FSF: Go-around Decision-making and Execution Project March 2017

[Don Hudson]

The Flight Safety Foundation has published a 54-page document / study on go-arounds. It's well worth downloading and reading, mostly for pilots but I think many here would find it interesting and educational. Our ATC and F/A members I'm sure would be interested.

From the introduction:

Quote

3.1 The Go-Around Noncompliance Problem Approach and landing is the most common phase of flight for aviation accidents, accounting annually for approximately 65 percent of all accidents. A Flight Safety Foundation study of 16 years of runway excursions determined that 83 percent could have been avoided with a decision to go around. In other words, 54 percent of all accidents could potentially be prevented by going around. It is generally felt that an unstable approach is the primary cause of landing excursions. However, within this 16-year period, just over half of the landing excursions followed a fully stable approach; in these instances, the flight became unstable only during landing. A critical industry policy designed to help prevent such accidents is the go-around policy. Interestingly, and sadly, the collective industry performance of complying with go-around policies is extremely poor — approximately 3 percent of unstable approaches result in go-around policy compliance. Why is a critical policy designed to prevent the most common type of accident ignored by flight crews, and why is that policy not being managed effectively by management? Improving the go-around compliance rate holds tremendous potential in reducing approach and landing accidents (ALAs). The go-around itself is not without risk. There is an increased risk in loss of control events during a go-around compared to exposure with all other phases of flight. It follows that we should only go around when the risk associated with an unstable approach is greater than the risk associated with a go-around. What is that balance, and how do we minimize go-around exposure to only those that are really necessary? The Flight Safety Foundation Go-Around Decision-Making and Execution Project was launched to research and answer the question “Why are we so poor at complying with established go-around policies?” It was also intended to improve our understanding of the risks associated with executing go-arounds and to make recommendations to improve compliance and mitigate risks associated with the go-around maneuver itself.

For those with FDA/FDM/FOQA programs, the study encourages use of data of examine the quality of the go-around for positive feedback to crews when flown well, and for trends where go-around pitch, flap & gear targets show as not quite being met.

The Report acknowledges that the go-around itself does carry some risk. Some interesting statements are made in the Report regarding when to go around, and I'll leave it up to others to find them and possibly discuss - there are some points made that will not be uncontroversial in my view.

https://flightsafety.org/toolkits-resources/go-around-project-final-report/ - FSF Project website
https://flightsafety.org/wp-content/uploads/2017/03/Go-around-study_final.pdf - PDF Document

[Moon The Loon]

Thanks Don. Have forwarded final study to my Flt Ops Dept.

Long time, no "see"! How ya doin'?

[Don Hudson]

Hi Moon - doing great. lurking and reading - focused on flight data work mainly and grandchildren - two-and-a-half, and 4 months!  - delightful. Sending the doc to ops people is great - there's some very valuable information and thinking in the study. There's also some things that I think warrant discussion!

 

[Say Again, Over!]

Hi Don,

I'll include this in the reference material for our ATC course.

Quick question, though, in light of the statistics quoted: Since over half of the runway excursions followed an otherwise stable approach could it be argued that the problem therefore does not lie with unstable approaches but rather with poor go-around decision skill?  Have we been looking in the wrong place all along?

I know there is a lot of pressure on operators to come in fully stabilized and we're seeing a difference of late with the "more manoeuvrable" aircraft coming in slower and farther out.  Could this have an effect on the philosophy?

Looking forward to the answer and discussion.

 

Felix

[Don Hudson]

Hi Felix - yes, I'm really looking forward to a discussion on this. With statistics like those in the article including the low-compliance stats, we have the opportunity to substantially reduce ALA's, (approach-landing accidents).

Just to quote the stats being referred to:

Quote

3.1 The Go-Around Noncompliance Problem
Approach and landing is the most common phase of flight for aviation accidents, accounting annually for approximately 65 percent of all accidents. A Flight Safety Foundation study of 16 years of runway excursions determined that 83 percent could have been avoided with a decision to go around. In other words, 54 percent of all accidents could potentially be prevented by going around. It is generally felt that an unstable approach is the primary cause of landing excursions. However, within this 16-year period, just over half of the landing excursions followed a fully stable approach; in these instances, the flight became unstable only during landing.

Re, " Have we been looking in the wrong place all along? "

Yes, I think so.

I've done flight data work for a long time and that question always nagged me...I saw lots of unstable approaches and uneventful landings and saw some "interesting" outcomes to very pretty right-on-the-numbers approaches. It begged the question: If unstable approaches were a problem, where were the incidents and accidents?

So it had to be the wrong metric (the stable-approach standard), and the wrong question to ask.

The Report recognizes the very low rate of compliance among professional, highly-qualified, experienced crews. Why is this the case?

I think both our own experience, and perhaps the FDA/FDM data may tell us why: 99% of landings were/are successful, with no apparent (to FDA), elevation of risk. So the traditional "mandate" to go-around if unstable falls away to experience of long histories of successful "re-stabilization" of the approach. In short, there isn't a "red light"...it could be the amber light all the way down to 500ft, (and in not-rare cases, lower), and a successful landing is the outcome.

That doesn't mean we should downplay the importance of the stabilized approach and the mandated go-around of course. But it's one of the questions that I think is really (really!) worthwhile discussing, mainly because the actual rate of compliance must improve AND the pilots must have reasonable, rational mandates which accommodate an approach which was a mess to begin with but "eventually" became stabilized. That means that the word, "eventually" must be defined, and I think this Report tries to do that.

That said, this isn't final and the Report acknowledges that this is an evolving shift in requirements.

To me, the most important parts of the Report are the first section which defines the issues, and the last sections where the stabilized approach criteria are redefined with a very rich discussion and points made.

