Solar portfolio monitoring has evolved from simple inverter dashboards into multi-layered platforms combining real-time electrical data, weather analytics, drone-based visual inspection, and AI-driven anomaly detection. For owners and operators of utility-scale portfolios from 50 MW to 1 GW, getting monitoring right is not a technology question — it is a financial one.
A portfolio running at 97% availability instead of 99% loses roughly $90,000 a year per 50 MW at $50/MWh and a 20% capacity factor. Undetected bypass diode failures and soiling below SCADA alert thresholds add further losses every day they go unfound. At scale, monitoring gaps compound.
Key takeaway: Complete monitoring requires 4 layers: SCADA + weather + drone inspection + AI analytics. Many financially significant faults sit below SCADA alert thresholds and are missed by SCADA alone. Top-quartile operators achieve >98.5% availability and $5–12/MWh O&M cost. Semi-annual IEC TS 62446-3 drone inspection is the highest-ROI gap to close.
This guide is the definitive reference for building or upgrading your portfolio monitoring programme — from architecture decisions to KPI frameworks to the inspection programmes that fill the gaps SCADA cannot cover.
What Solar Portfolio Monitoring Encompasses in 2026
“Solar monitoring” is frequently used to mean SCADA — the real-time electrical monitoring system that tracks inverter output, string current, and energy production. This is necessary but not sufficient.
A complete portfolio monitoring programme in 2026 includes four distinct layers, each covering fault types and performance gaps the others cannot detect:
| Layer | What It Is | What It Detects |
|---|---|---|
| Layer 1 | Electrical/SCADA monitoring | String outages, inverter failures, major production losses |
| Layer 2 | Weather and irradiance monitoring | Performance context for PR calculations |
| Layer 3 | Visual inspection (drone thermal + RGB) | Thermal anomalies, soiling, physical damage |
| Layer 4 | AI analytics | Cross-site anomaly detection, prediction, automation |
Most operators have Layer 1. Fewer have Layer 2 properly instrumented. A minority have Layer 3 on a scheduled basis. Layer 4 is emerging.
The 4 Layers of a Complete Monitoring Stack
Layer 1: Electrical / SCADA Monitoring
What it covers: String current, string voltage, inverter AC output, energy meter readings, availability status, inverter alarms, grid power quality.
What it detects: String outages, inverter failures, major production losses (>5% of expected output), grid connection issues, communication faults.
Detection limit: SCADA operates above a threshold. A fault that causes less than 2–5% production loss (the typical alert threshold) may never trigger an alert. Bypass diode failures, moderate soiling, and panel-level degradation typically fall below this threshold.
Data resolution: 1-minute or 15-minute intervals is standard for utility-scale. Research-grade monitoring uses 1-second intervals.
Communication protocols: Modbus TCP/RTU, SunSpec, IEC 61850, OPC-UA. Multi-brand portfolios require protocol normalisation — a key capability to verify in any solar panel monitoring software.
Layer 2: Weather and Irradiance Monitoring
What it covers: Plane-of-array irradiance (POA), global horizontal irradiance (GHI), ambient temperature, module temperature, wind speed, precipitation.
Why it matters: Production data without irradiance context is uninterpretable. A 10% production drop could be normal (cloudy day) or abnormal (inverter fault). The Performance Ratio calculation — the primary normalised metric in utility-scale solar performance monitoring — requires irradiance data.
IEC 61724-1 Sensor Requirements
| Accuracy Class | Level | Typical Use |
|---|---|---|
| Class A | High accuracy | Utility-scale plants; performance-guarantee and contract verification |
| Class B | Medium accuracy | Smaller commercial systems |
| Class C | Basic accuracy | Small systems with simple monitoring |
IEC 61724-1 defines monitoring-system accuracy classes (A, B and C). Utility-scale plants normally specify Class A, and the standard sets the sensor and data-acquisition requirements for each class.
Satellite irradiance data: Modern monitoring platforms increasingly supplement on-site sensors with high-resolution satellite irradiance data (e.g., SolarAnywhere, Solargis). Satellite data provides higher spatial resolution across a site and helps flag drifting on-site sensors, but it supplements rather than replaces the irradiance measurements the plant’s IEC 61724-1 accuracy class requires.
The IEA PVPS publishes monitoring and performance benchmarking data that informs these standards.
Layer 3: Visual Inspection
What it covers: Thermal anomalies (electrical faults), soiling patterns, physical damage, structural issues, delamination.
Why it matters: Layer 1 and Layer 2 together provide an excellent picture of how much electricity your site produces relative to what the weather allows. They cannot tell you why production is below expectation at the panel level. Visual inspection answers that question.
The detection gap: Many financially significant faults at utility-scale solar sites are not flagged by SCADA before visual inspection finds them. Bypass diode failures, moderate soiling, developing hotspots, and early-stage delamination all fall below SCADA alert thresholds.
