Air Safety

Introduction

The exercise was a 2v1 intercepts with tactics as part of a staff continuation programme covering basic fighter manoeuvres, air combat tactics and beyond visual range tactics. The exercise was preceded by a mass briefing on 2v1 tactics covering mission profiles, tactical considerations, rules of combat, no contact and breakaway minima and other safety aspects. This was followed by an individual mission briefing which again covered rules of combat and no contact/breakaway minima, as well as ‘domestics’, height blocks and specific tactics and procedures.

The mission callsign was Maple Hunter and consisted of Hunter 1, 2 and 3. Hunters 1 and 2 were allocated the 1 to 4 height blocks and Hunter 3 the 6 to 9 height blocks for the intercept phases and had individual controllers and GCI frequencies. A visual identification was to be obtained before any simulated weapons launch. The weather at the time was fine with no significant reports. Although visibility was excellent looking down, it was only poor looking up and at an angle.

The Accident

The initial unsuccessful engagement occurred with Hunters 1 and 2 heading about 100° in the FL210 to 240 height block and Hunter 3 heading about 290° in the FL360 to 390 height block. After the merge, Hunter 3 continued west, before turning left onto what was about east. Hunters 1 and 2 cross turned left and right respectively onto about west, with Hunter 2 rolling out in Hunter 1’s eight o’clock at two miles. Hunters 1 and 2 and Hunter 3 were split by about 10 miles during these manoeuvres. Approaching the merge again, Hunter 3 continued a left turn towards north, and Hunter 1 gained visual contact on Hunter 3 slightly left and high. Hunter 1 called the sighting and pulled up to engage. Hunter 2 also sighted Hunter 3 and pulled up to engage.

Hunter 3, turning left through north, sighted Hunter 2 and initiated a descending left turn. Hunter 3 did not have contact with Hunter 1. Hunter 1 attempted an unsuccessful Vertac lock on Hunter 3, and then attempted a boresight lock for a front hemisphere AIM-9 shot, at about 50° nose up, and at an estimated range of two miles. Significantly, Hunter 1 assumed Hunter 3 was reacting to him, but Hunter 3 had not seen Hunter 1 and was reacting to Hunter 2. Hunter 1’s attempted boresight lock was unsuccessful and he initiated a breakaway approaching one mile by rolling left to a slightly right wing down inverted position. During this manoeuvre Hunter 1 lost sight of Hunter 3 below his nose, but believing Hunter 3’s flight path to be predictable (incorrectly believing Hunter 3 had seen him), he assessed Hunter 3 would pass below him on his right side and well clear. Hunter 1, still believing Hunter 3’s flight path to be well clear, started a 2. 5 to 3 ‘g’ pulldown to move into a possible defensive position for Hunter 3. At the same time, Hunter 3 rolled out of his descending left turn and started to pull up.

About three seconds after losing sight of Hunter 3, Hunter 1 felt a shudder and assumed he had flown through Hunter 3’s wash. Hunter 3 felt a very mild thump, followed by the ‘Flaps Off Caution’ illuminating, with the right flap crossed out on the flight control display. Hunter 3 then noticed extensive damage to his right wing and realised a collision had occurred. The aircraft were at 27,700 feet.

Recovery

After the collision, Hunter 3 came onto the same frequency as Hunters 1 and 2 and informed them of the collision. Hunter 1 then discovered that the rear panel of his right LAU-7 launcher was missing. All three Hornets tracked towards home. Hunter 2 performed a visual inspection on Hunter 1 and then joined on Hunter 3, the more extensively damaged aircraft, to perform a visual inspection and escort him in the event of a possible ejection.

Both damaged Hornets performed controllability checks in the landing configuration. Hunter 1 had no handling problems; Hunter 3 experienced onset of yaw and roll at 150 knots. Following discussion with squadron operations, the decision was made to recover the least damaged aircraft (Hunter 1) first by a normal landing, leaving the approach and hook cable free for Hunter 3. Hunter 1 made an uneventful straight in approach and landing with half flap.

Hunter 3 then made an approach at 160 . knots (following the flaps off caution checklist), reducing to 155 knots at the threshold and made an uneventfularrested landing. Both aircraft had suffered Category 3 damage.

