Inertial Reference and Alignment

The IR is the only sensor on the aircraft that knows its own attitude and position with its eyes closed — no external signal, no air data, just integration of three laser gyros and three accelerometers. That independence has a price: before departure it must complete a "find down, find north" ritual on the ground (alignment), and thereafter it drifts, quietly, every hour.

This article works through the three-stage ritual, why the time explodes with latitude, what a tug bumping the aircraft does to it, what survives when alignment is lost in flight (ATT mode), and the prohibited-latitude lines the magnetic-variation table draws at high latitude.

1. A strapdown system — three gyros, three accelerometers, three modes

Per AMM 34-14-00:

The IR portion is a strapdown inertial system which provides a quality reference for attitude, heading (true and magnetic), angular rates and accelerations. The IR software also computes: - the inertial position - the ground velocity - the baro inertial vertical speed - the drift angle - the wind - the flight path data.

"Strapdown" means the gyros and accelerometers are bolted directly to the airframe — no rotating platform; attitude is solved by software integration of angular rate. The IR also contains a GPS partition. Per AMM 34-14-00:

The Global Positioning System (GPS) partition integrated in the IR portion provides both the GPS autonomous and hybrid navigation solutions.

The hybrid solution (GPIRS) is the subject of the next article; for now, note that on the subject of position, the IR and the GPS share a desk. The IR's three modes — ALIGN, NAV, ATT — are set jointly by the MSU rotary and internal logic, and are unpacked below.

2. Alignment in three stages — find down, find north, then sign your name

Alignment can only be done on the ground (a hard limitation, §6). It has three stages.

Stage 1 — coarse levelling (30 s). Per AMM 34-14-00:

The coarse level processing is engaged during the first 30 seconds of the IR alignment mode. This processing estimates the local vertical using the 3 accelerometers and the measured gravity.

On a stationary aircraft the only acceleration the accelerometers feel is gravity — wherever gravity points is "down," and pitch and roll follow.

Stage 2 — gyro-compassing (≥9.5 min). Per AMM 34-14-00:

Gyro-compass processing is engaged after the 30 seconds of the IR alignment mode (coarse level complete) and runs for a minimum of 9.5 minutes. Gyro-compass processing is used to orient body frame to North (using earth rotation detection by gyros).

The Earth's rotation axis points at true north, and a stationary gyro can sense that axis — so the IR finds true north, independent of the magnetic field (magnetic heading is computed afterwards, §6). A by-product: the horizontal/vertical ratio of the rotation component also yields an estimate of the local latitude — which is about to be used to test the crew's entry.

Stage 3 — position entry. The finish is to give the integrator a starting point. Four cases. Per AMM 34-14-00:

If a GPS position data is available and if there is no pilot entry, the GPS data is used to perform the alignment.

The configuration covered here carries the "IRS alignment based on GPS position" option, so normally the crew enters nothing — the GPS signs automatically. If the crew insists on a manual entry, the system checks it against GPS. Per AMM 34-14-00:

If a GPS position data is available and there is a pilot entry, the pilot entry is compared with the GPS position data. The discrepancies between the two latitudes and the two longitudes must be within 5Nm.

Beyond 5 NM the alignment annunciation flashes. Without GPS, the BITE uses the last recorded landing position as examiner (entry vs record ≤ 1° lat/long); gyro-compassing runs a second exam — the gyro's own latitude estimate against the entered latitude (sin/cos difference ≤ 0.01234). The harshest clause. Per AMM 34-14-00:

If sin/cos test fails two times with identical set latitude inputs then: - the FAULT legend flashes on the CDU - A warning message appears on the EWD: NAV IR 1(2)(3) FAULT.

Enter the same error twice and the system declares an IR fault — it would rather self-destruct than depart with the wrong position. That is the full court behind the NAV IR NOT ALIGNED family (POSITION MISMATCH → INSERT) in the abnormal chapter.

