Monitoring takeoff acceleration
 

Taken from an excellent post by arcniz over in the R&N forum.

I dont want to distract from the actual crash discussion, so posted it here.....

Any comments?

Mutt.

Great stuff - after all, didn't this kind of creative thinking produce the aeroplane to start with?

:cool:

It sounds like there is an unreasonable expectation of precision in the proposal; there's significant scatter in all test data for takeoffs, and more so operationally I suspect. I'm not sure what it's going to tell the crew usefully that they don't already know, without being set to a threshold that generates false warnings routinely.

The difficulty is, from brake release point or time to V1, the acceleration is not linear. This is because drag and thrust changes with speed. So acceleration at each point of ground run will be varying. If it were linear, it would be very simple and helpful to furnish the flight deck with an instrument comparing the actual accelaration with a preset expected or calculated acceleration value. If the actual acceleration is not in the slot, rejecting the take off well before V1 would be much safer.

In the air force, we used to check the distance markers on the sides of runway and compare it with current IAS to ensure the airplane is in the slot of acceptable minimum acceleration.

I thought I would re-post these 3 links discussing some development work that's been done on TOPMS (Take Off Performance Monitoring System), and the issues raised by the work that's been done so far.

NASA TOPMS research

University of Saskatchewan thesis on TOPMS system

NASA TOPMS cockpit display development

Fundamental to making this system work, is that it MUST be able to accurately predict (within an acceptable margin of error) if the aircraft will reach the calculated V1 speed at the calculated distance down the runway, at the present thrust settings. As a secondary function, the system should be able to predict if the aircraft can reach the calculated Vr speed before the end of the runway, at the present thrust settings.

It seems to me that there are three primary areas of technical development that must be done to make this system work.

1. Aircraft position on the runway must be known by the system to a fairly high degree of accuracy, and the runway parameters must be known by the system.

2. Current speed and acceleration must be known, and predictable changes in acceleration must be known with speed increase, thrust change, current aircraft configuration, etc. This concept of dynamic prediction is easy to grasp, but the reality of actually making it work is quite complex.

3. An acceptable display and pilot information and warning system must be developed.

The problem with the method used now, is that once calculated V1, Vr and thrust settings have been obtained, the pilot accelerates down the runway watching a buildup to these speeds, without any direct measurements of how much runway has been used while reaching those speeds.

(Epiphany moment) :D

In thinking this through, I recall in the old Jimmy Stuart movie about Charles Linberg's famous crossing of the Atlantic, that they put a painted white stick in the ground next to the runway at the point where the Spirit was supposed to reach Vr. I wonder if the NASA approach for a TOPMS system is too complex in trying to dynamically predict the down runway speed and distance of the aircraft during a takeoff roll, based on present up runway performance.

Maybe it would just be simpler for a TOPMS system to create computer generated "sticks" along the runway at set distances in a display, with expected speeds at each distance "milestone", after the takeoff parameters have been entered. Current speed data is easy to obtain, and current distance data down the runway should be easy to obtain. Then all the system has to do is compare and display expected speeds with actual speeds, as each of these runway "milestones" is reached. You would quickly know whether you were accelerating along the expected performance curve or not, and the reason for a slower than expected takeoff could be determined after the abort. This system would be MUCH simpler to design, compared to a dynamic performance predicting system.

An interesting one for me. Thirty years or so ago I returned from the US with an agreement with the inventor of a device that I saw immediately could be turned into an ‘acceleration achieved’ display—operating in real time.

The old Colonel in Texas had spent most of his life lecturing on blind flying…in the later years these lectures became highly regarded reminiscences at Randolph AFB etc., of attempts to make people believe that instruments were necessary at all. He was also the holder of several aviation patents.

I quickly drew up plans and proposals for the appropriate government offices, but was disappointed when they advised me that they were already financing ( 50% ) research on such a system. A BAC 1-11 was routinely charging up and down some runway using Doppler—the beam being squirted onto the concrete—and, I think, a dual needle on the ASI.

It all came to nowt, which was prbably a good thing for my finances, but a very interesint time. The old gent had done many things, including throwing hooded pigins out of aircraft (they paraglided) and using a stethescope to hear pings from an echo bounced off the fog covered surface...high tech days.

