The Go-Round Prompter

The Premise

In Air Safety Week 01 Nov 2004 a Take-off Performance  Monitoring System (TOPM) was discussed. This was an FMGS-tied system suggested to preclude the sort of take-off overrun accident that happened to the MK Airlines 747 freighter at Halifax on 14 Oct 04. It is thought possible that a fatigued crew may have re-used the take-off performance data card for their prior take-off ex Hartford or mis-set the derated power they (mis)calculated for the Halifax runway - and realized that far too late to retrieve the situation. It was postulated that mandatory data entry of the many take-off performance factors required would avoid mistakes and that the TOPM would real-time monitor the satisfactory progress of the take-off, avoiding any tail-strikes or panicky rotations in or beyond the overrun.

FAR 25.113 (Takeoff Distance and Takeoff Run) spells out with great precision the exact requirements for performance (screen heights, V1 etc) - however you should know that actual take-off performance "monitoring" in real time is a totally inexact science. That is why a take-off performance monitor would enhance safety by annunciating any acceleration discrepancy. But now the recent accident to an A340-300 at Toronto has highlit a similarly inexact situation in the approach and landing evolution. Is a Landing Performance Monitor technologically feasible? First, we shall examine the problem.

 “it's human nature to try and complete what you have started.” A very familiar jet accident is the overrun of a contaminated runway landing. Highly automated jets can now robotically handle pretty well everything but the take-off and the landing. Admittedly there is autoland - but that is restricted to the visibility limitations under relatively stable conditions of mist, fog and heavy drizzle.

 Autothrottle and ILS autoland are unable to cope adequately with unstable atmospheric conditions such as large gust factors and directionally variable strong winds found beneath a thunderstorm - or significant cross-winds. Autoland also requires a rated and categorised ILS equipped runway and a current and qualified crew. It also requires that multiple onboard systems be serviceable. So in point of fact, pilots are usually required to visually fly the last few hundred feet to touchdown and the landing roll-out. Unlike parametrically bounded systems, they are prone to errors of fatigue, indecision, uncertainty and illusions. In a 25 May 82 VASP 737 landing accident the probable cause was: "The pilot's misuse of rain repellent, causing an optical illusion". There can be potentially misleading visual cues when landing in severe precipitation, particularly at night. Heavy rain on the windscreen can change the refractive index significantly. For this reason one tendency is to flare high in these conditions which additionally might lead to excessive float prior to touchdown.

Add in a tailwind and, as with AF358 at Toronto, even though reasonably close to the correct threshold speed, that ensuing float can be very very wasteful of runway. What happens in those few moments between crossing the runway threshold and selecting reverse can make the difference between a successful arrival and a career in tatters.

You've been Set Up

Airline pilots very rarely carry out "bolters" (touch and go's). This has been recognized by the Flight Safety Foundation in a particular emphasis upon "Threat and Error" Crew management (TEM CRM). In short, sometimes a non-handling pilot can see something dire happening that the pilot flying (PF) will not. In some professions it's called task fixation. In times of stress it can become task saturation (the point at which situational awareness is superseded by adrenaline). A pilot is really only in control of his approach's "setup". Once he's crossed the threshold, he's wholly unaware of runway behind him and is looking fairly close in for his landing flare cues. He may be concentrating upon a smooth touchdown to the extent that he's blissfully ignorant of the rate at which he's running out of post-touchdown deceleration bitumen. In fact, once past the threshold, he's just not to know anyways accurately what roll-out distance remains - so his destiny is in his initial setup and thereafter rests wholly upon his early decision to press on or go-round. Many pilots just don't make such a conscious decision because they are (as yet) unaware of any impending peril. By the time fright arrives, flight has gone. Most runways only add the red runway lights for the last 1000 feet to the far threshold. That point is rather late for trying to do anything other than stand upon the brakes. Distance to run marker boards are usually found only on military bases. So what we are saying here is that a pilot doesn't need to commit gross errors to overrun, although admittedly many do. The difference between stopping and overrunning may lie in a few knots of headwind, the blanking lee side of a large hangar or whether the surface is bitumen or concrete. When wet, concrete whether grooved or not, is by far the poorer choice for effective braking. So illusions and other wild-cards aside, if we had to tabulate the factors within the pilot's control that determine the touchdown point and landing energy dissipation within remaining runway available, what would they be?

