The previous lesson ended on a quiet assumption: before the first integration, the system knew where down was, where north was, and where it was parked. This lesson is about how it learns all three. The learning is called alignment, it happens on the ground with strict rules, and once airborne there is no doing it again. The lesson closes with the hardware the pilot actually touches: the ADIRU promised back in module 2, the control panels, and the check flown crews use to judge each unit after landing.
Teaching the box which way is down and north
Alignment does two jobs. The first is the vertical: during the process the inertial platform is levelled (INS) or the local vertical is determined (IRS). On a platform INS the gimballed table is physically driven until the horizontal accelerometers read zero with the aircraft parked, which is what level means to an accelerometer. On a strap-down IRS nothing moves: the accelerometers feel gravity through the airframe, and the computer works out which way, through the fixed sensors, down points.
The second job is direction: true north and the aircraft's true heading are established. With the aircraft stationary, the only rotation the gyros can sense is the Earth's own turning, and the direction of that sensed rotation is the direction of the Earth's axis. Finding north by listening to the planet rotate is called gyrocompassing, and it completes a thought two modules old: module 3 said the poles' magnetic troubles get designed around, module 4 offered the free-mode directional gyro as the working crutch, and here is the full answer, a heading reference that never consulted magnetism at all.
The rules of alignment
The ritual has three rules:
The aircraft must be stationary throughout — the system is listening for the Earth's rotation and calibrating what level feels like, and any motion of its own, a tug pulling the aircraft, even taxiing, is indistinguishable from the signals being measured. Start moving and the alignment is spoiled.
The aircraft's position is entered during the alignment phase — integrated distances mean nothing without the starting point, and the corrections of the previous lesson need to know the latitude they are correcting for.
Patience — alignment takes around 10 to 20 minutes at mid latitudes, and longer at high latitudes. The reason sits in the gyrocompassing: nearer the poles, less of the Earth's rotation appears as the horizontal component the system steers north by, so the signal it is listening for grows weaker as the latitude grows.
Entering the position: latitude is the critical one
The two halves of the entered position are not equals. The split of the sensed Earth rotation between horizontal and vertical depends on latitude alone, so during gyrocompassing the system in effect measures its own latitude and compares it with the one entered. Enter a latitude that disagrees and the comparison fails: an incorrect latitude entry may lead to a loss of alignment, and is more critical than an incorrect longitude entry. Longitude leaves no signature in the sensed rotation at all. A wrong longitude is accepted without complaint and simply displaces every position the system reports afterwards.
In flight: no second chances
Once alignment is complete and the aircraft moves, the INS platform is maintained level and north-aligned: the Earth rate and transport corrections of the previous lesson continuously turn the reference so that what was true at the gate stays true across the planet. What cannot be done is starting over: in-flight realignment is not possible, because alignment needs a stationary aircraft, and in flight the system cannot separate the Earth's rotation from its own motion. Loss of alignment in flight therefore means the navigational data is lost, although attitude information may still be available: the sensors themselves are healthy, so a usable attitude reference can survive the death of the navigation.
How the error grows
Even a good alignment only slows the arithmetic of error, never stops it. The positional error of a stand-alone INS varies; a typical value is 1 to 2 NM (nautical miles) per hour of flight. Four contributions feed it:
gyro drift rate — the reference slowly wandering, as module 4 taught all spinning gyros do.
accelerometer bias — a small standing false acceleration that double integration turns into a steadily growing displacement.
misalignment of the platform — an alignment that finished slightly off level or off north, leaking gravity and mixing axes.
computational errors — the corrections and integrations are done to finite precision.
The hardware: ADIRU, panels and the CDU
Module 2 introduced the ADIRU alongside its air-data computer and promised that the inertial half would be explained here; that promise is now kept. On a modern aircraft there is likely to be an air-data inertial reference unit (ADIRU), an inertial reference unit (IRU) integrated with an air-data computer (ADC) in a single unit, the pitot-static and inertial worlds delivered together. The IRU inside is exactly the strap-down unit of the previous lesson, and the TAS its wind calculation needs now comes from the other half of the same box rather than an external feed.
The pilot meets the system in two places. The mode selector carries the IRU's master switch, typically running OFF, ALIGN, NAV and ATT, and the syllabus asks for two of those selections by name:
NAV is normal operation, the position for alignment and everything after it, delivering attitude and navigation together.
ATT is attitude only, the fallback selection that keeps the attitude reference alive when the navigation solution has been lost.
Day to day, though, the knob is not where the system is worked: the majority of the IRS data is accessed through the FMS control and display unit (CDU), or the multifunction control and display unit (MCDU) of a flight management and guidance system (FMGS): an FMGS is an FMS that also carries the guidance side. Position is entered there during alignment, and positions, velocities and status are read back in flight.