Visual Separation

According to FAA Order JO 7110.65AA - Air Traffic Control, paragraph 7.2.1 VISUAL SEPARATION, in a TERMINAL area visual separation can be either Tower-applied or Pilot-applied.  Pilot-applied visual separation requires the controller to:

(a)   Maintain communication with at least one of the aircraft involved and ensure there is an ability to communicate with the other aircraft.

(b)   The pilot sees another aircraft and is instructed to maintain visual separation from the aircraft as follows:

(1)   Tell the pilot about the other aircraft. Include position, direction, type, and, unless it is obvious, the other aircraft's intention.

(2)   Obtain acknowledgment from the pilot that the other aircraft is in sight.

(3)   Instruct the pilot to maintain visual separation from that aircraft.

(c)   If the pilot reports the traffic in sight and will maintain visual separation from it (the pilot must state both), the controller may “approve” the operation instead of restating the instructions.

(d)   If aircraft are on converging courses, inform the other aircraft of the traffic and that visual separation is being applied.

(e)   Advise the pilots if the targets appear likely to merge.

(f)    Control of aircraft maintaining visual separation may be transferred to an adjacent position/sector/facility. Coordination procedures must be specified in an LOA or facility directive.

Up to letter (c), the conditions were met in the case of the PAT25 helicopter and the CDA Local controller on the night of 29th January 2025; the controller was not required to inform the CRJ that there was traffic ahead and that the other aircraft would maintain visual separation; in this case, the CRJ could hear the controller’s instructions to the helicopter on another frequency but could not hear PAT25’s responses. Since the rule did not require PAT25 to read back the controller’s instructions, neither the controller nor the crew of the CRJ were aware that parts of the instructions were not heard by PAT25.

However, letters (d) and (e) contradict the notion that once visual separation has been established by the pilot of one of the two aircraft and approved by the controller, the controller is relieved of all responsibility for maintaining safe separation between the two aircraft. These rules require the controller to continue to observe the two aircraft on his screen, that he will warn the other aircraft if they are on converging courses, and that he will advise the pilots of both aircraft if the two targets appear likely to merge. Note that the rules do not mention vertical separation.

In the case of the Washington DC accident, the CRJ and PAT25 were on converging courses and finally merged on the radar screens, otherwise they would not have collided. Evidently, the controller believed that the conflict would be resolved by vertical separation but, as observed in a previous Blog, a vertical separation of less than 200 ft is not a safe separation. Had the Local controller expressly informed the CRJ that it would cross the path of a helicopter flying south along the river at 200 ft, the crew would have been made aware that a potentially dangerous situation was developing, would have been on the lookout for the helicopter and could have taken other actions, such as abandoning the approach or evasive manoeuvres if they saw the helicopter.

“Visual separation” is dependent on the “see and avoid” principle; since “see and avoid” cannot be relied upon to maintain safe separation between an airliner and a General Aviation or Military aircraft, even in daylight with good visibility, it follows that permitting a General Aviation or Military aircraft to unilaterally maintain visual separation from an airliner is not a safe practice and should be discontinued.

Where airliners are concerned, visual separation should only be permitted between two airliners following one another, in daylight in good visibility, when there is no possibility of misidentification.

See and Avoid

The principle of “see and avoid” has been a necessity since the early days of aviation and it is still relevant in operations at uncontrolled airports. However, it has not prevented accidents involving airliners and it should not be relied upon to maintain safe separation iof airliners n controlled airspace.

A typical failure of the “see and avoid” principle occurred in 1956, when a TWA Lockheed Super Constellation and a United Airlines DC-7 collided at 21 000 ft above the Grand Canyon in Arizona, killing all 128 passengers and crew aboard the two aircraft. The Civil Aeronautics Board (CAB) accident investigation report, states:

“The present concept for separation of aircraft and avoidance of collision in VFR weather conditions, regardless of flight plan or clearance, depends on the flight crews' ability to visually provide separation between aircraft. Civil Air Regulations expressly place this responsibility on the pilots and the concept is commonly referred to as the "see and be seen" principle.