Threshold IAS (& Density Altitude) each extra knot will cause your landing distance to increase by about 2%.(gust factors added to Vref IAS will similarly increase roll-out). In fact the entry argument here would be groundspeed - to accommodate the varying effects of density altitude (landing at Mexico City on a hot day for instance).
Threshold Crossing Height VASIS or PAPI based nominally upon a 50 feet pilot's eye height for a touchdown 500 to 2000 ft in on a 9000ft runway
Aircraft Mass which of course generates inertia - which is the problem (within pilot control - to an extent)
Power Levers to idle Point Pilot assessment based upon his target threshold speed (whether auto-throttled or not)
Spoiler Actuation Controlled by "Weight on wheels" ground/air sensing circuitry
Flap setting Full (or something less than - for handling considerations)
Autobrake setting / anti-skid or manual braking (either style braking is compromised by a low friction surface that's awash with water, hail, slush etc)
Flare height too high a flare will extend the touchdown point (especially with a tailwind)
Reverse Selection Point / Amount and deselection point (and whether a reverser or two is locked out and unserviceable)
Brake fade (during late roll-out) with steel brakes (not really a factor with carbon brakes)

Other Factors

Here we've disregarded tire tread condition, runway contamination (Mu or frictional coefficient), aerodynamic braking (nosewheel placement), wind component, runway slope, rubber deposits at the far end, runway flooding or hail/snow, up elevator (forcing main-gear into the runway for better braking) etc..... and of course, pilot error. There may also be the psychology of knowing what the runway overruns are (ravine, ocean, cliff, gravel arrester bed, barrier or highway).

But obviously, although beyond a pilot's control, they are still valid entry arguments for factorizing achievable stopping distances as well as quantifying and minimizing risk. There can be controllability factors affecting directional control and consequently degrading effective braking. Consider the heaping of water on the upwind side of center-line due to runway camber. Auto or manual braking can be affected by the port and starboard main-gear having a different grip upon the ground. In manual braking that can lead to directional control difficulty (as can too much reverse with rear-mounted pod engines). Yes it is also possible to have one side aquaplaning and the other not. Reverted rubber skidding can affect roll-out distance. That's the interaction between a heavily braked tire and a rubber deposit.

Time is of the Essence

Let us quickly address uncertainty or cross-cockpit disagreement. What might happen if a captain decides his F/O has landed too deep? In the QANTAS 747 QF1 scenario at Bangkok, the F/O's instinct to go round was countermanded by his commander, the handling became suddenly complicated by habit patterns and configuration - and they overran. Time is always acutely of the essence, communication is sparse and one never gets a second chance to get it right. Most Company Standard Operating Procedures have the selection of reverse as the last-chance point for conversion to a touch and go. But what happens if the F/O momentarily advances the thrust levers to convert to touch and go and the skipper instantly takes over and retards them to reverse? Well for starters, the spoilers will retract and not redeploy once a throttle goes forward of idle (A340 FCOM Vol 1). That's sowed the seeds of an overrun because braking on a wet runway will now be severely degraded. In fact this scope for an "on runway" fiasco has led to many a routine landing becoming an excursion. Part of the solution is good clearways alongside runways and obstacle free overrun safety areas. But perhaps technology has something to offer also. Nobody wants to see an A380 sitting in a ravine looking very forlorn if there's an optional extra that will assist pilots in making early go-round or touch and go decisions. In fact, you may ask, why is it that they don't? Task fixation, saturation, fatigue, ego, getaboarditis, reluctance to become #6 in the hold again or to divert whilst low on fuel..... i.e. there are many reasons why a pilot may, in the heat of the moment, throw caution to the wind and try to retrieve his predicament. But as Réal Levasseur, TSB investigator for Air France 358 commented: "There's no way, having touched down 4000ft in, that they could've stopped on that 5000ft of wet runway remaining." So what precisely is the answer to avoiding such career challenges? Reverse isn't very effective, particularly at low speed, so perhaps aircraft just need more stopping power. Perhaps Boeing's new electric nosewheel motor might be tossed into our deceleratory melting pot?

