Supervision and Monitoring of Solar Farms: IoT Guide 2026

· 14 min read
14 min read
Eziwan Team
Eziwan Team
IoT Infrastructure

France currently has more than 21 GWp of installed photovoltaic capacity, with a target of 100 GWp by 2050. Behind these figures lies an operational reality that is often underestimated: every watt-hour not generated costs money, and without active monitoring, a solar farm consistently leaves potential output on the table.

A short-circuited string, an inverter in silent alert mode, a temperature sensor outside the specified range: these minor incidents go unnoticed for weeks or months without monitoring, resulting in significant cumulative production losses. This guide explains how to set up effective IoT monitoring for a solar power plant, ranging from a 10 kWp industrial rooftop system to a ground-mounted solar field of several megawatts.

The Hidden Costs of an Unmonitored Solar Farm​

Failing to monitor a solar farm isn't a cost savings—it's a cost deferral with interest.

Undetected Production Losses: A partially shaded or soiled cell in a string can reduce that string’s output by 20 to 60 percent, depending on the module technology (with or without power optimizers). In a 500 kWp installation, a faulty string represents a loss of 5,000 to 15,000 kWh per year depending on sunlight exposure, or €600 to €1,800 in lost revenue (at the feed-in tariff or the EPEX spot price).

Undetected Accelerated Degradation: Photovoltaic modules degrade by an average of 0.5% per year. Certain defects accelerate this degradation: hot spots, delamination, and microcracks caused by hail. If detected early through thermography or IV curve analysis, these defects cost €200–500 per module to repair. If detected late, after 2 or 3 years, they require complete replacement.

Voiding of Manufacturer Warranties: Most performance warranties for modules (25 years) and inverters (5–10 years) require monitoring of operating conditions. Without time-stamped monitoring data, it is difficult to support a warranty claim.

ENEDIS Non-Compliance: Energy producers connected to the public grid are subject to metering and reporting requirements. A facility that lacks production data may have its feed-in tariff rights challenged during audits.

Representative figures: A European study (2023) covering 400 solar farms ranging from 10 kWp to 10 MWp showed that farms without monitoring systems exhibited performance discrepancies of 3 to 8% compared to monitored farms with equivalent characteristics. For a 1 MWp plant producing 1,200 MWh/year, this represents a loss of 36 to 96 MWh, or €3,600 to €14,400 per year, depending on the price of electricity.

IoT Architecture for Monitoring a Solar Farm​

An effective photovoltaic monitoring architecture is organized into three layers.

Field Layer (Sensors and Equipment):

  • Inverters: central inverters or micro-inverters, depending on the installation
  • String combiners with per-string measurement
  • Weather station: global horizontal irradiance (GHI), irradiance in the plane of the modules (POA), ambient temperature, module temperature, wind speed and direction
  • ENEDIS production meter (Linky PRO or approved B2B meter)
  • Temperature sensors on DC cables and circuit breaker panels

Communication Layer (IoT Gateway):

  • Local industrial gateway with Modbus RS-485, Modbus TCP, and/or SunSpec interfaces
  • 4G LTE connectivity for remote sites or as a backup to fiber
  • Local data storage in the event of a connectivity outage (store-and-forward)
  • Local data preprocessing (aggregation, calculation of performance metrics)

Cloud Layer (Monitoring and Analytics):

  • Real-time monitoring dashboard
  • Historical production and performance data
  • Alert and notification engine
  • Production reports for operators and investors
  • API for exporting data to accounting systems or energy aggregators

For an industrial rooftop installation of less than 36 kWp, the system architecture can be simplified: a single inverter with a built-in communication interface, a 4G gateway with Modbus RTU, and a SaaS monitoring platform.

For a 1 MWp ground-mounted solar farm, the complexity increases: multiple distribution boxes, a dedicated weather station, a separate ENEDIS meter, and potentially an industrial Ethernet subnetwork between the gateway and the equipment.

Communication Protocols: Modbus SunSpec, the solar industry standard​

The photovoltaic industry has adopted the SunSpec Alliance as its interoperability standard. SunSpec defines standardized data models implemented over Modbus TCP (port 502) or Modbus RTU (RS-485).

Why SunSpec Is Important: Before SunSpec, each inverter manufacturer had its own proprietary Modbus registry. Integrating SMA, Huawei, SolarEdge, Fronius, and Sungrow inverters into a single monitoring system required as many drivers as there were brands. SunSpec standardizes the addresses and format of essential data.

The main SunSpec models:

  • Model 1: Common (equipment identification, firmware version)
  • Model 101/102/103: Single/Split/Three-Phase Inverter (AC power, voltage, current, frequency, energy generated, alarms)
  • Model 160: Multiple MPPT Extension (data per MPPT tracker)
  • Model 307/308: Weather Station (irradiance, temperature)
  • Model 801–810: String Combiner (voltage, current, power per string)

In practice, SunSpec implementation varies by manufacturer and firmware version. Always check the list of supported SunSpec models in the inverter documentation, and test register reads before deployment.

