Airports — turnaround & on-time operations

The problem. An aircraft on the ground earns nothing, and every minute past its slot ripples outward — a late turnaround misses the departure wave, the next rotation slips, and by evening the delay has spread across the network. A turnaround is a relay of independent teams — deboarding, cleaning, fuelling, catering, loading, boarding — and if any leg runs long, the whole thing does. Operations needs to see a turnaround falling behind while it can still be recovered, not after pushback is missed.

What we solve here is delay propagation: catching the slow leg of a turnaround in time to throw resources at it.

Set up demo data

New workspace? Run this once (Python) to create the stand's turnaround feed, create the subscription before we listen, and push a leg-overrun reading. Safe to re-run.

import datahub_sdk, pandas as pd

client = datahub_sdk.DataHubClient.from_env()

client.timeseries.create([datahub_sdk.TimeSeries(external_id="stand_b12_fuelling_progress", name="Stand B12 fuelling", unit="pct", value_type="float")])
client.subscriptions.create([datahub_sdk.Subscription(
    external_id="stand_b12_turnaround", name="Stand B12 turnaround", timeseries=["stand_b12_fuelling_progress"])])
client.timeseries.insert_from_lists(timestamps=[pd.Timestamp.now(tz="UTC")], values=[1.0], ts="stand_b12_fuelling_progress")

1. Track each turnaround milestone live

Milestone timestamps and progress stream in per stand. When a leg overruns its planned duration, raise a turnaround_risk event so the duty manager reallocates before the slot is lost. See Consume live data.

Java

import ai.intellistream.datahub.sdk.subscriptions.SubscriptionMessage;

try (var stream = client.subscriptions().listen(List.of("stand_b12_turnaround"))
        .stream((SubscriptionMessage msg) -> {   // auto-acks after each message
            if (legOverrunning(msg.payload())) {          // e.g. fuelling past plan
                EventModel risk = new EventModel();
                risk.setExternalId("turnaround_risk_su204_" + System.currentTimeMillis());
                risk.setType("turnaround_risk");
                risk.setStatus("open");
                risk.setMetadata(Map.of("flight", "flight_su204", "leg", "fuelling", "stand", "stand_b12"));
                risk.setEventTime(ZonedDateTime.now());
                client.events().create(List.of(risk));
            }
        })) {
    awaitShutdown();   // your app lifecycle; closing the stream ends delivery
}

Python

with client.subscriptions.listen(["stand_b12_turnaround"]) as listener:
    for msg in listener:
        if leg_overrunning(msg.payload):              # e.g. fuelling past plan
            client.events.create([datahub_sdk.Event(
                external_id=f"turnaround_risk_su204_{int(pd.Timestamp.now().timestamp())}",
                type="turnaround_risk", status="open",
                event_time=pd.Timestamp.now(tz="UTC"),
                metadata={"flight": "flight_su204", "leg": "fuelling", "stand": "stand_b12"})])
        listener.ack([msg.message_id])

Rust

use dataplatform_rust_sdk::events::Event;
use chrono::Utc;

let mut listener = api.subscriptions.listen(&["stand_b12_turnaround"]).await?;
while let Some(Ok(msg)) = listener.next().await {
    if leg_overrunning(&msg.payload) {                // e.g. fuelling past plan
        let mut risk = Event::new(format!("turnaround_risk_su204_{}", Utc::now().timestamp()));
        risk.r#type = Some("turnaround_risk".into());
        risk.status = Some("open".into());
        risk.add_metadata("flight".into(), "flight_su204".into());
        risk.add_metadata("leg".into(), "fuelling".into());
        risk.add_metadata("stand".into(), "stand_b12".into());
        risk.set_event_time(Utc::now());
        api.events.create(&vec![risk]).await?;
    }
    listener.ack(&[msg.message_id.as_str()]).await?;
}

2. Trace the knock-on through shared aircraft

One late arrival delays the next flight on the same airframe, and the one after that. Model flights and the aircraft and crews they share as a graph; walking out from the delayed flight reveals the downstream rotations at risk — the blast radius of the delay — so re-timing decisions are made with the whole chain in view.

3. Learn from the day

On-time performance and per-leg durations are series; daily roll-ups by stand and ground-handler surface the legs that chronically run long. See Query & aggregate.

See the result

The overrunning leg trips the loop:

turnaround_risk_su204_…  → open   (fuelling running past plan, slot at risk)

See also

• Consume live data — live milestone tracking.
• Correlate alarms with the graph — delay knock-on across rotations.
• Query & aggregate — on-time-performance reporting.