Drone Solar Plant Inspection Services in India

Geocartis inspects solar plants across India using aerial thermography and high resolution visual capture. We find failed modules, hotspots, bypass diode failures, dead strings and inverter-level outages, and we deliver them mapped to your own table, string and inverter identifiers so your O&M team can act on the report rather than interpret it. Inspections are carried out to the conditions set out in IEC TS 62446-3. We also survey plants during construction and after storm damage. We are based in Ahmedabad and work nationally.

We are a survey and inspection services company. We do not build drones, we do not sell monitoring software, and we do not carry out electrical work.

The scale problem

A 100 MW plant built with 550 watt modules carries roughly 180,000 modules.

Inspecting those with a handheld thermal camera, at the rate a technician can walk a row and image each module properly, is not a job that finishes. In practice it does not get done. What gets done instead is inspection of the areas somebody already suspects, based on inverter-level or string-level monitoring data, which by definition finds only the faults large enough to show up in the SCADA.

Aerial thermography inverts that. Every module in the plant is imaged, in a few days rather than a few months, at a cost per module that makes complete coverage the normal case rather than the exception.

That is the whole argument for this service. Not that a drone sees something a handheld camera cannot, because it does not. That a drone sees all of it.

The money argument

Worth working through, because solar O&M budgets are tight and the case has to stand up.

Take a single dead string of 25 modules at 550 watts, so 13.75 kW of capacity producing nothing. At a specific yield of 1,600 kWh per kWp per year, which is a conservative figure for Gujarat and much of western India, that string would have generated around 22,000 kWh in a year. At a tariff of ₹2.50 per kWh, that is about ₹55,000 of lost revenue from one string, for one year.

A plant with fifty such faults undetected across a year is losing something in the region of ₹27 lakh. Those numbers are illustrative and your yield and tariff will differ, but the shape of the answer does not change: on any plant of scale, the inspection costs a fraction of what the undetected faults cost.

String-level monitoring catches some of this. It does not catch a single failed module inside a healthy string, it does not catch a bypass diode failure taking out one-third of a module, and it does not catch gradual degradation. Those show up thermally long before they show up in a generation report.

The second argument is fire risk. A hotspot is a thermal event on a live electrical asset, and module-level thermal inspection is explicitly framed in the standard as supporting fire protection, not just yield.

IEC TS 62446-3, and what a compliant inspection means

This is the reference document for the service, and it is worth understanding what it does and does not do.

IEC TS 62446-3:2017 is part three of the IEC 62446 series on testing, documentation and maintenance of grid-connected PV systems. Part three covers outdoor infrared thermography of PV modules and balance of system components on operating plants. It sets requirements for the measurement equipment, the ambient conditions, the inspection procedure, the evaluation of images, the contents of the report, and the qualification of the personnel carrying it out, and it provides a matrix of thermal abnormalities to guide diagnosis.

It is a technical specification rather than a full international standard, which matters for how it should be described. It sets out an agreed methodology and reporting framework, and industry practice has built a common interpretation on top of it. It is routinely referenced in EPC contracts, technical due diligence scopes and module warranty claim procedures, which is why an inspection that does not follow it is of limited use when you need to make a claim.

The condition requirements are the part clients most often do not know about, and they constrain when we can fly.

Irradiance in the plane of the module must be at least 600 watts per square metre. Below that, the temperature differences that reveal defects are too small to be reliable.

Cloud cover must be low and stable, commonly taken as no more than two oktas. Broken cloud is worse than full cloud, because passing shadow changes module temperature between one frame and the next and produces both false positives and missed defects.

Wind must be within limits, because wind cools modules and suppresses the temperature difference you are looking for.

The camera must view the module close to perpendicular, because reflected sky radiation at oblique angles corrupts the reading.

Modules should be reasonably clean, since soiling produces thermal patterns of its own.

What this means practically in India. The flying window is the middle of the day in clear conditions, which in much of the country means October to May, and it means a monsoon-season inspection is often not possible to the standard at all. Anybody offering you a compliant thermal inspection in July in Gujarat should be asked how.

It also means an inspection report worth having records the irradiance, ambient temperature, wind and cloud conditions at the time of capture. Ours does. A thermal report without those numbers cannot be assessed by anybody, and it will not support a warranty claim.

What aerial thermography finds

The standard’s abnormality matrix underpins the classification. In practice the recurring findings on Indian plants are these.

Whole plant and inverter level. An inverter or a combiner offline, showing as a large uniformly cool block of modules. Usually already visible in monitoring, and the aerial survey confirms extent and location precisely.

