A320 (2) / CRJX (2) / B738 (3) / A332, vicinity Madrid Barajas Spain, 2018
A320 (2) / CRJX (2) / B738 (3) / A332, vicinity Madrid Barajas Spain, 2018
On 27 May 2018, four losses of separation on final approach during use of dependent parallel landing runways occurred within 30 minutes at Madrid following a non-scheduled weather-induced runway configuration change. This continuing situation was then resolved by reverting to a single landing runway. The Investigation attributed these events to “the complex operational situation” which had prevailed following a delayed decision to change runway configuration after seven consecutive go-arounds in 10 minutes using the previous standard runway configuration. The absence of sufficient present weather information for the wider Madrid area to adequately inform ATC tactical strategy was assessed as contributory.
On 27 May 2018, after a series of seven go-arounds in 10 minutes at Madrid due to an excessive tailwind component on the ‘preferred configuration’ parallel landing runways in use, it was decided to change runway use to one providing landing runways more suited to the prevailing wind conditions. This change was then followed by 4 losses of separation between 8 different aircraft during final approach in day VMC each involving aircraft failing to meet the prescribed minimum of 2nm radar separation when vertically separated by less than 1000 feet and on approach to dependent parallel runways. All the aircraft involved were operating scheduled passenger flights.
An Investigation was carried out by the Spanish Commission for the Investigation of Accidents and Incidents (CIAIAC). All the required information was obtained from ATC and Meteorological sources and no recorded flight data was taken from the aircraft involved.
The event involved four separate losses of separation whilst runways 18L and 18R were both in use as designated landing runways (‘South Configuration’) following an earlier unscheduled change from operations with runways in ‘North configuration’ (landing on runways 32L and 32R). This change followed seven go arounds in northerly runway configuration due to unfavourable wind conditions (tailwind component and/or wind shear) in a period of 10 minutes.
The aircraft involved, all operating scheduled passenger flights into Madrid, were as follows:
- Event 1
- An Airbus A320 (CS-TNQ) being operated by TAP inbound and from Lisbon under callsign TAP1018
- A Bombardier CRJ 1000 (EC-LJS) being operated by Air Nostrum for Iberia and inbound from Alicante under callsign ANE99DJ
- Event 2
- A Boeing 737-800 (EI-FZR) being operated by Ryanair and inbound from Birmingham under callsign RYR290B
- A Bombardier CRJ 1000 (EC-LJR) being operated by Air Nostrum and inbound from Turin as ANE8825
- Event 3
- An Airbus A320 (EC-LYM) being operated by Iberia Express and inbound from Birmingham as IBS36BK
- A Boeing 737-800 (EI-FTY) being operated by Ryanair and inbound from Dublin as RYR54AH
- Event 4
- A Boeing 737-800 (EC-LXV) being operated by Air Europa and inbound from Lanzarote as AEA5BN
- An Airbus A330-200 (EC-LVL) being operated by Air Europa and inbound from Milan Malpensa as AEA7EY
All the events involved a failure to ensure that the separation minima for dependent parallel ILS approaches to runways 18L and 18R were met. In the absence of 1000 feet vertical separation, these required a minimum of 3nm horizontally between two aircraft when one is turning onto each ILS LLZ. Also, once established on the ILS LLZ, each aircraft was required to be a minimum of 3 nm from any aircraft ahead on the same LLZ and 2nm away from an aircraft making an approach on the adjacent LLZ. No evidence was found that flight crew non-compliance with ATC instructions had contributed to the occurrence of any of the losses of separation.
It was noted that the Madrid TAF (aerodrome forecast) current for the time of both the go-arounds prior to runway re-configuration and the losses of separation after it, had included as a PROB40 TEMPO (temporary) of variable wind directions at 15-27 knots with visibility 4000 metres in a thunderstorm with rain showers and SCT (scattered) CB at 2,500 feet aal. The METAR current just prior to the time of the 7 go-arounds from approaches to runways 32L and 32R gave the actual wind velocity as from 300° varying 270°-330° at a mean speed of 11 knots with a thunderstorm in the vicinity. After three of these go-arounds, it was decided to immediately change runway configuration so that landings would be on runways to 18L and 18R. Between this decision being taken and the ATIS being updated to advise the change of landing runways, four more consecutive go-arounds occurred. All these go-arounds were due to one or both of excessive tailwind or wind shear. By the time a new METAR was issued just after the seventh go-around had been initiated, the surface wind was from 120° at 7 knots gusting to 16 knots. The almost 180° change in wind direction was attributed to a gust front associated with a thunderstorm to the east of the airport.
- Event 1 occurred just over half an hour after the runway configuration change had been made and involved the CRJ 1000 on the 18L ILS remaining less than 2nm clear of the A320 behind it on 18R ILS for the final 12 nm with separation steadily reducing to reach 0.8 nm just before touchdown.