Standard Inspection Programme
| Inspection Type | Frequency | Coverage | What It Catches |
|---|---|---|---|
| Full drone thermal + RGB | Semi-annual | 100% of panels | All thermal faults, soiling, structural |
| Physical walk | Quarterly | Full site perimeter + spot check | Vegetation, security, visible damage |
| Targeted drone re-inspection | As-needed | Flagged zones only | Follow-up on SCADA anomalies |
| EL inspection | Every 3–5 years | Sample or full site | Micro-cracks, cell defects |
IEC compliance: Full drone thermal inspection per IEC TS 62446-3 provides the documentation required for insurance, financing, and regulatory compliance.
For operators conducting thermal inspections, ClearSpot’s solar panel thermal inspection provides AI-powered analysis.
Layer 4: AI Analytics
What it covers: Cross-site anomaly detection, fault classification, predictive modelling, automated work order generation.
Why it matters: At 50 MW, a skilled analyst can review SCADA data and inspection reports manually. At 500 MW, the data volume makes this impractical. AI handles the pattern recognition at scale, surfacing only the anomalies that require human attention.
Current AI Capabilities in Production Use
- Automated thermal anomaly classification (ClearSpot reports 99% fault-detection accuracy for bypass diode failures and hotspots)
- Cross-site statistical anomaly detection (identifying sites underperforming vs peer group)
- Predictive alerts based on trend analysis in SCADA data
- Automated work order generation from inspection faults
- IEC compliance report generation
Key Performance Metrics Every Portfolio Manager Must Track
| Metric | Definition | Formula | Target (Utility-Scale) |
|---|---|---|---|
| Performance Ratio (PR) | Actual vs theoretical yield normalised for irradiance | PR = Yield (kWh/kWp) ÷ (Irradiance (kWh/m²) × System efficiency) | 75–85% |
| Specific Yield | Annual energy per unit capacity | kWh/kWp/year | Location-dependent (1,100–2,000) |
| Availability Factor | Fraction of time system is operational and able to generate | Available hours ÷ Total hours | >98.5% (top quartile) |
| Soiling Ratio | Current PR relative to clean-panel PR | Measured PR ÷ Clean PR | >0.97 (trigger cleaning below) |
| Clipping Loss | Energy lost when inverter limits AC output | kWh clipped ÷ Total generation | <2% (design-dependent) |
| Capacity Factor | Actual generation vs maximum possible if generating at nameplate 24/7 | Actual MWh ÷ (Capacity MW × 8,760h) | 15–25% (location-dependent) |
| O&M Cost per MWh | Total O&M spend divided by energy produced | Total O&M cost ÷ Annual generation (MWh) | $5–12/MWh (top quartile) |
Review Cadence
- Daily: PR, availability, active alarms
- Weekly: Soiling ratio, open work orders
- Monthly: All KPIs vs budget, MTTD and MTTR trends
- Quarterly: O&M cost per MWh, inspection schedule progress
- Annually: Full performance review, IEC compliance documentation update
The IEC (International Electrotechnical Commission) publishes IEC 62446 and IEC 61724 standards for solar monitoring compliance.
Monitoring Architecture for 50 MW–1 GW Portfolios
50 MW (Single Site)
Monitoring needs:
- String-level SCADA with 15-minute data resolution
- 1–2 pyranometers + module temperature sensors
- Cloud-based monitoring platform with mobile access
- Semi-annual drone inspection
- Basic alert management (email/SMS for critical events)
Typical platform: Mid-market SaaS monitoring platform with drone inspection integration or separate drone inspection contract. Named options are compared in our guide to the best monitoring platforms for utility-scale solar.
Data storage: Cloud-based, 10-year retention minimum for lender compliance
200 MW (Multi-Site or Large Single Site)
Monitoring needs:
- Sub-string or string-level SCADA with 5-minute resolution
- IEC 61724-1 Class A sensor array (multiple pyranometers for large sites)
- SCADA historian with API access for third-party analytics
- Semi-annual drone inspection with AI analysis
- AI-assisted anomaly detection
- CMMS integration for work order automation
Additional considerations: Multi-site portfolios need a portfolio-level view aggregating data from each site, with drill-down to string level, typically delivered through an asset management platform built for multi-site solar portfolios. Communication redundancy becomes important — a SCADA communication failure at a 200 MW site costs more per hour than at 50 MW.
500 MW–1 GW (Enterprise Portfolio)
Monitoring needs — All of the above, plus:
- Dedicated data historian (OSIsoft PI, Aveva, or equivalent)
- OPC-UA or IEC 61850 for enterprise SCADA integration
- Cross-site AI benchmarking and fleet analytics
- Automated IEC 62446 and 61724 compliance documentation
- Agentic AI for work order generation and escalation
- Dedicated SCADA engineering team
Architecture consideration: At this scale, the monitoring platform is an enterprise IT system requiring security review, data residency compliance, and service level agreements equivalent to mission-critical infrastructure. Its outputs also feed the portfolio management layer that handles budgets, contracts and investor reporting.