Alternatives: For inverters that do not support SunSpec, proprietary protocols are available via RS-485 or through manufacturer apps (SMA Sunny Portal, Huawei FusionSolar, SolarEdge Monitoring). These solutions are functional but create a dependency on the manufacturer’s cloud, which can be problematic for data sovereignty and service continuity.

Key Data to Collect (strings, inverters, weather, meters)​

A monitoring system must collect data at levels of detail and frequencies appropriate for each use case.

Inverter Data (every 1 to 5 minutes):

  • Instantaneous AC power (W or kW)
  • Energy generated: daily / monthly / total (kWh)
  • DC voltage and current per MPPT
  • AC voltage and current per phase
  • Grid frequency
  • Internal inverter temperature
  • Status code and active alarm codes

String data (every 5 to 15 minutes):

  • DC voltage and current per string
  • Calculated power per string
  • Ratio relative to the reference string (anomaly score)

Weather Station (every 1 to 5 minutes):

  • Irradiance in the plane of the modules (W/m²) — POA pyranometer sensor
  • Ambient temperature (°C)
  • Module temperature (°C) — Pt100 or NTC sensor on reference cell
  • Wind speed (m/s) — for estimating cooling effects and detecting snow

Generation Meter (every 15 minutes):

  • Active energy fed into the grid (kWh) — ENEDIS billing reading
  • Instantaneous active power (kW)
  • Power factor (for power ratings > 3 kWp connected to a three-phase system)

Calculated Metrics (Performance Ratio, PR): The Performance Ratio is the fundamental KPI for a PV plant. It is calculated as the ratio of the energy actually generated to the energy that could theoretically be generated under the same irradiance conditions. A PR of 80% is typical for a well-maintained installation in France. A PR below 75% should prompt an investigation.

Predictive Alerts: Detecting Clogging and Deterioration Before Loss Occurs​

Real-time alerts flag failures that have already occurred. Predictive alerts allow for intervention before production losses become significant.

Soiling Detection

Gradual soiling of the modules is often the primary cause of underperformance in solar farms located in agricultural areas, near roads, or in semi-arid regions. Detection is based on monitoring the Soiling Index:

Soiling Index = Measured PR / Reference PR for Clean Modules

When the Soiling Index falls below 0.97 (a 3% loss), cleaning is economically justified in most scenarios. For a 1 MWp installation, a 3% loss represents approximately 36 MWh/year, or €3,600 to €7,200 depending on the price of electricity—the cost of professional cleaning is generally €0.01 to €0.02/Wp.

String-Based Anomaly Detection

Comparing strings with the same orientation, tilt, and number of modules makes it possible to detect anomalies:

  • A string with a voltage 15% below the median of similar strings warrants investigation (partial shading, loose connection, faulty module)
  • A string at 0 V and 0 A indicates an open circuit (fuse, connector, cable)
  • A string with normal voltage but low current suggests an open cell or an activated bypass diode

Detection of Gradual Drift

A 30- or 90-day trend analysis normalized by irradiance can detect a gradual decline of 1 to 2 percent below the normal annual rate, which is an indicator of accelerated aging (hot spots, microcracks).

4G LTE Connectivity for Off-Grid Solar Farms​

The vast majority of ground-mounted solar farms are located in rural areas or on the plains, far from urbanized areas. Wired connectivity (fiber, ADSL) is often unavailable or not economically feasible for the small amount of daily data generated by a modest-sized solar farm.

4G LTE connectivity is the standard solution for the vast majority of installations:

Advantages:

  • Nationwide coverage of over 98% in France (Orange, SFR, Bouygues)
  • Deployment requires no civil engineering work
  • Low monthly cost (€5 to €30 for an M2M SIM with low data usage)
  • Failover possible with a secondary SIM from another carrier

Recommended Architecture:

  • Industrial gateway housed in an on-site communications cabinet, powered by the facility’s AC grid (or via battery + dedicated solar panels if an independent power supply is desired)
  • External LTE antenna with low cable loss (LMR-240 cable or equivalent, max. 3 m between antenna and gateway)
  • M2M SIM on a private APN to isolate the park’s traffic from the public network

Bandwidth Sizing: For a 1 MWp solar farm with 20 inverters and a weather station, data collected every 5 minutes amounts to approximately 200–500 MB per month, which is easily covered by a 1 GB/month M2M plan.

Store and forward: The gateway must store data locally in the event of a loss of connectivity (carrier network maintenance, weather-related incident) and automatically transmit it when connectivity is restored. A local history of at least 72 hours is recommended.

ENEDIS Compliance and Generation Metering​

Facilities connected to the French public grid are subject to the metering requirements defined by ENEDIS in its Technical Connection Specifications (PTR).

Generation Meter:

  • For single-phase or three-phase installations ≤ 36 kVA: a Linky meter or approved equivalent, read by ENEDIS.
  • For installations > 36 kVA: an approved remote-reading meter with a communication interface (IEC 62056, Modbus TCP, or M-Bus, depending on the version), read by ENEDIS or via self-reported readings.