3. Alignment time — the cosine of latitude in the denominator

Per AMM 34-14-00:

The typical alignment time will be calculated for a given latitude as follows: - ABS (5.0 minutes/cosine (latitude)) for latitudes between 60S and 60N

Between 60° and 73° it is fixed at 10 minutes; above 73°, 17 minutes. Why slower at higher latitude? Gyro-compassing relies on the horizontal component of the Earth's rotation, which scales with cos(latitude) — strongest at the equator, zero at the pole (where the rotation axis points straight down and there is no "north" in the horizontal plane to find). Numbers: 40°N ≈ 6.5 min; 43°N ≈ 6.8 min; 30.6°N ≈ 5.8 min; a 75°N polar alternate goes straight to 17 min. That is where the pre-flight time budget comes from: on high-latitude turnarounds, alignment is the long pole in the flow.

4. Getting bumped, and the half-minute reprieve

Alignment is a ritual that hates interruption. A velocity step from taxi or towing exceeding 0.5 ft/s triggers, per AMM 34-14-00:

Thirty seconds after motion detection, the system reverts to a full alignment (time to the end of alignment will revert to 9 min 30 s). It is not necessary to re-enter the position.

In the cockpit this shows as EXCESS MOTION on the MCDU + NAV IR NOT ALIGNED on the EWD + the PFD attitude flagged. It costs time, not position — so the handling of "a tug just moved us" is clean: stop, wait out the 9.5-minute re-run.

The reverse — the time-saver — is the 30-second realign. Per AMM 34-14-00:

This mode is selected by moving on the 221VU the OFF/NAV/ATT selector switch from NAV to OFF then to NAV within five seconds, when the aircraft is on ground (ground speed less than 20 knots).

It zeros the accumulated velocity error and trims the attitude/heading from the previous NAV — clear velocity, keep attitude, still need position (a valid position must be received). That is the mechanism behind the SOP rule "fast alignment is normal, full alignment only in five specific cases." Speed matters: exceed five seconds and the system thinks you are shutting down, restarting the full 9.5 minutes. And this NAV→OFF→NAV trick is meaningful only on the ground — in flight, moving out of NAV is the one-way street of the overview article.

5. What NAV mode produces — how wind, VS, and FPA are computed

Per AMM 34-14-00:

The NAV mode is the primary operating mode for the IR and is implemented in software as unaided strapdown inertial navigation computation.

Beyond the triple integration (attitude ← angular rate, velocity ← acceleration, position ← velocity), several everyday displays are NAV-mode by-products.

Wind — the numbers in the ND corner are a subtraction. Per AMM 34-14-00:

Wind speed and direction are computed from the GPIRS Ground Speed vector and the selected ADR True Airspeed (TAS). If GPIRS Ground Speed is not available, then the IR Ground Speed vector is used.

The IR does not compute the wind if the TAS is less than 100 knots or if the air data sources are no more available.

Ground-speed vector (how I move over the ground) minus airspeed vector (how I move through the air) = how the air moves over the ground = wind. So the previous article's "CAS < 60 → TAS = NCD," compounded with "no wind below 100 kt TAS," means **the wind is blank from the take-off roll to shortly after lift-off**; the wind arrow additionally waits for wind speed > 2 kt.

Baro-inertial vertical speed — the VS on the PFD is a blend. Per AMM 34-14-00:

The IR software contains a baro-inertial loop to compute the Inertial Vertical Speed and Inertial Altitude. This loop permits to take advantage of the different qualities of the inertial and air data systems. The IR brings its better behaviour in dynamic maneuvers while the ADR brings its stability in time (no drift of the outputs like in IR).

Inertial provides "fast" (the needle moves the instant you pull); baro provides "true" (no long-term drift). Lose the selected ADR and the loop is starved — inertial VS goes NCD and the display degrades to pure barometric rate. Two indication details: beyond ±6000 ft/min the needle freezes and turns amber; and on approach the digits and needle turn amber when **RA < 2500 ft with VS beyond −2000, or RA < 1000 ft with VS beyond −1200** — the display watching the stabilised-approach gate for you (the same origin as the SINK RATE > 1200 callout).

FPA / drift — FPA = arctan(inertial VS / ground speed); drift = track − heading. The TRK/FPA "bird" symbol is treated by the FCTM as an "energy microscope" in windshear because both quantities are inertial and immune to pitot/static deception.

6. Magnetic variation and the polar region — one table, two sets of latitude lines

The IR lives in a true-north world (gyro-compassing finds the rotation axis); magnetic heading is translated. Per AMM 34-14-00:

To create the labels which are magnetic referenced (magnetic heading, magnetic track angle ; labels 320, 317) the IR computes a magnetic variation which is added to the true values (labels 314, 313).