The Answer Man

Don Bateman, father of the EGPWS (Enhanced Ground Proximity Warning System), took a basic GPWS (or systems-tied radar altimeter) and endowed it with the ability to know where it was at all times courtesy of SATNAV (GPS/GNSS). EGPWS can identify a terrain threat in all weathers courtesy of its terrain data-base. No EGPWS equipped airliner has yet flown into a hill. It's a true testament to what technology is capable of doing to assist a fallible pilot - as is TCAS (Traffic Collision Avoidance System). With EGPWS, as in real estate, it's location, location. With TCAS, the algorithms resolve relative movements of potentially conflicting aircraft. With LPM we'd need a system that can suck on all the relevant factors, meaningfully embrace some data-scatter or variability, resolve whether the energy dissipation challenge is feasible, apply a safety margin and then only tell us (audibly) NO (i.e. we just don't need it to say YES). If it was to err on the safe side, that would be no bad thing. Re-examine the tabulated and non-tabulated arrival factors discussed above. Determine whether a modern computer could factor those in and calculate the chances of your arrival going bad (i.e. better than you could by crossing your fingers, clenching your buttocks or landing with the toe-brakes on) - if it had a runway data-base.  Would such a system be worth avoiding an A380 or lesser investment sitting in a ravine with large loss of life? What would it be called?

IMPETUS -

  1. An impelling force; an impulse.
  2. The force or energy associated with a moving body.
    1. Something that incites; a stimulus.
    2. Increased activity in response to a stimulus: The approaching deadline gave impetus to the investigation.

The IMPETUS deadline is always that far end of the runway and the inky blackness beyond. Most pilots sooner or later get to see it approaching at a speed they never want to see again. Whether you're one of us, or yet to join the club, you will nevertheless understand the momentum behind IMPETUS.

from the Answer Man
 I enjoyed reading your excellent discussion on a "prompter".
 Everything you have said is well said. I also thank you for your kind remarks concerning the "Answer Man".
 
A comment on the "You've been Set Up" paragraph is that we have existing practical means to provide advisories for the remaining runway distance to the far threshold. Runway end coordinates are stored in the E-GPWS database and these continue to be precisely up-dated from E-GPWS flight history and satellite data and photos to WGS-84 geodetic reference. We use this database for providing the pilot visual and aural voice alerts / warnings to a flight path that would land the aircraft short of the runway end.  We use this database for additional safety functions.
 
 Many operators have added or are adding a software function, which we call Runway Awareness and Advisory System (RAAS I) to the existing E-GPWS. RAAS I can call out the remaining runway distance after the aircraft has passed half the total runway length, when the aircraft is less than 100 feet radio altitude above the runway and the ground speed still exceeds 40 KTS.  The aural voice calls out thousands of feet remaining (or in 300 meter increments). These advisories are to encourage a possible go-around decision.
 
We are developing Unstable (excess energy--"Too High! and or "Too Fast") Approach advisories at the 500 foot gate above the runway, along with an excess tail wind advisory as a software option to E-GPWS.
 
We owe a lot to President Clinton's order to remove "Selective Availability" from GPS.
 
We are also developing RAAS II (which gives an aural voice advisory when the aircraft is about to land or about to enter the runway of Traffic on the runway or on a short final to land).
 
I like what you wrote. Your article is refreshing and thought provoking.
 
Don Bateman.

don.bateman#honeywell.com
Comments upon Don Bateman's words of wisdom feedback (below):
 

"RAAS I can call out the remaining runway distance after the aircraft has passed half the total runway length, when the aircraft is less than 100 feet radio altitude above the runway and the ground speed still exceeds 40 KTS (???)."  Not sure that I can see this facility as being of much use "after the aircraft has passed half the total runway length". i.e. probably already into reverse and unable to safely get back into the air (i.e. go round) because of company SOP's - whilst being told by RAAS1 that you are likely to go off the end? A late go-round at that stage would be more likely to replicate the MK 747 at Halifax (a high-powered accelerating crash off a runway overrun). IMPETUS (as I see it) should be capable of an overall assessment of the factors at work as you pass over the landing threshold - i.e. not much, including the wind component, is likely to change significantly after that juncture....... yet the negative input info of IMPETUS would be very timely and ensure a safe go-round/touch and go transition.