“The Board determines that the probable cause of this mid-air collision is that the pilots did not see each other in time to avoid the collision.”

Other examples of failure of "see and avoid" concerning airliners are the 1979 collision between a Pacific Southwest Boeing 727 and a Cessna 172 near San Diego (144 deaths) and the 1986 collision between an Aeromexico DC-9 and a Piper PA-28 in the Los Angeles Terminal Control Area (TCA) resulting in 82 deaths.

These accidents, in which the NTSB cited the failure of visual separation and “see and avoid” in controlled airspace, led to more stringent rules for General Aviation aircraft operating around major airports and the present classification of airspace in the USA.

Also in 1986, a collision occurred between a Twin Otter and a helicopter over the Grand Canyon National Park. In addition to the failure of “see and avoid” and relevant to the 2025 Washington DC accident, in the report the NTSB said:

“Also contributing to the accident was the modification and configuration of the routes of the rotary-wing operators resulting in their intersecting with the routes of Grand Canyon Airlines near Crystal Springs.”

Another risk factor which must be considered is the presence of military aircraft operating in airspace used by civilian airliners. The Aviation Safety Network database lists 58 midair collisions involving airliners with more than 3300 fatalities since 1954; 18 of these accidents involved a military aircraft.

We will now examine a possible reason why “see and avoid” failed in the case of the 2025 Washington DC accident:

At 20:46:01, when the CDA Local controller called PAT25 and advised them that a CRJ was at 1200 feet just south of the Woodrow Wilson Bridge, circling to runway 33, the helicopter was about 0.7 nm south of the (Arlington) Memorial Bridge and about to cross over the George Mason Memorial Bridge. At this point the helicopter was north of the airport and practically lined up with runway 01-19 so that the landing lights of any aircraft on approach to runway 01 would appear very bright. However, at this time, the CRJ had begun a slight turn to the right of the runway 01 centreline in order to initiate the circling visual approach to runway 33.

Behind the CRJ were two more airliners which the controller cleared to land on runway 01. To someone lined up with runway 01-19, the landing lights of these aircraft would appear brighter than those of an aircraft which was now pointing in another direction. According to the NTSB, the word “circling” was not recorded on the helicopter’s CVR; it is therefore possible that the crew of PAT25 misidentified the CRJ for the aircraft behind it and thereafter visually followed the path of this aircraft as it continued towards runway 01. As the helicopter continued flying south towards Hains Point, the aircraft lined up to land on runway 01 would appear on the right hand side of the helicopter, while the CRJ would be on its left. If the crew of the helicopter were wearing night vision goggles, these would restrict their field of vision, making it difficult to see the CRJ if they were looking to the right.

The above explanation underscores the difficulty (not to say the impossibility) of making a positive identification of another aircraft at night.

This tragic accident, which took the lives of 60 passengers and seven crew, dramatically demonstrates once again that “see and avoid” cannot be relied upon to maintain safe separation between airliners or between an airliner and a General Aviation or Military aircraft, even in daylight with good visibility.

In controlled airspace, the responsibility for maintaining safe separation between aircraft corresponds to the respective controller and should not be delegated to a pilot.

The next Blog discuses “visual separation” which in turn depends on the “see and avoid” principle

15th of February 2025

Update on the 29th of January 2025 midair collision between an American Eagle CRJ-701ER, and a US Army Sikorsky UH-60 Black Hawk helicopter, callsign PAT25.

NTSB news conference on 14th of February 2025 based on preliminary information downloaded from the CVR/FDR from the helicopter.

American Airlines Flight 5342 took off from Wichita Airport, Kansas (ICT) at 17:38 local time (23:38 UTC). Prior to descent, the flight crew briefed the expected arrival procedure (TRUPS FIVE) for an ILS approach to runway 01 at DCA. Descent began at 20:15 ET.