String level. An entire string cool relative to its neighbours, indicating an open circuit, a blown fuse, a disconnected or damaged cable, or a connector failure. This is the highest-value single finding, because a dead string is significant lost generation and is frequently invisible in plants monitored only at inverter level.

Module level. A whole module warmer than its neighbours, indicating a short circuit, an open circuit within the string configuration, or a serious internal fault.

Substring level. Roughly one-third of a module hot, which is the classic bypass diode signature. Common, easily missed on the ground, and it costs a third of the module’s output.

Cell level. Individual hot cells, from cracks, cell mismatch, solder failure or localised shading. Single hot cells are also the most direct fire risk finding.

Patterned degradation. Potential induced degradation typically shows a characteristic pattern concentrated at one end of a string, and it matters because it indicates a systemic problem rather than a one-off failure.

Physical and environmental. Soiling, bird droppings, vegetation shading, structural shading from adjacent rows or equipment, cracked or broken glass, delamination, and hail impact damage.

Balance of system. Where in scope, thermal anomalies at combiner boxes, cable terminations and junction boxes.

Alongside the thermal, high resolution visual capture records physical condition: broken glass, frame damage, module misalignment, tracker position anomalies, structural corrosion, cable management failures, and vegetation growth.

What thermal will not tell you

Being clear about this matters, because a thermal survey is a screening tool and clients sometimes expect it to be a diagnosis.

Thermal shows the symptom, not always the cause. A module running hot has several possible explanations and the abnormality class narrows them rather than resolving them. Confirmation of the cause needs ground work: I-V curve tracing at string level, insulation resistance testing, visual inspection at close range, or electroluminescence imaging.

Microcracks are largely invisible thermally until they progress far enough to cause a measurable temperature difference. Electroluminescence imaging is the technique that finds them, it requires current injection and darkness, and it is a specialist service that we do not provide.

Degradation rate is not measurable from a single thermal survey. Comparing this year’s survey against last year’s, on the same plant with the same asset mapping, does show progression, which is one of the strongest arguments for buying this annually rather than once.

Absolute temperature is not the point. Thermal cameras of this class read temperature differences reliably and absolute values less so, and the standard’s approach is built around differences and patterns rather than absolute readings. A report that presents absolute module temperatures to a decimal place is presenting precision it does not have.

The deliverable problem: asset identification

This is where most solar drone inspections fail to be useful, and it is the part we would ask you to judge us on.

A thermal orthomosaic of a 100 MW plant with red markers on it looks impressive and is close to worthless to an O&M team. The technician who has to fix the fault needs to know which module, in which table, on which string, under which combiner, under which inverter, in your numbering, not ours.

Producing that requires the plant’s own asset structure: the layout drawing with table and row numbering, the string mapping showing which modules sit on which string, and the combiner and inverter allocation. With those, every anomaly we find carries a full address and the report drops straight into a work order.

Without them, the best anybody can deliver is a coordinate and a picture, and somebody at the plant then spends days working out what that corresponds to.

So the first thing we ask for on any solar enquiry is the as-built layout and string mapping. On plants where those documents are missing or out of date, which is more common than it should be, we can build a positional table and module index from the survey itself and reconcile it with whatever records exist. That is additional work and it is quoted separately, and it is worth doing once because every subsequent inspection then has an asset framework to report into.

Beyond thermal: other solar work

Pre-construction survey. Terrain model, contours, drainage and access for a proposed site, feeding grading design, row layout and pile design. Ground slope drives inter-row spacing and shading, so the terrain data is not incidental.

Construction monitoring. Pile position and verticality survey, table and tracker installation progress, module installation counts by block, grading and earthwork quantities, access road and drainage progress, and inverter station and substation civil works. On a large solar EPC contract, module counting from the air is a fast and unambiguous progress measure. Our construction survey page covers the method.

Commissioning support. A full thermal baseline at handover, before the plant enters its warranty period, establishing the condition of every module on day one. This is the single most valuable inspection in a plant’s life and it is regularly skipped. Without it, a claim two years later about modules that were faulty from delivery has no baseline to point at.

Storm and hail damage assessment. Rapid post-event survey with damage extent mapped by block and table, module counts affected, and a dated georeferenced record. For an insurance claim on a large plant, this is the difference between a negotiated estimate and a measured one.

Vegetation and site condition. Growth encroaching on rows, drainage failure, erosion and washout after heavy rain, and boundary condition.

Rooftop and C&I plants. Smaller, and the constraints are different: urban airspace, surrounding buildings, and often a shorter flying window because of surrounding shading. Thermal inspection works the same way, and the roof itself can be surveyed for membrane condition at the same time.