They're obviously thinking about it at Honeywell - but we are apparently on a different wavelength. Under an IMPETUS regime, Jeppesen (if they ran the runway data-base) would rate runways for all their inbuilt characteristics but also their decrepitude (of having allowed rubber deposits to accumulate in the touchdown areas, no EMAS etc).


Link Subject
The FSF Approach & Landing Risk Reduction Kit
the Burbank SWA 737 Overrun ("The captain stated to Safety Board investigators that after touchdown, the end of the runway appeared to be closer than it should have been and that he thought they might hit the blast fence wall.")
Landing Long
VASP 737 accident  wet-runway-ops.html
Wet versus "Contaminated" Runway Ops

RAAS to the rescue?

The pilots of a Northwest Airlines A319 with 122 passengers who mistakenly landed June 19 on a U.S. Air Force base instead of their nearby destination commercial airfield were fired about a month later (see ASW, June 28). The two airfields, Ellsworth AFB and Rapid City airport, are about seven miles apart, the Rapid City field being described as "just over the hill" from Ellsworth and the crew was descending through clouds. Details of the case have been the subject of recent coverage by the Pioneer Press newspaper of St. Paul, Minn. The paper obtained copies of the air traffic control tapes, in which surprised Air Force watch standers remarked that the base commander was going to be "a little upset" over the unexpected landing of the commercial jetliner on the B-1 bomber base.

The two pilots were replaced by another crew, who flew the A319 the few miles to its correct destination about three hours after the inadvertent landing.

The incident pilots have filed a grievance against Northwest for terminating their employment.

There are a number of interesting aspects of the case:

  • Was this incident the culmination of precursor events? There are reports of several previous instances of pilot confusion between Ellsworth AFB and Rapid City Airport.
  • How'd the crew lose situational awareness? Descending toward Rapid City, the flight crew was flying a VOR (very high frequency omnidirectional range) approach, intended to guide them to the general vicinity of the runway. Apparently, when they broke out of the clouds, they sighted what they believed was Rapid City's Runway 32, pushed the nose over (suggesting an unstabilized approach) and mistakenly landed on Ellsworth's Runway 31. With all the high-tech navigation equipment with which their A319 was equipped, including GPS and moving map displays, the crew still landed on the wrong runway, at the wrong airport.
  • What did the air traffic controller advise? In cases of airports in close proximity, aren't controllers supposed to say, "Northwest 319 Ellsworth Air Force Base is at your 11:00, Rapid City Regional is at your 12:00. Report Rapid City Regional in sight." At least, that is the kind of guidance that should have been radioed, according to Federal Aviation Administration procedures (FAAO 7110.65), Chapter 7, Section 4 on visual approaches, subparagraph 7-4-3g, which says, "In those instances where airports are located in close proximity, also provide the location of the airport that may cause the confusion."

The hypothetical question that comes to mind is whether the incident pilots would have been sufficiently alerted if the Honeywell [HON] Runway Awareness Advisory System (RAAS) had been fitted to their airplane. In addition to its pre-takeoff and after-landing advisory messages, the system is designed to aurally advise pilots of the runway on which they are about to land when the airplane is some three nautical miles distant and within + or - 20 degrees of the runway centerline (see ASW, June 16, 2003). A typical voice advisory would be, "Approaching two-five." In cases where two runways are within + or - 20 degrees heading of each other, then the message "Approaching runways" is generated.

In this case, the pilots mistakenly landed on Ellsworth's runway 31. Their intended touchdown was at Rapid City's runway 32. According to Honeywell officials, their systems database for RAAS and the Enhanced Ground Proximity Warning System (EGPWS) is "enabled" for Rapid City. However, the database for Ellsworth is enabled for EGPWS only. The incident airplane was equipped with EGPWS.

What this all means is that even if the incident airplane had been equipped with RAAS, it would not have issued an advisory that the pilots were approaching Ellsworth's runway 31. They would have heard the standard altitude callouts from EGPWS, which might have subtly reinforced their belief that they were landing at Rapid City. (For more on RAAS and EGPWS, see http://www.egpws.com/raas/raas.html)

Approaching Runway - RAAS In Air Advisories

RAAS equipped aircraft provide the flight crew with an aural advisory when the aircraft is airborne and approaching a runway. This advisory is enabled when:

  • The aircraft is between 750 feet and 300 feet above the airport field elevation (AFE), and
  • The aircraft is within approximately three nautical miles of the runway, and
  • The aircraft track is aligned with the runway within + or - 20 degrees, and
  • The aircraft position is within approximately 250 feet, plus runway width, of the runway centerline.

