The solar energy sector in India is moving towards automation and digitalisation, with special focus on the operations and maintenance (O&M) segment. In the initial years, O&M activities were limited to site management involving module cleaning and grass cutting. However, later, optimal use of the balance of system and delivery of the promised generation gained importance. This is known as predictive maintenance. Earlier it was done manually, but with time O&M players have started investing in the automation of predictive maintenance services.
There is a strong business case for the automation of O&M services. One, automation helps in reducing manpower costs. The low tariffs discovered in the recent solar auctions have squeezed the margins of O&M players as well. Automation will help them maintain profitability.
Two, O&M players can deploy technology to predict the future of the components as well as the plant performance in order to avoid penalties. To integrate renewable energy with the grid, various state governments have come up with strict scheduling, forecasting and deviation settlement regulations that penalise over- and under-generation of electricity from renewable energy plants. These two trends have encouraged O&M players to invest in the automation of predictive maintenance services.
Automation and digitalisation can assist in predictive maintenance on the field as well as in offices. While the use of drones, robots and smart sensors is decentralised, remote monitoring systems are set up at a centralised location. These systems monitor the solar plant from a remote location in real time and help in taking appropriate and quick actions in case of any uncertain incident. Remote monitoring also helps identify underperforming components, operate automatic ticketing systems and detect soiling losses.
Inspire Clean Energy is a Mumbai-based O&M service company that has developed a software platform called EIRA. The model of EIRA is based on the three-second and three-click concept, wherein a client is made aware about the status of the solar plants in three clicks. EIRA logs into every project at 15-minute intervals to check the performance parameters of each inverter, energy meter and transformer. Once a flaw is detected, it automatically generates a ticket and sends it to the nearest engineer. As soon as the problem is solved, a report is sent to the client. The platform also has a dashboard with smart tiles. The tiles stay green in normal conditions and turn red if the corresponding inverter is down. It takes just three seconds for an investor to get this status and three clicks for the operator to procure more information about the generated ticket.
There are a number of such examples where advanced technologies are deployed to improve predictive maintenance. These technologies have matured from computer-based automated monitoring systems to drones, robots and wearables, and now finally to artificial intelligence (AI) and internet of things (IoT).
Automated monitoring and big data analytics
Real-time monitoring of equipment performance and measurement of energy generation is necessary to ensure that the plant is performing at an optimum level. Further, plant yield needs to be remotely monitored in order to identify faults without hindering plant operations. Thus, supervisory control and data acquisition (SCADA) systems have become an inherent part of all solar plants. Smaller distributed plants are monitored remotely and not by a dedicated O&M team. They typically depend on web-based monitoring systems that are more economical than the large expensive SCADA systems. The main difference between SCADA and web-based systems is that the former provides remote control functions, in addition to the monitoring and data collection functionalities.
Robotics, drones and wearables
Automation technologies are being used in the solar industry in the form of drones, robots and wearables in order to reduce manpower costs. Drones are excellent for site assessment and O&M, providing greater detail than ground crews. According to industry experts, drones can inspect all the modules in a 2 MW plant in about 15 minutes, while the same activity will require more than three hours if carried out manually. Moreover, thermal imaging cameras on drones can detect malfunctioning modules, specifically hotspots, which reduce electricity generation. They can point out faulty modules or strings with great reliability and can save up to 30,000 hours of hazardous work each year. Meanwhile, crawling robots can get quite close to a structure’s surface, and use microwave, and ultrasonic transmitters and receivers to penetrate equipment structures and reveal faults in materials.
Apart from drones and robots, IoT-enabled wearables such as watches, headphones and armbands are being used by many O&M players for remote monitoring of solar plants. While their use is currently limited to very small rooftop plants, their scope in large utility-scale plants is immense.
Growing role of AI
Various programs are being developed in the AI and machine learning domain to enable monitoring and operating machines to self-learn, evolve and become self-reliant to an extent that they can take care of any present and future problems arising in a solar power plant.
Energy storage coupled with AI is being explored to maintain the grid frequency and avoid outages. Maintaining the frequency of solar power at the accepted grid range becomes difficult due to the intermittent nature of solar. AI can help in predicting weather patterns by analysing large real-time weather data sets from multiple sources like satellites, weather stations and other devices and comparing it with historical weather data. It can also predict how this weather would impact solar production, thus allowing power producers and operators to adjust and schedule power accordingly.
Further, AI can be applied to improve solar project performance as it can collect data from multiple solar assets and analyse these data sets with respect to factors such as region, system, radiation, equipment specifications and weather. AI can spot inconsistencies and diagnose faults in a much shorter time frame and more accurately than humans.
While solar power automation systems have various benefits, they also involve significant costs. The cost largely depends on the level of automation. The use of robotics, drones and other automated tools for O&M is more expensive than asset management tools and machine learning applications. However, if applied on a certain minimum scale and across regions, it may prove to be cost effective in terms of total efficiency gains.
In fact, investment in automation helps devise a more focused O&M strategy. In addition, it reduces replacement costs by extending component life. However, it takes time to develop human resources with the right expertise.
A key positive trend is that industry players are now more aware about the scope of automation and digitalisation. In the O&M segment, automation to facilitate predictive maintenance has been a priority area. This has already shown positive results with respect to decreasing manpower costs. As per industry estimates, the technical manpower requirement in solar plants is 5 MW per person and should reach 10 MW per person by 2019. “As the O&M segment becomes more technology-driven, the composition of the total cost is likely to shift from personnel to digital in the years to come. As per estimates, the share of personnel in the total cost will come down to 17 per cent and the expenditure on digital equipment will comprise more than 50 per cent of the total cost by 2028,” says Puneet Jaggi, director, Solarig Gensol. It is believed that if the current pace of automation continues, the majority of the O&M activities will be carried out through AI-enabled field assistants instead of manpower.
With positive results on the ground, O&M players are keen to incorporate futuristic technologies such as AI-based digital twins. These AI applications can reduce the time and effort required for planning and analysis. A replica of an actual physical asset or a digital twin will be created to be used as a benchmark for identifying faults through data anomalies. However, in the short term, the high cost associated with these futuristic technologies may pose a challenge, particularly for smaller players working on thin margins.
By Sarthak Takyar