The magnetic variation is computed as a function of the present position (latitude and longitude) within the range of 82°N to 60°S excluding the magnetic polar region (latitude exceeds 73°N and longitude between 120°W and 92°W).

Note: the AMM writes the eastern edge of the magnetic polar region as 92°W, while the FCOM limitations write 90°W. Both are reproduced verbatim; the operating limitation (§ below) governs.

The grid data is interpolated from a NOAA source. Per AMM 34-14-00:

The grid data comes from the National Oceanic and Atmospheric Administration and must be updated every 10 years in order to meet the accuracy requirements.

Note: the ADIRU chapter (AMM 34-12) states this update cycle as five years. The two figures are inconsistent within the AMM and are both recorded.

What happens beyond the table's domain? LIM-NAV gives two sets of lines and two consequences. Per FCOM LIM-NAV:

IR alignment in NAV mode is possible on ground only. Ground alignment of the IRS is possible in latitudes between 82 ° North and 82 ° South.

First set (when all ADIRUs carry the same magnetic-variation table) — crossing the line only means "change reference." Per FCOM LIM-NAV:

In NAV mode, the IR will not provide valid magnetic heading and magnetic track angle: ‐ North of 73 ° North, between 90 ° West and 120 ° West (magnetic polar region) ‐ North of 82 ° North ‐ South of 60 ° South.

When flying at latitudes beyond these limits, the TRUE reference must be selected.

Second set (when any one ADIRU carries a different table) — the limits contract sharply (60°N between 30–160°W, 75°N, 55°S) and the consequence escalates. Per FCOM LIM-NAV:

Flying at latitudes beyond these limits is prohibited.

Why prohibited? Three ADIRUs consulting different tables resolve three different "magnetic norths" at the same position — the comparison monitor (HDG DISCREPANCY) is swamped by the spurious difference, and a real fault would hide behind it. The physical object behind this limitation is data table 2: if maintenance loads software with mismatched table versions, it can create an aircraft with "one table different."

The display once TRUE is selected. Per FCOM DSC-34-10-10-20:

When the NORTH REF pb-sw is set to TRUE, the ADIRUs replace magnetic heading by true heading on EFIS and DDRMI . In addition, the GRID track appears on ND.

Per AMM 34-14-00:

Grid information appears automatically in digital form (green) at the top of the ND when the NORTH REF pushbutton switch is selected TRUE and the latitude is above 65° (North or South).

The entry caution (NAV EXTREME LATITUDE → NORTH REF SEL TRUE) and the expected NAV HDG DISCREPANCY when the three IRs switch to TRUE out of step (delay the ECAM action; it self-clears in a few minutes) belong to the polar-operations article.

7. ATT mode — the attitude life-ring after navigation is lost

Per AMM 34-14-00:

The mode can be activated on ground or in flight and is intended to provide a rapid attitude/heading restart capability if the IR has experienced total power shutdown or failures which do not disable the mechanization of the attitude computation.

Its start-up is a 20-second freeze. Per AMM 34-14-00:

This mode needs a 20-second initialization phase with the aircraft in level flight. During this phase all the ARINC bus outputs are sent with their status matrix coded NCD (No Computed Data).

Level flight lets the accelerometers repeat the "gravity = down" coarse-levelling — this time airborne, called attitude erection. What if the aircraft is not level? There is a guard. Per AMM 34-14-00:

an erection cut-out function delays erection when yaw rate exceeds 0.5 deg/s and permits erection to continue when yaw rate drops below 0.25 deg/s.

The "false gravity" of a turn (centripetal acceleration) would tilt the horizon, so erection pauses in the turn and resumes when wings-level — it needs 20 seconds of quiet before it delivers. ATT-mode heading has no source (north was lost long ago); an ENTER HEADING prompt waits for a manual MCDU entry, thereafter re-checked against the standby compass every ten minutes. And one recommendation that is easy to state backwards. Per AMM 34-14-00:

However, it is recommended to stay in NAV mode even with excessive navigation errors because of higher accuracy of attitude signals and a more complete signal processing.

Drifted position ≠ time for ATT. ATT is a life-ring for someone who has lost attitude; with a large position drift, NAV-mode attitude is actually better than ATT. Only a flashing IR FAULT light, or loss of power in flight, calls for it.