The Black Hawk helicopter, callsign PAT25, was on a combined annual and night vision goggle check-ride for the pilot flying (PF) and the crew was probably wearing night vision goggles throughout the flight.

At 20:30 local time (01:30 UTC), PAT25 began flying north to south over the Potomac River after manoeuvring near Laytonsville, MD. Cockpit voice recorder (CVR) data indicates that the instructor pilot was pilot monitoring (PM) and the check-ride pilot was pilot flying (PF).

At 20:33:41, PAT25 requested the Reagan National Airport (DCA) Tower for helicopter Routes 1 to 4 to Davison Army Airfield (KDAA), which was approved by the controller.

The Washington DC helicopter routes designated by the FAA stipulate maximum altitudes and indicated routes but not lateral boundaries. For the 6.3 nm sector of Routes 1 and 4 between Memorial Bridge and the Woodrow Wilson Bridge, the maximum permitted altitude is 200 ft. Opposite the approach path for aircraft landing on runway 33, Route 4 is depicted over the East side of the Potomac River.  

At 20:38:39, PAT25 reached the intersection of the DC Beltway and the Potomac River near Carderock, MD, began descending and picked-up Route 1 over the river towards downtown Washington, DC.

At 20:39:10, the Potomac Approach controller cleared Flight 5342 for the Mount Vernon visual approach to runway 01. Four minutes later the crew contacted DCA Tower. The controller asked if the crew could switch to runway 33. After a brief discussion between the crew, they agreed to runway 33.

At 20:43:48, PAT25 was 1.1 nautical miles (nm) west of Key Bridge. At this point, the PF had an indicated altitude of 300 ft, while the PM had an indicated altitude of 400 ft. Neither pilot commented on the discrepancy. In this sector of Route 1 the maximum permitted altitude is 700 ft.

At 20:44:41, the CRJ selected flaps 30, then flaps 40. Eight seconds later, 6.2 nm south of the airport, the landing gear was down and locked, with the aircraft fully configured for landing.

At 20:45:27, the CRJ was 5 nm south of the airport, the autopilot was disconnected and the aircraft began a shallow right turn off the path of the runway 01 localizer at a radio altitude of 1700 ft and with an airspeed of 134 knots.

At 20:45:30, PAT25 passed over the Memorial Bridge. The PM told the PF they were at 300 ft and needed to descend, with the PF responding they would descend to 200 ft.

At 20:46:01, the Tower controller called PAT25, advising them that a CRJ was at 1200 feet just south of the Woodrow Wilson Bridge, circling to runway 33. This communication was audible in the CRJ, but the portion stating the traffic was "circling" may not have been heard by the crew of PAT25.

At 20:46:08, PAT25 responded that they had the traffic in sight and requested visual separation, which was approved by the controller.

At 20:46:29, the crew of the CRJ received a 1000 ft automated call-out.

At 20:47:26, the CRJ began to roll left to line-up with runway 33 at 133 knots and two seconds later the crew received a 500 ft automated call-out. According to ADSB data, the airliner would have been about 1.2 nm from the runway 33 threshold at this point.

At 20:47:27, PAT25 passed the southern tip of Hains Point, about 0.9 nm from the point where helicopter Route 4 crosses the extended centreline of runway 33.

At 20:47:39, DCA Tower called PAT25 asking them if they had the CRJ in sight. This was audible in both CVRs. A separation conflict alert was audible in the background of the Tower's transmission.

At 20:47:40, the crew of the CRJ received an automated Traffic Advisory call-out: “Traffic, Traffic”.

At 20:47:42, the Tower instructed PAT25 to pass behind the CRJ. This was audible in the CRJ’s CVR but the portion of the transmission stating "pass behind the" may not have been heard by the PAT25 crew because their microphone was keyed for 0.8 seconds, stepping on the Tower's transmission.

