Risk

Risk and Reliability in Spaceflight: Why Many Small Odds Add Up

A rocket is thousands of parts that all have to work. Reliability engineering is the discipline of working out how likely that is, and how to make it more likely.

Reliability is a probability

In reliability engineering, reliability is defined as the probability that a product, system or service will perform its intended function adequately for a specified period of time, or operate in a defined environment without failure. That definition matters: it turns "is this rocket safe?" into a question about numbers, which can be estimated, tested and improved.

Parts in series

If a system needs every one of its parts to work, and the parts fail independently, the system's reliability is the product of the parts' reliabilities. Multiplying numbers less than one always gives a smaller number, so long chains of individually reliable parts can be surprisingly fragile.

Illustration only. Imagine a hypothetical system of 100 independent parts, each 99.9% reliable, all of which must work:

R(system) = 0.999^100 ≈ 0.905

Each part looks almost perfect, but together there is roughly a one-in-ten chance that something fails.

Parts in parallel

The answer is redundancy: duplicating critical components so that the system keeps working if one fails. If two independent units can each do the job, the system fails only if both fail, so you multiply the failure probabilities instead.

Illustration only. Two hypothetical independent units, each 95% reliable, either of which is enough:

P(both fail) = 0.05 × 0.05 = 0.0025
R(pair)      = 1 - 0.0025  = 0.9975

Redundancy costs mass, power and money, all of which are scarce on a spacecraft, so engineers choose carefully where it is worth it.

When independence breaks

All of these sums assume failures are independent. In real systems they often are not. Two "redundant" units fed by the same power supply, or made in the same batch, or exposed to the same cold, can fail together. These common-cause failures are one reason why real reliability analysis is harder than textbook multiplication, and why designers separate backup systems physically and use different designs where they can.

Risk outside the rocket

Orbit itself has hazards. Space debris includes defunct satellites, spent rocket stages and fragments from collisions. In 1978 NASA scientists Donald J. Kessler and Burton G. Cour-Palais proposed what is now called the Kessler syndrome: a scenario in which the density of objects in low Earth orbit becomes high enough that collisions cascade, each one creating debris that makes further collisions more likely. It is a picture of risk that feeds on itself, rather than adding up independently.

Reading risk claims

Ask what is being counted: per launch, per mission, per hour or per part.
Ask whether the numbers come from flight history, tests or models.
Be wary of small samples. A vehicle with few flights can have a perfect record by chance.

Estimating risk well also depends on understanding randomness itself.

REV 2026-11-09 / Tobias Wren