Floating solar. Thermal and visual inspection over water, with the additional constraint that a lost aircraft is generally not recoverable, so flight planning is conservative.

How we do it

Step 1: Plant data before the flight

We ask for the as-built layout, the table and row numbering scheme, the string mapping, the combiner and inverter allocation, module make and rating, the plant’s own asset naming convention, and any previous inspection report.

We also ask for recent monitoring data on known underperforming inverters or strings, not to bias the survey, but because a thermal finding that corroborates a monitoring anomaly is more actionable than either alone.

Step 2: Scheduling to the conditions

The condition requirements set out above decide when we fly, and they are non-negotiable if the report is to mean anything.

We schedule for clear conditions in the middle of the day, with irradiance above the threshold, and we build weather contingency into the programme rather than committing to a date and then flying in the wrong conditions to meet it. On a large plant that may mean the survey runs across several days, and we record conditions per block so the report reflects when each part was captured.

We also ask whether the plant is due for cleaning. Surveying a heavily soiled plant produces thermal patterns from soiling that mask module defects, and the sensible sequence is to inspect after a cleaning cycle.

Plants with trackers need the tracker position agreed. Modules should be at a consistent orientation across the survey, and a tracker moving during the flight changes the viewing geometry.

Step 3: Airspace and site permissions

We check the plant against the Digital Sky airspace map before quoting. Green zone to 120 metres above ground level without prior flight permission, yellow zone with permission and a lower ceiling, red zone not at all. Large solar parks in Rajasthan, Gujarat and Ladakh can sit in border or restricted airspace, so this is checked rather than assumed.

Plant permission, safety induction and the site’s own rules about flying over live electrical equipment are arranged with the O&M team. Our aircraft carry active Unique Identification Numbers and our pilots hold DGCA Remote Pilot Certificates.

Step 4: Flight planning

The thermal sensor’s resolution and the required detail level set the flying height, and it is a direct trade against area covered per day.

A scanning-level survey, sufficient to find string outages, dead modules and clear substring failures, is flown higher and covers a large plant quickly. A detailed survey, resolving individual cell-level anomalies reliably, needs enough thermal pixels across each cell and is flown considerably lower, covering much less ground per sortie.

Those are different products at different prices and we quote them separately rather than blurring them. A great many “IEC compliant” inspections sold in this market are scanning-level surveys described as detailed ones, and the difference shows up as missed cell-level findings.

Flight lines are planned so the camera views modules close to perpendicular, which on a tilted fixed-tilt array means the flight geometry is set by the array tilt rather than by convenience. Overlap is set so every module appears in multiple frames.

Alongside the thermal pass we fly a visual pass, because the two together resolve ambiguity that either alone leaves open.

Step 5: Capture

Conditions are logged continuously: irradiance from the plant’s own pyranometer where available, ambient temperature, wind and cloud state. These go in the report.

Flying over a live plant means the aircraft is over energised equipment throughout, so take-off and landing sit outside the array, transit routes avoid inverter stations and the substation, and the crew works to the plant’s electrical safety rules.

On large plants the survey is blocked so that each block is captured within a stable conditions window, rather than spreading one block across a morning of changing irradiance.

Step 6: Processing and analysis

Thermal imagery is processed into a georeferenced thermal mosaic and analysed against the abnormality classes. Anomalies are identified, classified, and then, critically, matched to the plant’s asset structure using the layout and string mapping.

Each anomaly gets: a class, a location by table, row, module position, string, combiner and inverter, the measured temperature difference against a reference, the thermal image, the corresponding visual image, and a severity or priority.

We separate findings we are confident in from findings that warrant ground verification, and we say which is which. A thermal signature that could be soiling or could be a cell defect is flagged for a technician to look at, not resolved on our side to make the report cleaner.

Step 7: Reporting

The primary deliverable is a defect register as a structured table, sortable and filterable by the plant’s own identifiers, with imagery linked. Alongside it, the thermal and visual mosaics, an anomaly map over the plant layout, and a summary with the estimated capacity affected.

On estimated losses. Where we present an estimate of capacity or generation affected, the assumptions are stated: which anomaly classes were counted, what capacity was assumed per module or string, and what yield figure was used. It is an estimate derived from a count, not a measurement, and it should be labelled that way.

On a repeat programme the format is identical between cycles, and the report includes findings that are new, findings that persist from last cycle, and findings that have been rectified. That last column is what tells you whether your O&M contractor is closing out the work.

Deliverables and output file formats

Where we work

We are based in Ahmedabad, in the state with more large-scale solar development than any other.

