On-Orbit Failure Analysis of 1,200 Satellites: Common Failure Modes by System Type

A 10-year study of on-orbit anomalies and failures across 1,200 satellites identifies that power system failures and attitude control software faults account for 67% of mission-limiting anomalies, with SEU-induced software faults increasing proportionally with altitude.

What Actually Fails on Satellites: A Decade of Data

The space industry has historically been conservative about sharing failure data. The increasing commercialisation of space — and the corresponding growth in new entrants who need this knowledge — has made data sharing more common. This study, covering 1,200 satellites across LEO, MEO, and GEO orbits from 2015-2024, provides the most comprehensive failure analysis published to date.

Power System Failures: Still the Leading Cause

Despite decades of improvement, power systems account for 31% of mission-limiting anomalies. The dominant failure modes: solar array degradation beyond design predictions, battery capacity loss in orbits with high eclipse fraction, and power distribution failures in the PCU.

Notably, solar array degradation was the leading single cause — and in 73% of analysed cases, the degradation rate had been observable in telemetry for 6-18 months before mission impact. The implication: improved trending analysis of power telemetry would have enabled proactive response in the majority of cases.

Attitude Control Software Faults: The Rising Category

Attitude control software faults increased from 11% to 24% of mission-limiting anomalies over the study period — the largest relative increase of any category. The dominant mechanism: Single Event Upsets (SEUs) in onboard processors causing software state corruption. SEU rates increase with altitude (higher radiation environment) and with newer, smaller semiconductor process nodes (lower critical charge).

The study finds that satellites at GEO experience SEU-induced software faults at 3-4x the rate of LEO satellites. Mitigation approaches in use by higher-reliability missions: radiation-hardened processors (expensive, performance-limited), redundant voting processors, and software-implemented SEU mitigation (scrubbing, triple modular redundancy in critical state registers).

Structural Findings Relevant to SE Practice

Two findings with direct SE implications: first, anomalies detected within the first 3 months of operations were 4x more likely to be attributable to integration and test (I&T) escapes than to design defects. Second, missions with digital thread implementations — continuous configuration management linking design models to as-built documentation — had 40% lower rates of I&T escapes than missions using traditional document-based configuration management.

The first finding argues for more rigorous pre-launch I&T, particularly for software-hardware integration testing under realistic conditions. The second argues for investment in digital thread infrastructure as a reliability tool, not just an efficiency tool.