereby an exact digital replica or ‘twin’ of the physical installation can be used for maintenance insights. Assets, devices, processes and systems at the solar farm can be analysed and even monitored remotely using the digital twin, particularly by testing for potential faults via data anomalies.
Solar energy has steadily become cheaper per kilowatt hour, yet its technology-related limitations persist. Predictive maintenance through autonomous inspection is an excellent way of bridging the ‘power curve’ caused by downtimes. Not only does it result in cost savings for solar farms, but will surely continue to transform the industry by increasing the safety and efficiency of operations.