Gujarat. The Khavda renewable energy park area in Kutch, Charanka in Patan, Dholera, Radhanesda and the Banaskantha belt, plus distributed and C&I installations across Ahmedabad, Gandhinagar, Surat, Vadodara, Rajkot and Mundra. Kutch and Banaskantha are within straightforward reach of our base, which matters on a service where mobilisation is a real share of the cost.

Rajasthan. Bhadla, Jaisalmer, Bikaner, Jodhpur, Barmer and Nagaur, which between them hold a very large share of India’s utility-scale capacity.

Central India. Rewa, Neemuch, Agar and Shajapur in Madhya Pradesh, and sites across Chhattisgarh.

Southern India. Pavagada, Tumakuru and Kalaburagi in Karnataka, Kurnool, Anantapur, Kadapa and the Andhra Pradesh parks, Telangana, and the Tamil Nadu belt around Ramanathapuram, Tuticorin and Virudhunagar.

Northern and western India. Uttar Pradesh, Madhya Pradesh, Maharashtra including the Solapur and Dhule areas, Punjab, Haryana and Uttarakhand.

Eastern India. Odisha, West Bengal, Jharkhand and the north-east, where plants are smaller and more dispersed and the economics favour clustering visits.

What we do not do

We do not carry out electrical testing. I-V curve tracing, insulation resistance testing, continuity and earthing tests are ground-based electrical work.

We do not do electroluminescence imaging, which is what finds microcracks.

We do not carry out any electrical work, switching, isolation, cleaning, repair or module replacement.

We do not certify plant performance, warrant a performance ratio, or issue any statement of compliance. We inspect and report; the determination is the owner’s, their independent engineer’s, or their insurer’s.

We do not diagnose the root cause of every anomaly from thermal evidence alone, and we say which findings need ground verification.

We do not fly to IEC TS 62446-3 conditions when the conditions are not there, and we will move a survey date rather than deliver a report that does not meet them.

We do not fly in red zones.

Frequently asked questions

How long does a plant inspection take? It depends on plant size, the detail level and the weather window. A scanning-level survey covers ground quickly; a detailed cell-level survey covers far less per day. We quote days on site for your specific plant and detail level rather than a general figure, and we build weather contingency in.

What is the difference between a scanning survey and a detailed one? Flying height and therefore thermal resolution at the module. A scanning survey reliably finds string outages, dead modules and clear substring failures. A detailed survey resolves individual cell-level anomalies. They are different products at different prices, and a scanning survey sold as a detailed one will miss cell-level findings.

Can you inspect during the monsoon? Generally not to the standard. The conditions requirement of at least 600 watts per square metre irradiance with stable low cloud rules out most monsoon days in most of India. We would rather move the survey than produce a report that does not meet the conditions.

Do you need our plant drawings? Yes, and this is the single thing that most affects how useful the report is. The as-built layout, table and row numbering, and string to combiner to inverter mapping let us address every finding to your own identifiers. Without them you get coordinates and pictures, and somebody at the plant has to translate.

Does the plant need to be shut down? No. The opposite: the plant must be operating and generating for thermal inspection to work at all, because the defects reveal themselves through current flow.

Should we clean before the inspection? Ideally yes. Soiling produces its own thermal patterns that can mask module defects. Inspecting after a cleaning cycle gives a cleaner result in both senses.

Will you tell us how much generation we are losing? We give an estimate of affected capacity by anomaly class with the assumptions stated. It is derived from a count of findings, not a measurement of output, and the report labels it as such.

Can this support a module warranty claim? It can form part of one. Warranty procedures commonly reference IEC TS 62446-3, which is why the conditions record and the classification matter. Whether a claim succeeds depends on the warranty terms and usually on further testing. We provide evidence into the process.

Should we do this every year? Most plants benefit from an annual inspection, and the comparison between cycles is where a lot of the value sits, because it shows progression and tells you whether previous findings were actually fixed. The most valuable single inspection, though, is the one at commissioning, which establishes the baseline before the warranty period runs.

Can you inspect rooftop plants? Yes, subject to airspace and surrounding buildings. The roof structure and membrane can be surveyed at the same time.

Get a quote

Send us the plant location and capacity, the module make and rating, whether the array is fixed tilt or tracking, and the detail level you need. Attach the as-built layout and string mapping if you have them, and tell us if you do not.

Tell us whether this is a commissioning baseline, a routine annual inspection, a troubleshooting survey on a known underperformer, or a post-storm damage assessment, because each is scoped differently.

We will return the airspace position, the survey window against seasonal conditions, days on site, and a fixed price.