Crew aspects

The pilot of Hunter 1 was a Category D QFI on Hornets. He had almost 3,000 hours, including nearly 100 hours on Hornets. The pilot of Hunter 3 was a Category B FCI on Hornets. He also had nearly 3,000 hours, including over 800 hours on Hornets. Both pilots were considered qualified and current for the flight and were medically classified A1G1Z1 (extremely medically fit). There was nothing remarkable in either pilot’s history which could have any bearing on the accident.

Accident causes

The primary cause was the failure of Hunter 1 to maintain visual contact with Hunter 3 during the breakaway. The contributing causes were a wrong assumption by Hunter 1 that Hunter 3 was in visual contact with him and Hunter 1’s wrong assumption that Hunter 3’s flight path was predictable and would give safe separation. The final contributing cause was that Hunter 3 did not have visual contact with Hunter 1.

Although Hunter 1 had infringed the ASI breakaway minima for a front hemisphere missile attack, this was considered not to have contributed to the accident, and that all reasonable steps had been taken to avoid the accident. The poor visibility looking up in conjunction with the grey camouflage scheme made visual range estimation difficult, particularly in a head on situation at a high closing speed.

Conclusions

It was concluded that airspeed limits on intercept and air combat training were basically sound, but that further refinements were needed for front hemisphere tactics. The problem is that modern fighter aircraft engaged in air combat tactics are operating very close to the limits at which the human brain, nervous system and muscles can respond to visual stimuli within the necessary time. The judgement required of pilots of distances, closing speed of often over a thousand knots and rapidly varying aspects, when coupled with human reaction time, leaves little room for error.

The rules of combat and breakaway minima must allow for the risk of human error, while at the same time permitting realistic operational training. To do so, the breakaway minima must have some margin for error, while the rules of combat must stress that the breakaway minima are exactly that- minima which are not to be infringed. The breakaway minima assume ideal conditions of visibility and preparedness for initiation of breakaway. Very high closing speeds, less than ideal visual conditions and stress factors such as workload, may dictate termination of an attack at ranges greater than the specified minimum in order to assure safe separation. Ultimately, collision avoidance depends on the pilot’s judgement to break away early enough.

The factors which determine a safe but practical breakaway distance are the risk of collision, the missile envelope and the IR acquisition envelope. For visual head on missile attacks, the collision risk is the principal factor. The ASI (airspeed) limits for visual head on missile attacks within the remaining portion of the front hemisphere as well as the definitions were considered to be adequate. However, the safe separation minima was later changed from 600′ horizontally and 200′ vertically, to a 500′ sphere, except for low closure rate manoeuvres such as horizontal sissors, where the limits remained at 600′ horizontally and 200′ vertically.

Finally, the following points should be stressed:

• During a front sector visual attack with missiles, the target must be displaced immediately after lock on to ensure lateral and/or vertical separation develops.

• Pilots of attacking aircraft must maintain visual contact during the breakaway manoeuvre until no possibility of confliction exists.

• Pilots must always assume the aircraft they are engaging has not sighted them, and must assume responsibility for maintaining safe separation.

ACCIDENT REPORT: McDonnell Douglas MD-80 and Piper PA28, Cerritos, 31 August 1986

This article is taken from the NTSB report on this accident. The lessons for us are in the difficulty to see both small and large aircraft in good visibility and the fallibility of radar and its human operators. The Los Angeles basin is one of the most dense flight traffic areas in the world. Apart from Los Angeles International (LAX) Airport there are four other major airports, a number of military airfields and many light aircraft airports. In this congested airspace, aircraft on IFR and VFR flight plans operate aircraft of vastly differing size and speed, with some on full ATC reporting and some not. In the summer months, visibility is often very poor due to smog in the lower levels.

Synopsis

At about 1140 local time a Piper PA28 Cherokee departed Torrance, just south of LAX, on a VFR flight to Big Bear, about 60 miles to the east and at about 6,000 feet in the mountains. At 1120 the Aeromexico MD-80 departed Tijuana for LAX. These aircraft collided over Cerritos, a suburban area of Los Angeles, at about 1152, both aircraft falling to the ground. The three occupants of the PA28 were killed in the collision; 58 passengers and six crew on the MD-80 were killed by ground impact forces and 15 were killed on the ground.