Load Curve: Producers subject to the purchase obligation with a capacity greater than 36 kVA must provide ENEDIS with load curves at 10-minute intervals. This requirement necessitates a local metering system with data storage and automatic transmission.

ENEDIS Interface for Self-Consumption: For self-consumption systems that sell surplus energy, the ENEDIS meter simultaneously measures both feed-in and consumption. The IoT monitoring system can retrieve this data via the ENEDIS API (Metering Data Access Feed) to integrate it into the dashboard.

Purchase Agreement: Production data is generally self-reported or recorded by the operator, depending on the type of agreement. Monitoring data with a time-stamped history serves as supporting documentation in the event of a dispute over production volumes.

Typical ROI for a Solar Monitoring System​

Investing in a monitoring system is justified for installed capacity of approximately 10 kWp or more, depending on the value of the energy and the type of sales contract.

Typical Costs of a Monitoring System (2026):

  • Lightweight SaaS solution for installations ≤ 100 kWp: €300–800 installation fee + €15–50/month subscription fee
  • Professional solution for a 100 kWp–1 MWp array: €1,500–5,000 installation fee + €50–200/month
  • Operator solution for installations > 1 MWp with a dedicated weather station: €5,000–20,000 installation fee + €200–800/month

Quantifiable Benefits:

  • Early detection of anomalies: 1.5% to 3% increase in production (conservative range based on available studies)
  • Reduction in unnecessary preventive maintenance interventions (scheduling based on actual performance rather than a fixed schedule)
  • Optimization of cleaning schedules: 0.5 to 1% additional production
  • Reasonably documented total gain: 2 to 4% of annual production

Simplified calculation for a 200 kWp system:

  • Typical annual production: 200 kWp × 1,100 h/year = 220,000 kWh
  • Monitoring benefit: 2.5%: 5,500 kWh × 0.10 €/kWh = 550 €/year
  • Monitoring cost for 200 kWp: ~100 €/month = 1,200 €/year
  • Result: neutral in terms of direct savings, but offset by the value of preserved warranties, ENEDIS compliance, and a reduction in emergency service calls.

For a 1 MWp plant, the calculation clearly shows a positive result: monitoring savings of 2.5% × 1,100 MWh = 27.5 MWh, or €2,750 to €5,500 depending on the price of electricity, for an annual monitoring cost of €2,400 to €9,600. The ROI is achieved in 1 to 2 years for most projects.

Example of an Eziwan architecture for a solar farm​

The typical architecture of a solar monitoring system covers the entire chain, from the string to the dashboard:

Inverters and trackers (Modbus SunSpec, RS-485) + energy meter + weather station → Eziwan industrial 4G gateway → multi-carrier M2M SIM card → outbound VPN tunnel → Eziwan cloud → performance ratio, low-output alerts, and remote access to inverters.

In a ground-mounted solar farm, the gateway is powered by the auxiliary network at the delivery substation and operates completely autonomously: no telecom infrastructure needs to be installed, and no network intervention is required on the operator’s end. Remote access allows O&M teams to diagnose a faulty inverter before dispatching a field crew—often the difference between a 2-hour and a 2-day loss of production.

Checklist for Implementing Supervision of an Existing Fleet​

  1. Inventory of inverters: brands, models, firmware versions, and availability of the SunSpec Modbus profile.
  2. Audit of the existing RS-485 bus: topology, length, terminations—the number one cause of missing data.
  3. Connectivity selection: multi-carrier M2M SIM (fleets are almost always in rural areas), outdoor antenna if the delivery station is made of metal.
  4. Definition of KPIs: expected performance ratio per string, under-production alert thresholds, inverter availability.
  5. Metering connection: consistency between production measured at the inverters and energy metered at the delivery point.
  6. Alerts and O&M on-call duty: Routing of critical alarms (inverter failure, loss of communication) to the operations team.

FAQ​

Can Enphase, SMA, Fronius, and ABB inverters be monitored using a single gateway? Yes, provided that the inverters expose their data via Modbus TCP or SunSpec. SMA (SunSpec Modbus), ABB (Modbus TCP), Fronius (Modbus TCP SunSpec), and certain Enphase models via their local gateway are compatible. Eziwan offers preconfigured profiles for common brands, reducing the initial setup time to just a few hours.

Does the gateway work on a non-fiber network—4G only? This is the most common deployment scenario for ground-based stations or industrial rooftops located far from urban centers: the built-in 4G LTE gateway connects directly via the mobile network. For solar farms in rural areas, the Dual SIM ensures uninterrupted data transmission even if one carrier’s service goes down.

What is the minimum data collection frequency required to detect a failure in real time? For rapid detection of a string with no power: every 5 minutes is sufficient. For calculating the Performance Ratio: every 15 minutes. For ENEDIS compliance (10-minute production curves): every 10 minutes. The Eziwan gateway allows you to configure different sampling frequencies for each variable based on its importance.

Further Reading​

Want to learn more? Eziwan offers IoT monitoring solutions for solar farms, from small industrial rooftops to large ground-mounted plants. Discover our energy monitoring solutions or contact us for a quote tailored to your installation.

Additional Resources​