- Event 2 involved the CRJ 1000 on 18L ILS being less than 2 nm clear of the A320 behind it on the 18R ILS for the final 12 nm as it approached touchdown, with separation steadily reducing to 1.2 nm just before touchdown.
- Event 3 involved the A320, already established on the ILS 18R being less than 3nm ahead of the 737-800 as it joined the 18L ILS and once it was established at around 8nm, less than 2nm in front, this separation then steadily reducing to 0.9nm just before touchdown.
- Event 4 involved the 737-800 on the 18R ILS being less than 2nm behind the A330-200 on the 18L ILS from 11 nm prior to touchdown and then steadily reducing to less than 1nm shortly before touchdown.
The Investigation considered various factors which were likely to have collectively contributed to the losses of final approach separation, noting that the decision to change runway configuration without prior notice was made by the TWR Supervisor at the airport in response to the situation there without explicit regard to the ad hoc complexity which the change would create in the TMA.
It was noted that the TWR Supervisor had begun considering a runway configuration change as soon as the sudden surface wind change occurred and aircraft on approach began reporting an increased tailwind component, wind shear and the presence of storm east of the airport. However, after discussing it with the ACC Supervisor, he had decided to “wait about five minutes in case the wind direction changed again” during which time the sequence of consecutive go-arounds began. He then spoke with the forecaster who advised that the situation “involved convective currents of variable direction and that he estimated they would dissipate in one hour”. Nine minutes after the first go around had occurred, the decision to change the runway configuration was made. It was considered that the TWR Supervisor had had insufficient weather information to make a timelier configuration change decision. It was also considered that such a decision should be taken and timed in conjunction with the ACC Supervisor since there would be an unavoidable workload increase for the TMA sectors affected.
In respect of airport weather information, it was noted that although a SPECI report had been issued for the airport 25 minutes prior to the wind direction change that led to the go-arounds (which occurred between the issue of two routine METARs), “a SPECI is not issued for a sudden change in wind speed or direction”. It was also noted that the ICAO Annex 3 requirement for the validity of a TAF or METAR was only that it should cover a radius of 5 miles from the aerodrome reference point (ARP). It was noted that the State weather agency had now acknowledged that “other type of products, not defined in Annex 3 of the ICAO Convention, are needed”.
A previous similar event and the ANSP response to it
It was noted that a related convective weather event at Madrid on 29 April 2016 had previously been investigated by the CIAIAC. The published Final Report (published only in Spanish) had not made any Safety Recommendations because the measures considered by the Commission for the Study and Analysis of Notifications of Air Traffic Incidents (CEANITA) and taken by the organisations involved in the event “were deemed appropriate”. However, it was found that the recommendations made by the CEANITA had not been fully progressed and that the intent of some of them had been met by analysis and ‘debate’ which had not led to action which fully addressed the ATC issues that are liable to arise in relation to episodes of convective weather like the one which had now been investigated.
The Cause of the losses of separation was determined as “the complex operational situation in the airspace of the Madrid TMA”.
Three Contributory Factors were also identified as follows:
- The unavailability of weather information tools that cover the entire airspace of the Madrid TMA.
- The sudden change in weather conditions, and specifically, in the wind speed and direction.
- The time needed to make the decision to change the runway configuration.
Eight Safety Recommendations were made as a result of the Investigation as follows:
- that ENAIRE (ANSP) collaborate with AEMET (Spanish National Weather Agency) to improve the weather forecast available at the airport control tower and at the airport. [REC 16/19]
- that AEMET collaborate with ENAIRE to improve the availability of information on forecast weather conditions information at both the Madrid-Barajas control tower and at the Madrid-Barajas weather office. [REC 17/19]
- that ENAIRE revise the procedure in place for changing configurations and involve the supervisor of the Madrid TMA in this decision. [REC 18/19]
- that ENAIRE train its air traffic controllers on the procedure for swapping sectors when changing configurations. [REC 19/19]
- that ENAIRE specify the stormy weather procedure taking into account the operational and geographical features of Madrid-Barajas airport. [REC 20/19]
- that ENAIRE continue collaborating with AEMET to prepare weather measurement and forecasting products that cover the entire volume of the Madrid TMA airspace. [REC 21/19]
- that AEMET continue collaborating with ENAIRE to prepare weather measurement and forecasting products that cover the entire volume of the Madrid TMA airspace. [REC 22/19]
- that ENAIRE improve how weather data are displayed on the screen such that they span the entirety of the Madrid TMA airspace. [REC 23/19]
The Final Report of the Investigation was approved on 18 December 2019 and released in English translation on 1 April 2020, the definitive Spanish language version having been released on 5 February 2020.