I think there are some controversial statements as well, and I think they should be challenged for thought and discussion. Example:

Quote

5. Touch down just beyond the touchdown aim point following a normal flare, and not beyond the touch down zone (TDZ). If not touched down within the TDZ (or revised touchdown point limit) — go around.¹
6. Touch down on the runway centerline with the main landing gear on both sides of (straddling) the runway centerline. If all main landing gear are on one side of the centerline — go around.

^1. Touchdown aim point is defined by the U.S. Federal Aviation Administration as 1,000 ft from the runway threshold. The International Civil Aviation Organization defines touchdown aim point in reference to the available landing area, as follows:
Table            Available landing area        < 800 m           800–1,200 m         1,200–2,400 m           > 2,400 m
                      Touchdown aim point              150 m           250 m                      300 m                              400 m

To me anyway, the Report is saying that if you're not on the runway by 400m, you should go-around.

In my view, except on 5000ft runways and shorter, particularly contaminated runways, a go-around from post-touchdown is a higher risk maneuver for an airline crew than staying on the ground and using heavy braking, but that is just an opinion for discussion and is not based in anything or any data.

The proposal is to create a new point at 300ft for the decision-making process. I think the Report explains the rationale quite well but again, discussion is required.

As I say, well worth reading, discussing and even studying.

[J.O.]

Thanks Don, I look forward to reading through it thoroughly. If it's as well thought out as their ALAR tool (wow, it's almost 20 years already!), it should help the industry to take the next steps to a lower accident rate.

Speaking of unstable approaches, have you seen the TSB report into the Jazz undershoot at CYAM? I'd be curious to hear your thoughts on their attempts to apply swept wing jet stable approach criteria to an aircraft that's closer to a C172 than a B737.

[Say Again, Over!]

Hi Don, I believe that the best way to increase safety is by making pilots and other key players safety aware as opposed to safety compliant.  We are only (really) safety minded when we accept, assess and mitigate risk.  Trying to proceduralize risk assessment only leads, IMHO, to complacency. 

Not to say of course that we should now encourage unstable approaches!  What I do see is that we made "the stabilized approach gate" the holy-grail of safety when it is only a part of it.  Reading the study, I can see that ALAs were incurred when something happened after the decision to land was made "because we are stable".  The check marks had replaced solid risk assessment and valid and important cues were missed.  Are we not building a form or confirmation bias by insisting on such procedures?

This being said, I understand that in a busy, task saturated environment, the procedure is supposed to make risk assessment more efficient by making a few complex decisions simpler (i.e.stable approach criteria).  But would seat of the pants flying, in controlled situations, not be conducive to honing those decision making skills?

A few (ok, many) years ago, in CYYZ, we used to stay on the dual configuration throughout the day.  This meant that during some periods, all 5 positions were opened (North and South towers and Grounds plus Clearance Delivery) and we all shared the 2 or 3 airplanes that were at the airport all with a very low alertness level.  We found that this is when the stupid incidents happened (this is true of a majority of incidents in ATC).  We proposed that we combine positions and make the operation slightly more complex to keep people stimulated at the proper level and I do believe we had some improvements.

I'd be curious to know the metrics of alertness for the ALAs in the study.

In the meantime, I forwarded this study to our safety people as well as the instructors in the school.  Especially wrt to recommendations 1, 18 and 19.

[Don Hudson]

Hi Felix - I've always strongly believed in recommendation 19! The more familiarity with each others' jobs, the better. I've always felt that controllers should get fam rides. I'm not sure whether the regs permit that today but they should.

The recommendation regarding communication with the flight is also a good one. From pilots' perspective it permits concentrating on the correct execution of the go-around. SOPs I'm familiar with require a verbal call from the PM, (pilot monitoring) of "positive rate" before the gear can be selected up. If an interruption occurs at that point, then the call may be omitted and the gear left down until the noise level catches the attention of the crew.

Once the call for the go-around is made, (normally by the captain), the execution, (performance) of the go-around involves four immediate things:

• thrust increase, (manual or automatic) to TO/GA setting;
• pitch change, normally to 15deg up;
• flap retraction to the go-around setting, (on command of the PF);
• "positive rate" call followed by gear retraction, (on command of the PF)

The pitch and thrust settings vary according to circumstances. A go-around from minimums would follow the above sequence. A go-around from altitude, say near the FAF, would be handled differently - probably lower thrust, lower pitch. (On the Airbus, the thrust levers must be set to "TO/GA" to initiate the go-around mode for the engines and the FMS; they are then brought back either to the Climb detent or operated manually).

Some companies are moving beyond the "formula-based" stabilized approach towards the decision-making process expressed in the Report. That means that an approach which might be characterized as unstable-by-snapshot, (going through the red light at the 500ft "gate" for example), ultimately becomes stable when examining the data; - for example when high tailwinds exceeding the certification limits of the aircraft reduce to slight headwinds in the last few hundred feet. If the airport is a high altitude one and it's a hot day, the groundspeed (and energy level of the "mass" on approach) obviously can be very high. But if the crew has knowledge that the wind reduces or turns to a headwind, should a go-around be conducted? I think the 300ft point discussed in the Report helps resolve the decision and a go-around can be executed at that time if the energy level of the aircraft remains too high for a safe landing and stopping.

Perhaps the thinking and the SOP could be, "below 1000ft, be prepared to go-around at any time an instability of the approach exceeds the ability of the crew to return to stability, the limit being 300ft which would be the mandatory go-around point if unstable according to the usual targets.

If one has good GPS data, one can then reliably set the touchdown point (both longitudinally and laterally), and then determine a required deceleration rate from which a calculation of say, "runway remaining at 70kts" can be made. I know this works; as a variation, one can also determine risk levels if the airplane is above a certain groundspeed with 1000ft remaining, etc., etc.