An ISS solar string can fail, disappear, and seem to return

What the audit establishes

Each ISS power channel begins with 82 strings distributed across two original blankets. Losing one string costs a few hundred watts of usable generation, small beside station-scale totals but material when aging arrays and higher utilization tighten operating margin. The audit's central point is methodological: no single observation gives a stable count of failed original strings.¹

The complication became sharper after roll-out solar arrays were installed in front of the original wings. Some iROSA circuits connect into legacy circuitry. When an installation electrically replaces a failed original string, station telemetry can look as though that string returned. The old hardware did not repair itself; the electrical boundary moved.

Why state estimation is harder than fault detection

The original blankets shunt strings in a defined order as demand falls. That switching sequence creates a recognizable power signature, but shunted strings produce no electrical output and heat under sunlight. Sensors, infrared imagery, astronaut photographs, and inspection of trend breaks therefore observe different consequences of the same circuit state.

The 2022 anomaly demonstrates the difference between a component failure and a network failure. Visible arcing damage on an inboard ribbon cable, thought to follow micrometeoroid or orbital-debris damage, removed twelve strings from useful service. A later EVA unplug removed eight more. Because two strings share a return trace while each has its own hot trace, damage location determines whether a single string or a larger group disappears.

Where a skeptic should push

This is an engineering audit presentation, not a controlled reliability experiment. The slides document known events and diagnostic methods, but do not estimate a universal string-failure rate or prove that the apparent MMOD cause explains every damaged conductor. Counts also depend on the chosen boundary: original blanket, augmented wing, or power channel.

That limitation is the useful result. A health metric that changes meaning after an upgrade is not a clean lifetime indicator. Forecast models need configuration history alongside telemetry, particularly when new generation is wired through old distribution hardware.

The bottom line

High confidence: ISS usable-power forecasting must track circuit topology and maintenance events, not only measured output. Moderate confidence: the same lesson generalizes to long-lived lunar grids assembled in increments. It would be broken by a diagnostic architecture that can uniquely identify every source and conductor across every reconfiguration, which the reviewed ISS methods do not provide.

For space power, configuration history becomes a sensor

The non-obvious implication is that a digital power model needs event provenance as much as voltage and current. The opportunity is more accurate margin planning from data already produced by operations. The threat is false recovery: a replacement source can mask an unresolved legacy fault, leaving planners to over-credit damaged distribution paths during later contingencies.