atc-discussion Hogmanay (EGPH & EGPF) [30th Dec] - Debrief

[Harry Sugden]

Summary of Lessons

• Edinburgh is always the busier airfield and should have FIN rostered (or at least on standby) from about an hour into events (or when the arrival rate is predicted to pick up). Glasgow does not need FIN rostered for dual-airport events.
• AIR and FIN coordination needs work, but this is not unique to this event. It may be worthwhile to produce guidance on what kind of constant conversations need to be had during events. The positions tend to sit in different coordination rooms which isn’t ideal, and muting on TeamSpeak prevents more effective communication.
• The APC controllers at Edinburgh did a fantastic job, given the traffic levels. IMPRESSIVE! @Fergus Walsh @Thomas Greer @Reece Buckley
• Holding is not fair without a ‘grand plan’ or a way of keeping track of the delay across different inbound routes. It may be worth discussing whether this is something that matters on VATSIM, and thus whether it is worth taking some (relatively simple) steps to resolve this.
• LAC North is worth rostering for any ScTMA ‘overload’ events - they played an essential role this event in regulating the flow to ScAC.

Traffic Levels

	Edinburgh (EGPH)		Glasgow (EGPF)
Time (Zulu)	Departure Rate (per hour)	Arrival Rate (per hour)	Departure Rate (per hour)	Arrival Rate (per hour)
1900-2000	1	12	1	5
2000-2100	8	24 😮	2	6
2100-2200	6	21	2	4
2200-2300	9	21	1	4
TOTAL	24	78	6	19

 

Aerodrome / Approach

The Edinburgh controllers did a fantastic job at handling the level of traffic that they did. With no rapid exit taxiways, handling a high number of departures and arrivals at Edinburgh is a difficult challenge - yet for three hours, at least 20 arrivals were consistently handled.

😓 We were lucky that the demand for departures at Edinburgh wasn’t stronger. Some thought that coordination between TWR and FIN could have improved to help get the departures that did show up with minimum delay, but this is a common feature of events.

The Edinburgh INT/FIN split was only rostered 1945-2115, but was required until close. It worked well given the extremely high pressure from the top in terms of having two enroute holds waiting to be emptied! There were some occasions where aircraft who were slow descending in the TARTN hold prevented a bunch of aircraft from dropping down. In these cases, coordination by INT with Galloway to bring the aircraft toward STIRA so they could descend on their way to the north side / in the STIRA hold may have been useful. This is as much ScTMA’s responsibility to notice as it is APC, however.

Streaming before holding began could apparently have been a little better, with 5 NM on a few occasions (especially when on consecutive aircraft) leaving APC with little room to play with. But this is the responsibility of all controllers down the chain to ensure aircraft are sufficiently in trail, and the merging of 3 routes inbound to TARTN does sometimes lead to unavoidable < 10 NM streaming to APC.

😴 Glasgow FIN was rostered - it was not (and is never) needed in events with both Edinburgh and Glasgow. If anything, a standby could be rostered that jumps on for 10 minutes here and there if required if there are any unmanageable waves of traffic.

 

Area Control

Sectorisation, Frequency Congestion & Coordination

Having 7 controllers was just the right amount for this event and enabled a reasonably optimal sectorisation.

❗ The original plan was to split Deancross from ScAC Bandbox at 2015Z to catch the expected main wave of inbounds from the South. A controller dropout meant that the planned Deancross controller took all of the Scottish upper sectors, which quickly became unworkable. Handling a strong stream via INPIP and AGPED to Edinburgh, as well as being required to provide top-down at Inverness and Aberdeen led to a temporarily degraded service north of the TMA. Luckily, despite the drop-out, a quick reshuffle enabled ScAC East to be split off from Bandbox, so each ScAC controller became responsible for one of the enroute holds.

✔️ ScTMA was split at the right time - there were very few departures this event, so Galloway was consistently underworked, but it was still important for them to take some workload off the Talla controller. Talla had an extremely busy frequency from opening as the TARTN hold filled up well above the published FL140 max (highest was FL210 at one point). To avoid overloading the sector, a stop on all further traffic entering the sector was put in place until this level could be brought down.

 ✔️ Antrim was also very useful to have on throughout, handling a total of 19 flights from Dublin/Belfast/Derry inbound to Edinburgh.

🤩  ✔️ Finally, LAC North was an essential addition to the roster. Without them, ScAC Bandbox’s job would’ve been made one hundred times harder before the split. ‘Overload’ style events in the ScTMA should always have AC North rostered.

Generally, all controllers reported frequency congestion of around 3, with peaks of 4 (out of 5, being ‘overworked’). This is with the exception of the ScAC Bandbox shift (before splitting), which was too busy. All controllers reported having sufficient time to coordinate. There was no coordinator for this event.

Spacing & Holding

Once the peak began, aircraft were generally having to hold for about 20 minutes in the TARTN hold. But with holding in place, AC North ( @Matthew Burton ), ScAC Bandbox ( @Daniel Button ) and ScAC East ( @Trevor Hannant ) did a fantastic job with in trail streaming of inbounds to Edinburgh. The picture below was taken during the event - thanks @Cathal Boyce 😄 Even when there is both terminal and enroute holding, having aircraft in trail is essential to avoid overloading the next controller and increasing complexity any further. There were no complaints from anyone who gave feedback related to enroute streaming.  ✔️ 

Holds at INPIP and AGPED were established as soon as Talla reached capacity and had to put a stop on all further inbounds into the sector. There were unpublished holds, but were in good positions and managed well by their respective controllers to hold the traffic. Reflecting on my own Talla controlling, there was too much vectoring of traffic between INPIP/AGPED and TARTN, given the traffic was going to re-enter a hold anyway. This increased the workload further unnecessarily, but was my attempt at accounting for the long distance between INPIP/AGPED and TARTN when calling aircraft off the stack. ❗

Miles-in-trail requirements were used between Talla and Bandbox/East to empty the aircraft from the enroute holds, adjusted depending on how full TARTN was. These worked well, and were also adhered to brilliantly by ScAC.  ✔️ 

 ❗ However:

• Aircraft via INPIP with a cruise level below FL255 avoided the INPIP enroute hold and were put straight into TARTN. This enabled them to effectively skip the queue.
• Aircraft managed to avoid any significant holding by coming via Galloway and then being routed onto STIRA. One aircraft did a spin at MAC, but otherwise these aircraft experienced no delay.

❓ 🤔 Neither of these loopholes are ideal, but they are also somewhat unavoidable without a coordinator. It may have made sense to re-route aircraft planned via AGPED round to the north of Edinburgh and into the STIRA stack, but then aircraft in INPIP would still have held longer than these STIRA aircraft. The answer could lie in recording at what time an aircraft starts holding as soon as the inbound delay is above around 20 minutes, potentially using the SPad. This could then be used to determine the order in which they then enter the terminal hold. 

If aircraft are routed via STIRA, then APC should theoretically continue to empty planes from TARTN (and by extension, INPIP) with an earlier holding start time, thus ensuring STIRA planes get a fair share of the delay. (With AGPED -> STIRA re-routes, the ‘start time’ could be set as the point they are taken off route for the trip round). It does appear as if some other locations use this method - start times for holds have been recorded in the FP remarks for Zurich’s extremely popular Tuesday Nights in the past. This may be something to explore.

There were 3 diversions to Glasgow from INPIP/AGPED towards the end of the event. By this point, the delay at TARTN actually only turned out to be about 5-10 minutes!