ART-2025 • Module 7 • Case Study: Delhi Airprox

Story / What Happened at Delhi Narrative

“Normal” parallel ops turned abnormal instantly HF / SA Case

Morning at Delhi IGI Airport, 10 Nov 2023. Hazy ~2000 m, heavy traffic. RWY 29R was used for arrivals, RWY 29L for departures. These parallels are dependent, meaning they are not fully separated like independent parallels.

Vistara A320 (VT-TQL) was on very short final to 29R. Around 100 ft AGL, the crew got a WINDSHEAR WARNING and immediately initiated a go-around.

Almost at the same time, Ethiopian B787 (ET-ATL) was cleared for take-off from 29L. Tower expected that the departure and the go-around would remain safely apart.

Parallel Runway Geometry (Simplified)

RWY 29R (Arrivals): Vistara on short final → “GO-AROUND” → starts climbing runway heading

RWY 29L (Departures): Ethiopian rolling / airborne → climbs out

Both are now climbing in almost the same airspace near the airport, in haze.

Their vertical / lateral spacing began collapsing within seconds. Both crews received a TCAS Resolution Advisory (RA) to avoid collision. That was RA #1.

After they followed TCAS instructions, ATC started giving headings and climb profiles. But those headings later caused the two aircraft paths to converge again in the terminal area. A second TCAS RA (RA #2) was triggered before things finally stabilised.

Key observation

This was not a single “near miss.” It produced two TCAS RAs within about two minutes.
The first RA was go-around vs. departure. The second RA happened after initial vectors.
Dependent parallels are not “independent”; assumption of safe divergence was wrong.

Minute-by-Minute Timeline Sequence

How fast it went critical Seconds Matter

~T0: Vistara A320 on final RWY 29R (arrival).
~T0+30s: Windshear alert → immediate go-around announced.
~T0+35s: Ethiopian B787 cleared for take-off RWY 29L (departure).
~T0+60s: Both aircraft climbing. Spacing compresses rapidly.
~T0+70s: TCAS RA #1 — collision avoidance commands issued.
~T0+120s: Tower / Approach issue headings, altitudes, SIDs to resequence them.
~T0+130s: Headings and climb levels start to converge again in terminal airspace.
~T0+140s: TCAS RA #2 — second RA to prevent conflict.
~T0+180s: Stable separation finally achieved. Situation normalises.

Single decision point

Go-around is called on short final.

Tower must instantly consider BOTH runways “linked” and unsafe for any takeoff.

If departure is not stopped immediately, you now have two heavy jets climbing together in haze.

AAIB India documented this as one serious incident with two successive TCAS RAs, not two separate events.

Why Did We Get Two TCAS RAs? Root Cause

RA #1 vs RA #2 Separation Loss

TCAS RA #1 — Initial climb conflict

Vistara: go-around off 29R, climbing out.
Ethiopian: departure off 29L, climbing out.
Both jets are basically in the same terminal airspace volume at the same time.
Tower assumed “natural divergence,” but did not enforce it.

TCAS RA #2 — Re-convergence after first RA

After RA #1, ATC issued headings / SIDs to resequence.
But vertical “who occupies which level” was not locked down with a hard block.
Those headings then caused the two aircraft to come together again in the TMA.
That triggered a second RA before stable spacing was fully built.

Human factor meaning

We partly fixed the first conflict, but we did not fully rebuild protected separation before we “went back to normal vectoring.” Normal vectoring too early caused the second RA.

Human Factors / Situational Awareness Why SA Failed

SA breakdown in Tower / TMA HF Focus

Task saturation: Tower/APP was already near workload limit: haze, high movement rate, parallel runways.
Channelised attention: Focus locked onto the go-around aircraft (Vistara), while the departure (Ethiopian) became “just routine”.
Expectation bias: Subconscious belief “these two will naturally separate” instead of issuing a hard stop / divergence instruction.
Team SA gap: No immediate “STOP IMMEDIATELY [(repeat aircraft call sign) STOP IMMEDIATELY];” call .
Coordination lag: Approach did not cordinate / did altitude blocks and divergent headings before resuming standard headings. That allowed the second convergence.

SA Breakdown Pyramid

Level 1 — Perception: I HEARD “go-around”, I SEE a departure.

Level 2 — Comprehension: ⚠ Only partial. Did we truly accept that both are now climbing in the same airspace?

Level 3 — Projection: ✖ Failed. We did not predict “In 10 seconds these two will merge.”

ATC Mental Check During Abnormal

What just changed?
Who is now in potential conflict?
Did I STOP anyone else from moving?
Did I rebuild real separation, or am I just hoping?

VEGT / Guwahati Action Points Local Application

What we apply locally Operational Practice

VEGT Immediate Practice COMPLETE

Any go-around = runway / climb path is LOCKED.
Treat it as occupied airspace. No new departure until you assign who climbs and who stays clear.

Anyone can say STOP.
“STOP IMMEDIATELY [(repeat aircraft call sign) STOP IMMEDIATELY];.” do remember 4NM rule.

give traffic information cordinate with approach.
Make sure both read back.

VEGT Single-Runway / Parallel-Like Teaching Flow

GO-AROUND / WINDSHEAR / ABORTED LANDING

→ “STOP IMMEDIATELY [(repeat aircraft call sign) STOP IMMEDIATELY];

→ Assign which aircraft climbs / which levels with cordiantion with approach (altitude + heading blocks)

→ Only resume next movement after positive separation is confirmed

The Delhi event teaches us: TCAS is not our plan. TCAS is the last layer. We (Tower / APP) are supposed to prevent the situation where TCAS even needs to speak.