8. The IR's health record — comparison cautions and removal criteria

In flight, the FWC compares the two sides. Per AMM 34-14-00:

When a difference higher than 5 deg. is detected by comparison inside the FWCs between the roll angle or the pitch angle provided by two IRs,

When a difference higher than 7 deg. (or 5 deg. in true heading) is detected by comparison inside the FWCs between the heading value provided by two IRs,

The matching PFD/ND local information is CHECK ATT / CHECK HDG (amber), handled by "the standby instrument judges, IR 3 steps up."

After landing, the crew writes the IR's health report. The drift algorithm sits in the AMM. Per AMM 34-14-00:

The IRS average drift rate is computed from the first IRS alignment and each time the MLG shock absorber is compressed by comparing IRS 1, 2, 3 positions recorded at touchdown with a known geographic landing point.

The "known landing point" is taken 400 m along the runway axis from the destination threshold; the time denominator is the greater of "time since last alignment" and one hour. The result shows on the IRS MONITOR page (Done phase). Removal uses a two-strike rule for radial position error; residual ground speed is more direct. Per AMM 34-14-00:

- if the residual ground speed error is 15 kts or greater after each of two consecutive flights, replace the ADIRU - if the residual ground speed error is 21 kts or greater at the end of any flight, replace the ADIRU.

A stopped aircraft's ground speed should be zero; an IR that still "feels" 15 kt of motion has a sick gyro/accelerometer. The SOP parking check (POSITION MONITOR deviation) and the GPS-spoofing "use the IRS MONITOR page to avoid a spurious removal" guidance later both cite this standard. One last note on the IR 3 "eats where the rotary points" logic: ADIRUs 1 and 2 always prefer their own ADR data, but IR 3's air-data source follows the AIR DATA + ATT HDG rotaries (CAPT ON 3 on the IR side only → IR 3 uses ADIRU 1's ADR; F/O ON 3 → uses ADIRU 2) — so for channel 3, "the selected ADR" in the baro-inertial loop and the wind computation is your choice.

Key numbers

Self-test

[!note]- Q1. Why does gyro-compassing fail near the pole, and how does that explain the alignment-time formula? It relies on the horizontal component of Earth's rotation, which scales with cos(latitude) — zero at the pole. Hence time ∝ 1/cos(latitude): fast near the equator, exploding toward high latitude.

[!note]- Q2. With GPS available you enter a position manually. How is it adjudicated? Without GPS? Consequence of the same error twice? With GPS: manual vs GPS must agree within 5 NM. Without GPS: entry vs last recorded position within 1°, plus the gyro's sin/cos latitude check. The same wrong latitude entered twice → NAV IR FAULT.

[!note]- Q3. A tug pushes the aligning aircraft two metres. How much time do you lose, and must you re-enter position? EXCESS MOTION → 30 s later it reverts to a full alignment (9 min 30 s). Position need not be re-entered.

[!note]- Q4. What does a fast realign clear, keep, and still require? Clears the accumulated velocity error, keeps the previous attitude/heading, still requires a valid position — done NAV→OFF→NAV within 5 s, on the ground, below 20 kt.

[!note]- Q5. Name the two ingredients of the ND wind, and its two blank-out reason chains. GPIRS (or IR) ground-speed vector and the selected ADR TAS. Blank when TAS < 100 kt, and when CAS < 60 kt makes TAS itself NCD.

[!note]- Q6. The two sets of magnetic-variation latitude lines — when does each apply, and how do the consequences differ? Same table on all ADIRUs → cross the line and "TRUE must be selected." One ADIRU with a different table → the limits contract and flying beyond them is prohibited (the comparison monitor loses credibility).

Key takeaways

References

• FCOM LIM-NAV — ground-alignment latitude limits, dual magnetic-table clauses, TRUE requirement.
• FCOM DSC-34-10-10-20 — NORTH REF / TRUE / GRID display.
• AMM 34-14-00 — strapdown mechanisation, three-stage alignment, time formula, fast realign, NAV outputs, magnetic variation, position criteria, EXCESS MOTION, ATT mode, comparison cautions, removal criteria.

Independent study material, not an Airbus publication and not endorsed by the manufacturer. Always defer to the current operator FCOM, FCTM, and QRH for operational use.