SCADA vs Cloud-Based Monitoring: The Technology Transition
Traditional utility-scale solar monitoring was built on on-premise SCADA systems — industrial control systems hosted in site control buildings, operating on dedicated communication networks. This architecture is reliable and secure but expensive to operate and difficult to scale across multiple sites.
Cloud-based monitoring platforms have largely replaced on-premise SCADA for all but the largest and most sophisticated operators:
| Dimension | On-Premise SCADA | Cloud Platform |
|---|---|---|
| Upfront cost | High (hardware + integration) | Low (SaaS subscription) |
| Ongoing cost | IT staff + maintenance | Per-site subscription |
| Multi-site rollout | Complex, slow | Days to weeks |
| Data accessibility | On-site or VPN | Any browser, any device |
| Software updates | Manual, disruptive | Automatic, continuous |
| AI capability | Limited (static algorithms) | Continuous improvement |
| Redundancy | Requires investment | Built-in (cloud infrastructure) |
| Cybersecurity | Operator-managed | Vendor-managed (with due diligence) |
The hybrid reality: Many large portfolios maintain a thin on-site data acquisition layer (for reliability and communication) feeding a cloud analytics platform. The SCADA in the control building handles real-time control; the cloud platform handles monitoring, analytics, and reporting.
The Role of Drone Inspection in Portfolio Monitoring
Drone inspection of utility-scale solar farms is not a replacement for SCADA — it is the layer that answers questions SCADA cannot. The two systems are complementary:
| SCADA tells you: | Drone inspection tells you: |
|---|---|
| “String 7 on Inverter 3 is producing 8% below its neighbours” | “String 7 has 4 panels with bypass diode failures and 2 panels with significant soiling — here are their exact GPS coordinates and thermal images” |
Without drone inspection: You know something is wrong but must dispatch a technician to investigate. Mean time to identify: 3–7 days.
With drone inspection: You receive a prioritised fault list with exact locations within 24 hours of the flight. Mean time to identify: <1 day.
Scheduling Principle
- Semi-annual drone inspection provides complete site coverage twice per year
- The optimal timing is spring (before peak summer generation) and autumn (before low-irradiance winter period)
- SCADA anomalies can trigger targeted follow-up drone flights between scheduled inspections — a 50 MW section of a large site can be re-inspected in 1–2 hours
Compliance Monitoring: IEC 62446, IEC 61724, and Lender Requirements
Two IEC standards govern solar monitoring compliance:
IEC 62446 — Inspection and Documentation Standard
- Part 1: Commissioning records and ongoing inspection requirements
- Part 3: Thermal imaging inspection protocol (drone inspection standard)
Relevant for: insurance claims, lender condition covenants, asset sales
IEC 61724 — Performance Monitoring Standard
- Part 1: Sensor and data acquisition requirements (Classes A, B, C)
- Part 3: Energy yield assessment methodology (bankability)
Relevant for: project finance, performance guarantees, refinancing
Minimum Compliance for Most Utility-Scale Project Finance
| Standard | Requirement | Frequency |
|---|---|---|
| IEC TS 62446-3 | Compliant thermal inspection | Annually or semi-annually |
| IEC 61724-1 | Class A monitoring | Continuous |
| IEC 61724-3 | Yield assessment | Available on request for refinancing/sale |
Data Management: Historians, APIs, and Integration Standards
Data volume: A 200 MW portfolio combining SCADA time series, irradiance sensors, and drone thermal and RGB imagery accumulates a large data set every year. At 10-year retention, this becomes a multi-terabyte data management challenge.
Key Integration Standards
- Modbus TCP/RTU: Most common for inverter communication
- SunSpec Alliance: Open standard for solar-specific device communication; simplifies multi-brand integration
- IEC 61850: Substation automation standard used for grid-connected protection systems
- OPC-UA: Enterprise SCADA integration standard
API access: Modern monitoring platforms provide REST APIs for data export. Verify that your platform provides:
- Historical data export without row limits
- Real-time data streaming for third-party analytics
- Standard format export (CSV, JSON, Parquet) without proprietary lock-in
Alert Management: Avoiding Alarm Fatigue at Scale
A utility-scale solar site can generate hundreds of alerts per day — most of which require no action. Alarm fatigue (operators tuning out alerts because most are noise) is one of the most common causes of late fault detection.