The NTSB determined that the probable cause of the accident was the limitations of the ATC control system to provide collision protection. Factors contributing were the inadvertent and unauthorized entry of the PA28 into the Los Angeles Terminal Control Area (TCA) and the limitations of the see and avoid concept to ensure traffic separation.

History

The PA28 route was planned to keep it well south and east of LAX climbing to 9,500 feet. The PA28 pilot did not activate his flight plan, nor was he required to. The transponder was set correctly to the VFR code though does not give height information. The pilot, flying from the left seat, was familiar with the area. He had two passengers, one seated in the front right seat.

The MD-80 was on an IFR plan and was cleared to approach from the south at 7,000 feet. The aircraft was then cleared by approach control’s arrival radar controller to reduce to 190 knots and descend to 6,000 feet. Just over one minute later, the two aircraft collided at about 6,500 feet. Both pilots of the MD-80 were familiar with the area and the captain was highly experienced.

The controller was extremely busy at the time of the accident. He was controlling several other aircraft and, during the period between clearing the MD-80 to 6,000 feet and the accident, had a conflict to sort out. A Grumman Tiger called, requesting a clearance, and the controller informed the pilot he was in the middle of the TCA and suggested that ‘in the future you look at your TCA chart. You just had an aircraft pass right off your left above you at 5,000 feet and we run a lot of jets through there right at thirty five hundred’. He then noticed the radar no longer tracking the MD-80 and was unable to establish radio contact.

All aircrew and ATC controllers involved held current ratings and were in good health. As far as could be determined, both aircraft were serviceable up to the time of impact. Weather was clear, visibility to 14 miles, with no cloud. The sun was at nearly 62° elevation to the south east, putting it high in the sky, behind the MD-80 and about the two o’clock position on the PA28.

The Accident

From the radar recording, the MD-80 Flight Data Recorder (FDR) and the PA28 altimeter witness marks from the mid-air impact, the collision occurred as the MD-80 was descending through 6,560 feet. It was tracking north west and the PA28 was tracking east, crossing the MD-80 track from left to right. The collision damage on the MD-80 was confined to the vertical and horizontal stabilizer, the horizontal stabilizer separating from the MD-80.

The engine of the PA28 struck the left side of the MD-80 vertical stabilizer. The left horizontal stabilizer shearing the roof off the PA28 cabin, the damage indicating the PA28 was in an almost wings level attitude at impact.

Ground Radar Data

The arrival radar controller stated that the PA28″ . . . was not displayed. It is my belief that he was not on my radar scope”. He considered his workload to be ‘light’. Though he was busy at the time checking two different scopes for conflicting traffic for two flights and taking a change of details for the MD-80 from the arrival co-ordinator.

Seven months after the accident a DC-9 and a Cessna passed within 1,000′ to 600′ of each other almost over the accident site at a similar altitude and under similar conditions. The DC-9 crew reported the near miss as it was happening. The controller rechecked his radar display which did not depict the Cessna. He then had another controller confirm the Cessna was not depicted.

Despite the absence of a return on the smaller aircraft in each incident, when the radar recording was played back both the PA28 and Cessna were depicted. Numerous tests were conducted in an attempt to resolve this anomaly. A satisfactory solution was not reached. The report stated in part: ” . . . the Board concludes that the positions of the Piper airplane were depicted on the arrival radar controller’s display by, at the least, an alphanumeric triangle, but that the controller did not observe the Piper’s radar target”.

The Board considered the following three factors, either individually or in combination, to have been the cause of the controller not seeing the PA28.

”The possible distraction of his attention from the critical area of his radar display caused by the projected entry of the Grumman Tiger into the TCA and the change of landing runways for the Aeromexico MD-80; the possibility that the controller may have unintentionally discounted the non-mode CVFR radar return of the Piper as a threat because it was located within the lateral confines of the TCA; and the possibility that the primary radar return of the Piper either did not appear on his display or the strength of the return was compromised by atmospheric interference”.

The arrival radar controller testified that, with regard to TCA intrusions, the number varied and “it could be anywhere from zero to 1 O or 15 a shift that I will observe”.

Visibility and Vision Studies

A visibility study was conducted to determine the physical limitations to visibility from each of the pilot ·seats. The time histories of both aircraft flight paths and attitudes were combined with binocular photographs of the cockpits. The binocular photographs are taken by a camera with two lenses, the spacing between the lenses being equal to the average distance between human eyes. The viewing angles for each aircraft were then calculated and plotted at five second intervals -in relation to the design eye reference (DER) points for’each windshield.