All EGPWS aurals have priority over this RAAS advisory. The "Approaching Runway - In Air Advisory" is suppressed between 550 feet and 450 feet above runway elevation to allow normal 500-foot altitude call outs and/or crew procedures without conflict. If the advisory is triggered while the aircraft is between 550 and 450 feet above field elevation (AFE), the advisory is suppressed until the aircraft descends below 450 feet AFE, where the message will be annunciated.

If the criteria above are not satisfied before the aircraft descends below 300 feet AFE, the advisory is aborted.

Source: Honeywell, RAAS product specification.

from Air Safety Week of 16 Oct 2004

Real time take-off performance monitoring (as suggested by an ASW contributing editor from the International Aviation Safety Association, with acknowledgement to the collective ruminations of the Pprune Pilots’ Forum):

“Real-time acceleration analysis during each and every departure roll might just be the 21st Century
way to solve the problem, even if airport weight bridges or other passive controls are added as a further
means to assure load limits.
“Many factors can influence the ability of an airplane to lift off the runway in the time and distance
available: weight, airfield density altitude, runway slope, humidity, thrust, tire friction, winds, turbulence,
runway surface contamination, etc. Some are fixed, many are variable, yet all are potentially conclusive
factors in a maximized load marginal operation. When calculating what’s possible, some factors (such as
wind component) won’t be known exactly. Few will be controllable once the take-off roll begins. The
choices mid-roll devolve to power, airfoils (quickly selecting more flap), and reject/reverse/braking.
“Departure performance is pre-calculated with an expected weight, thrust, slope, winds, temps, etc.
The calculation, as done manually, creates a few numbers such as MTOW and V-speeds. The calculations
also will determine whether a reduced power take-off is possible (with increased engine life). But the same
information can just as easily create a computerized moment-by-moment profile, expressed as speed versus
distance, or acceleration vs. time, or roll distance vs. time from the application of thrust.
“Having derived this computed curve of expected performance at hand, it is the most natural thing in
the world for a little bit of electronics to compute a moment-by-moment ‘actual performance’ during the
take-off roll from groundspeed, GPS, wheel spin-up, acceleration forces, or all of the above. By
continuously comparing the projected performance with the observed values, the resulting real-time
performance quotient would provide a validated comparison of expected versus actual progress in getting
airborne within the runway available. Any significant departure from expected norms would be alerted in a
timely manner, rather than, at present, suddenly confronting the pilot with a huge quandary – to firewall it
and go, or to try and stop. Such snap decisions by their nature have a 50 percent probability of being
wrong. Actually, in a misconfigured marginal take-off, the calculated last point of abort will also be in
error, so any such decision may have a 100 percent chance of being wrong (cannot get airborne within
runway remaining but also cannot stop within the abort distance available).
“This progressive roll data should be presented to the pilot in a readily interpretable form. It can be a
single white LCD digit in a separate display that will illuminate after a standard trigger distance (perhaps
after 3000ft of wheel-spin). If it shows zero, you are ‘on the money.’ If it indicates +1(green) you are
lighter than you thought, it’s more humid or the temperature has suddenly dropped. If it shows -1 (amber)
you are heavier etc.; -2 (amber) and you are worried (it has your attention), drops to -5 (red/flashing) and
the decision that ‘discretion is the better part of…’ is already made for you and it’s made in time – and so
you reject early.
“What this decision aid would do is add technological logic to what a pilot must instinctively do
now: monitor the progress of the takeoff run and determine whether it ‘feels’ right or not. The difference is
that it would add quantifiable precision to a seat-of-the-pants process. It would work just as well in cold
dark places, in rain and snow and other cases where the crew’s sensory perceptions may be constrained or
illusions present (such as humps in the runway).
“Getting something wrong now and again is human nature. Allowing that error to create a calamitous
accident chain is bad risk management. Such a ‘Take-off Tattle-tale’ could easily be integrated into present
aircraft electronic systems. <<Air Safety Week of 01 Nov 2004>>