At 20:47:44, PAT25 stated they had the traffic in sight and requested visual separation, which was again approved by the controller. The Pilot Monitoring (PM) told the Pilot Flying (PF) that he believed ATC wanted them to move left, towards the East bank of the Potomac River.

At 20:47:58, the CRJ started increasing its pitch after a verbal reaction from the crew, with almost full deflection of the elevators. The collision occurred one second later, at 20:47:59.

While the Tower transmitted simultaneously on two VHF radio frequencies, PAT25 and the CRJ were communicating separately in one of each of those frequencies. This meant they could both hear the Tower, and vice-versa, but the two aircraft could not hear each other's transmissions.

The last radio altitude recorded by the CRJ, two seconds before impact, was 313 ft, and its descent rate was 448 ft/min. The pitch of the CRJ reached 9° nose-up after the crew's reaction, as well as a roll to 11° left wing down.

The last radio altitude recorded on the PAT25 FDR was 278 ft, steady for the previous 5 seconds. The pitch of the helicopter at the time of collision was about 0.5° nose-up with 1.6° of left roll.

The NTSB provided an explanation of the altitudes used by the aircraft crews and the radio altimeter altitudes recorded in the respective flight data recorders (FDRs). Flight crews normally rely on the barometric altitude displayed on altimeters installed on both sides of the cockpit. The displayed barometric altitude depends on the atmospheric pressure sensed by the static ports on the side of the aircraft and, when flying below 18 000 ft, on the reference atmospheric pressure (QNH) for the airport or region; the QNH is provided by ATC and must be manually set by the crew in each altimeter window. Barometric altimeters are very precise instruments but atmospheric pressure varies with time and weather conditions; an incorrect QNH value may result in an error of hundreds of feet. Neither the displayed barometric altitudes nor the QNH are recorded in the FDRs. The FDRs record pressure altitudes and, at low altitudes, radio altitudes. The pressure altitude is determined by the aircraft’s systems based on a standard atmosphere model and the NTSB has established that the pressure altitude recorded by the PAT25 helicopter was inconsistent (“bad data”). This pressure altitude is also transmitted by the aircraft transponder and displayed on the ATC radar screens, but as a “Flight Level” in hundreds of feet. Thus, if an aircraft has a pressure altitude between 101 ft and 200 ft, the controller will only see FL 002.

The radio altimeters bounce a signal off the surface of the earth and give a precise reading of the height of the aircraft’s belly above the ground, in this case the surface of the water of the Potomac River, which is essentially sea level. The radio altimeter value is also displayed in the cockpit, but although pilots may use it as a reference, their primary flight instruments are the barometric altimeters.

The NTSB also confirmed that PAT25 was not broadcasting ADSB data, which contains more information than the standard transponder, but it is not known at this time if this was due to a failure or because it was switched off.

Based on this preliminary but official information, the following can be concluded:

• The CRJ was on a normal flightpath and glidepath for a visual approach to runway 33, correctly lined-up and configured to land when the collision occurred at an altitude of about 300 ft.

• The visual approach to runway 33 (and the corresponding published RNAV (GPS) instrument approach) requires flying over helicopter Route 4 at an altitude below 400 ft. This theoretical vertical separation of less than 200 ft is NOT a safe separation even under ideal visual flight conditions (VMC) and should never have been permitted.

• When it collided with the CRJ, PAT25 was near the middle of the river channel when it should have been hugging the East bank of the river and it was nearly 100 ft above its maximum permitted altitude of 200 feet.

• A separation conflict alert was triggered in the DCA Tower about 19 seconds before the collision; this sounded an audible alarm and would also have been displayed on the controller’s radar screen.

• The CRJ crew received a Traffic Advisory (TA) from its TCAS about 18 seconds before the collision; a Resolution Advisory (RA), which would have commanded the crew to ascend, was not issued because these are inhibited below 1000 ft.