Challenges in Solar Farm Operations and Maintenance

Photovoltaics (PV), also called solar cells, are electronic devices that convert sunlight directly into electricity. Accumulation of dust, bird droppings, atmospheric particular matter, dry leaves, that gets deposited on the solar panel reduces the power generation capacity of the panel. This is a major problem in solar farms. Firstly the solar panels do not convert 100% of the light source into power source, due to the inherent nature of the solar cell technology. Add to this the soiling of the panels will further reduce the solar power generation capacity. As per statistics and research done in middle east, the output power generation from solar panels decreased as much as 50% when not cleaned for 6 months.

The Cleaning Imperative

In a typical commercial solar PV panel there are about 72 solar cells. The total dimension of this panel is 78in X 39 in.(3.25ft X 6.42 ft).  All the PV panels are connected in series in a farm. If one module fails, it can effect the entire string. Typically each such solar panel generates about 320 watts (discounting hours of irradiation, panel efficiency).  Each such solar panel needs 20.8 sq.ft. area. To generate 1 Megawatt of solar power, 3,125 solar panels are required. The typical land area required for 1 Mw solar farm is about 4 acres. For a 1 Gigawatt solar farm ~3 million solar panels are required. Compare this with the hydrocarbon based fuel plant, roughly 10% of the land area of a solar farm is needed.

In a typical dry and arid zone like India where solar farms are growing exponentially, these panels need to be cleaned once a week. Results indicate that solar energy production can be reduced by ∼17–25% if not cleaned.

The Cleaning Challenge

The current method of cleaning adopts laborers where they use microfiber cloths fixed at the end of an extendable stick or brushes. Adoption of mechanical motorized brushes or microfiber is used as an alternative, to clean the dust from the panel.  This is a huge challenge currently for the solar farms.

  • Use of Water – As per the council of energy, environment and water estimates that about 3-4 litres are needed per panel in a normal land area and 7-8 litres in arid zone land area. At this rate, 24,000 litres of water would be required to clean a 1 Mw solar farm. The temperature of the water should be around the same as the solar panel, else it results in cracking of glass due to thermal shock.
  • Use of Chemicals or Detergents – Use of water is not a sustainable practice. If chemicals and detergents are used, ground contamination is a risk. Hard water (borewell water) causes additional issue of glass residue deposits, which is another blocker of sunlight. Softening water plants have to be installed at the solar farms which is additional cost.
  • Risk of surface damage – use of hard brush, or un-scientific techniques by laborers while cleaning the solar panel often causes scratches that reflects the solar irradiation and also causes damage to the surface of the solar panel.
  • Highly Labor intensive – Solar farms have to employ a huge workforce for manually cleaning the panels. The work is also very menial in nature, although it might generate employment for low skilled people.
  • Hazard of electrical shock – Since manual labor is employed who are low skilled workers, there is a risk of getting exposed to electric shock, especially if done during the peak of the day and power production.


Keeping the efficiency of the solar panels high is a business challenge for the solar farm companies. To be profitable they have to produce to the highest efficiency. Things that are not under control can be weather (cloudy). However de-soiling the panels is an imperative. The following solution techniques can be adopted by the solar farms

1. AI/ML based intelligent solution for identifying cleaning panels

A predictive technique should be adopted for cleaning the panels instead of periodic. This can optimize the work and the workforce. The following are the options available.

  • Analyzing Performance Ratio (PR)  – The solar photovoltaic (PV) plant performance is measured in PR as defined in IEC 61724. This unit of measurement PR captures the inefficiency and the losses including due to soiling and fouling of solar PV panels. A reference solar PV module is deployed at the actual solar farm site. This panel data is used to model the impact of soiling on the PR and the PV performance. The soiling rate, washing cycle detection, etc., is modelled using AI/ML from the data available from the reference panel. Using this method, the algorithm developed from the reference solar module, it is used on the actual solar panels to determine the PR degradation and the soiling affect. This is then used to determine and localize the solar PV that needs cleaning.
  • Using UAV drone – Drones to capture image of the solar panel. Using computer vision and image analytics, AI/ML, the severity of soiling can be determined. Based on the exact soiled solar panel and its location a automated work order is generated for cleaning the soiled solar panel.

2. Sustainable options for cleaning the panels

Alternative to water based cleaning approach can be explored. Some of these technologies still need maturity and economics. The options that can be explored are dry cleaning methods

  • Surface Vibration Technique – Use of piezo devices to create vibration  or ultrasonic sound waves for loosing the dust and then vacuuming the same.
  • Pressured Air Flow – A pressured air flow over the panel using compressed air can be used to loosen the dust with vacuuming.
  • Electrostatic dust shield – The panels can be coated with a thin film of hydrophobic solution that converts into nanofilm. Coatings like Titanium Dioxide, that reacts with UV light and the dust is dispersed. Which can be used to optimized

3. Adoption of AGV and Robotics to automate

There are many companies that have devised rover type robots that can crawl on the solar panels autonomously and clean the panels. If these rovers can be delivered to the panels using Autonomous trucks with all the cleaning paraphernalia(water, compressed gas, rover, vacuum, etc) which can be guided to the panel that needs cleaning (based on the efficiency). The autonomous truck if has LiDAR it can be driven in the dark using the LiDAR enabled path planning and routing. LiDAR technology works i low light conditions, allowing cleaning to be done either early dawn or dusk time. The truck can have a waste water tank, a waste bin for collecting the dust.

The Autonomous truck with a robotic arm can place the rover on the panel. This can be remotely monitored or operated. In order to achieve this a private 5G network can be deployed to get a high bandwidth video connection from the autonomous truck.


Pavagada the 3rd largest Solar farm in the world located in Tumkur,  Karnataka, India has a capacity of 2050 Megawatt. Approximately having 6 million (6,406,250) solar panels, is spread over 13,000 acres of land. Even if 100ml of water is used per panel per month, a total of 6 lakh liter of water is needed per month. That is roughly 0.04% of daily water supply of Bangalore, having ~12 million population.

The operations and maintenance of solar farms will not be economical in the long term if technology is not adopted. The production cost will be high. While some of the technologies still need to mature, there is a scope of exploring Robotics, dry-cleaning and optimizing the cleaning effort as explored in this PoV.


Author Details

Krishnananda R. Shenoy

Krishnananda R Shenoy is the Chief Architect of Internet of Things (IoT) practice, Engineering Services at Infosys. Responsible for solving customer IoT requirements and solve the problems by proposing the right technology solution. This includes Edge, Cloud and Hybrid architecture. Industry 4.0 Consulting and Advisory for helping customers transform towards digital manufacturing. Implement emerging technologies like AR/VR, Digital Twin, 5G and Industry Cyber Security. Responsible for setting up Go-To-Market with product partners, hyperscalers and startups.


Leave a Comment

Your email address will not be published.