The study showed that for about the last 90 seconds of controlled flight, the PA28 was about 15° to 300 left of the DER point on the captain’s windshield and the first officer’s windshield. This meant that the PA28 was primarily in an area where the captain could see it with both eyes and, for the first officer, it was in the MD-80 centre windshield and in an area, for about 50% of the time, where he could see it with both eyes.

For the PA28 pilot, the MD-80 was about 50° to the right of the DER point and could only be seen by him on the far right side of the copilot’s windshield. For the passenger in the right seat of the PA28, the MD-80 was about 55° right of the DER point and would have appeared at the left edge of the right side window.

Previous studies by the Massachusetts Institute of Technology had been conducted to determine target acquisition parameters for pilots. Sixty four tests were conducted with pilots who did not know the test objectives. Intercepting aircraft were flown head on, 90° and 135°, predominantly from the left. Visual acquisition was achieved in 36 attempts (56%) and the median range was 0. 99 miles. The greatest acquisition range was 2. 9 miles.

A further 66 tests were conducted in a TCA with the pilots aware of the test objectives. Visual acquisition was obtained in 57 attempts (86%) and the median range was 1. 4 miles.

Using these tests and pertinent data from the Cerritos mid-air, an acquisition graph was constructed. These graphs do not allow for limitations such as cockpit structure or the monocular regions. The NTSB probability of visual acquisition graphs indicate the PA28 pilot had an 80% probability of seeing the MD-80 15 seconds before the collision and the airline pilots a 30% probability. These graphs attempt to allow for aircraft size, relative positions, closure rates and other factors.

Regardless of this, however, both aircraft were operating in visual flight conditions and were required by FAA regulations to see and avoid each other, even though the MD-80 was under radar control in a TCA and on an IFR flight plan. The failure to see and avoid must be considered in relation to the limitations of angles of closure, target sizes, conspicuity of target and the physiological capabilities of the eye.

Of particular interest, the probability of acquisition graphs Indicate that had the controller seen the PA28 and been able to provide altitude information as well as range and bearing to the MD-80, the MD-80 crew’s probability of acquisition would have risen from 30% to 95%. In any case, some avoiding action would have obviously been initiated.

NTSB conclusions

The major NTSB findings were that:

a. Both pilots wore required to see and avoid. There is no evidence that either pilot attempted an evasion.

b. The PA28 pilot was not cleared to enter the TCA but his entry was inadvertent.

c. Both aircraft were displayed on the arrival controller’s radar but that the PA28 primary target may not have been displayed or displayed weakly.

d. The arrival controller did not see the PA28 radar return. e. The Los Angeles area was not equipped with an automated conflict alert system. (Board tests had indicated that such a system would have alerted the controller).

The NTSB probable cause was the limitations of the ATC system to provide collision protection. Contributing factors were the inadvertent and unauthorised entry of the PA28 into the TCA and the limitations of the see and avoid concept.

Relevance to Australia

Although this accident occurred in a high traffic density area, the traffic level was only light on the day and could easily equate to many radar controlled areas within Australia. Of importance is the possible limitations of the radar system and its operators. It is possible the radar may miss a target or may not show the target on a scope. It is possible, for various reasons, that the controller may not notice a target on the scope or may, consciously or subconsciously, ignore the target.

Aircrew lookout is the only present method available to overcome the operational limitations of radar. A Caribou crew operating in the Bankstown area on a Sunday will have all available eyes outside. Will the same crew be as vigilant 10 miles away on a visual approach into Sydney or 1,000 miles away approaching Ceduna? The Cerritos mid-air not only shows the limitations of radar but also the limitations of the eye. Lookout and a good scan technique are the only real answers. Keep vigilant in areas where you least expect other traffic; that is, in controlled airspace and the back of beyond. Back this vigilance up with a sound knowledge of where you are in relation to likely busy spots – small airfields, glider areas, hang glider jump-off spots, etc – and listen for other traffic on the radio.

The only good that comes out of an horrific accident like this mid-air is the lessons learnt to prevent a recurrence. Apply the lessons to yourself and stay alert and aware.