Best Practice Alert Hierarchy
| Level | Trigger | Response Time | Who Is Notified |
|---|---|---|---|
| Critical | >50% inverter capacity offline, safety system trip | Immediate (<15 min) | On-call engineer + manager |
| High | String outage, inverter alarm, PR <80% | 2–4 hours | O&M team |
| Medium | String underperformance (10–20%), communication fault | Next business day | O&M team |
| Low | Minor anomaly, data quality warning | Weekly review | O&M team |
AI alert filtering: AI-based anomaly detection reduces alert volume by distinguishing true anomalies from normal variation. A string performing at 92% of expected output is not an anomaly on a partly cloudy day — but it may be on a clear day. AI applies context; threshold alerts do not.
AI and Automation: Where Solar Monitoring Is Heading
The monitoring platforms entering production at new utility-scale sites in 2026 include capabilities that were experimental three years ago:
- Automated work order creation: Faults from drone inspection and SCADA anomalies automatically generate CMMS work orders with priority, location, and recommended action pre-populated
- Predictive maintenance alerts: Early-warning signals from SCADA trends 30–60 days before expected failure
- Autonomous inspection scheduling: AI identifies which zones of a large site show anomalous SCADA behaviour and schedules targeted drone follow-up automatically
- Continuous IEC compliance documentation: Every inspection event and fault resolution automatically updates the compliance record
The direction of travel is clear: from platforms that show operators data to platforms that act on it on operators’ behalf, within defined decision boundaries.
Building Your Monitoring Strategy: A Step-by-Step Framework
Step 1 — Audit your current monitoring stack
Map each of the four layers (SCADA, weather, visual, AI) against what you currently have. Identify gaps.
Step 2 — Define your KPI set
Choose 6–8 KPIs from the table above. Assign data sources for each. Confirm each KPI can be calculated from your current data.
Step 3 — Establish your inspection programme
If you don’t have scheduled drone inspection, this is typically the highest-ROI gap to close. Obtain quotes for semi-annual IEC TS 62446-3 compliant drone inspection for your portfolio.
Step 4 — Integrate inspection data with CMMS
Ensure faults identified by drone inspection automatically create work orders. Manual data transfer between inspection reports and maintenance systems is an error-prone bottleneck.
Step 5 — Set up alert hierarchy
Review your current alert configuration. Remove or reclassify any alert that fires more than 3 times per week without resulting in action — it is creating noise, not signal.
Step 6 — Establish compliance documentation process
Verify you can produce an IEC 62446 inspection report and an IEC 61724-3 yield assessment within 48 hours if a lender or insurer requests one. If not, identify the gaps.
Step 7 — Review annually
Portfolio monitoring requirements evolve as assets age, lender requirements change, and technology improves. Conduct an annual monitoring strategy review against this framework.
Conclusion
Portfolio monitoring for utility-scale solar is not a set-and-forget infrastructure decision. It is an active operational discipline that directly determines how much revenue your assets generate over their 25–30 year project life.
The operators who treat monitoring as a compliance checkbox — minimum SCADA, annual report — are systematically leaving 2–5% of generation revenue unrecovered every year. At $50/MWh and 200 MW of capacity (about 350,000 MWh a year at a 20% capacity factor), that is roughly $350,000–875,000 annually.
The operators who build all four monitoring layers and review them rigorously achieve top-quartile performance and build the documentation record that supports favourable insurance terms, successful refinancing, and premium asset valuations.
Get a free portfolio monitoring assessment from ClearSpot. Book a call below and we’ll review your current monitoring stack against this framework and identify your highest-value gaps.
For detailed aerial inspection workflows, ClearSpot’s aerial solar panel inspections guide covers step-by-step procedures.
Frequently Asked Questions
Solar portfolio monitoring is the centralised tracking of performance, health and output across multiple utility-scale solar plants, typically in portfolios from 50 MW to 1 GW. A complete programme combines SCADA, weather and irradiance data, drone thermal and RGB inspection, and AI analytics.
SCADA alerts trigger above a production-loss threshold, typically 2–5% of expected output. Bypass diode failures, moderate soiling, developing hotspots and early delamination fall below that threshold and are only found by visual inspection.
Performance ratio, specific yield, availability, soiling ratio, clipping loss, capacity factor and O&M cost per MWh, reviewed daily, weekly, monthly, quarterly and annually depending on the metric.
Semi-annually is the common baseline, in spring before peak generation and in autumn before the low-irradiance season, with targeted re-flights when SCADA flags an anomalous zone.
IEC 61724-1 sets sensor and data acquisition classes for performance monitoring, IEC 61724-3 covers energy yield assessment, and IEC TS 62446-3 defines the thermographic inspection protocol used for drone surveys.
AI classifies thermal anomalies, compares sites against peer groups, filters alert noise by applying weather context and creates prioritised work orders automatically, so analysts only review the anomalies that need action.
Leaving 2–5% of generation unrecovered at $50/MWh and a 20% capacity factor costs roughly $350,000 to $875,000 a year for a 200 MW portfolio.