We already have all the gear on board for an effective takeoff acceleration monitor and alert system...IRS for accel achieved + Runway data in the FMC database (to calculate accel required)
All it requires is some clever team to write up the program. Speed tapes have a trend vector - if required acceleration to 80kts is not achieved this vector arrow could start flashing red and a timely stop could be made. It would be inhibited above 80kts. Get on to it Guys!
There have been many T/O performance accidents that have not involved engine failure - the DC-8 60 series (was it ONA or Capitol?) comes to mind that started T/O with the parking brake set comes to mind, and what about the recent 747 at YHZ? - although I don't know the details of that one.
Some time ago there was anecdotal evidence of an interesting T/O by a major 747 carrier at a far-eastern airport. Thrust was set and the a/c trundled off - albeit slowly, down the strip. Things looked pretty tight but it eventually staggered into the air and dragged its @rse all the way to flap retract, when climb thrust was selected and - the a/c came rapidly to life and accelerated away nicely! For some reason the set thrust had been nowhere near t/o thrust and it had gone unnoticed.
We need a simple T/O acceleration monitoring and alert system

Monitoring takeoff acceleration
 

Once upon a time there was a parameter to be checked during the take off run.
It was by first seventies, when I was flying LA CARAVELLE;
The "Time to 100 Knots".
Perhaps it's time to resume it?:ok:

Take of acceleration
 

In Jaguars on hot and high days we had a speed check at every marker (300m) and at the middle marker if we didnt cross above a certain speed, it was mandatory to reject. we had a couple of airfields at very high altitudes where take offs were really hair raising.

The similar for landing at the middle marker we had to be less than 100 Kts or we would deploy the tail chute.

There are accelerometers available linked to the INS which could be checked by the PNF about 2-3 times and if the acceleration drops below a certain value than one could be warned. In addition this would help during deceleration on landing.

The reason I think a "speed at distance" system would be easier than a "dynamic performance predicting" system, is the basic equation a=f/m.

In this equation, acceleration is changed by either an unexpectedly low thrust level (including unexpected drag) or an unexpected weight. Depending on the level of sophistication desired in a dynamic predicting system that might cue pilots for actions during takeoff (such as increasing thrust, etc), the system has to accurately figure out whether there is a mass problem, or a thrust problem. This would be complex and the system could never be wrong for safety's sake. Then the system would have to dynamically change and display its prediction based on performance changes made by the pilot.

The "speed at distance" system would be much simpler to design, certify and implement, as there would be no prediction (after the initial speed at distance plots were generated), and no action cues would be generated by the system for the pilot (thus no possibility of false warnings or false action cues). The pilot (PNF)would simply read the displayed numbers and use his judgement instead (and follow SOPs etc).

A HUD display generated by this system would be possible as well. :O

Time to V1
 

I like this because you can simply add the calculated/measured/interpolated time to V1 column to the takeoff charts.

Start the stop watch at throttles up and abort if V1 is not achieved by that time plus a tolerance.

If acceleration is insufficient when the time is up, then you do have more runway ahead at the expected time and have more distance to stop and less speed to get rid of;)

This is a better deal than being 30 kt. short of V1 while rapidly consuming remaining runway and beginning to wonder what's going on:uhoh:

In a glass cockpit a/c, the time remaining could be represented as diminishing line hanging below V1 that would flash appropriately as the decision point is approached and sometimes exceeded.

An alternative representation could be based on the familiar remaining circular seconds symbol we see on movies that would sit in the middle of the attitude display -- we're still on the ground and won't be flying until past Vr.

One could also choose a speed such as V1-10 so that possibly distracting displays would be gone when it's time to fly the thing.

Start the clock ticking automatically when the throttles go past what's normally used to begin taxi.

Time to 100 knots on the 'bus works very well!!

The 332 - set 1.03 EPR, start the clock and voila - 30 secs to 100 kts.
The 343 - set 50% N1, start the clock and voila - 30 secs to 100 knts.
The 345 - set 1.05 EPR, start the clock, and also voila - 30 secs to 100 kts.

This works for every take-off that I've tried it on:ok:

PS - RR Trents on the 332!

I don't think there is much rocket science in this. It is simply time vs airspeed. You have all the tools you need in the cockpit. An ASI and a stopwatch. All that is missing is the performance data from the manufacturer.

When I was flying C130's in the military we used this system to great effect when doing max effort takeoffs from limiting strips. I can't remember the exact procedure, but it was time to 80kts i think. The nav would start the time and call "time" when the calculated time had elapsed. The co-pilot would monitor speed and call "go" at 80kts. If the "time" call came first we would reject. If the "go" call came first then we would continue. A simple acceleration check.

Admittedly this particular procedure had room for confusion, but it could easily be adapted to a modern jet.

Time to 100k;
 

British Airways [BOAC] were checking the time to 100kts a thousand years ago on their Britania fleet. There was a graph in the flight manual for various weights but it always seemed to be 23 seconds . It was the Captains responsibility to start the stop watch. I've never been able to understand why it was not used on the modern fleets. Perhaps it's just too do. !!!!! 828a

Don't like the idea. There is enough to look at and monitor whilst handling an accelerating aeroplane ready to handle abandon take-offs, gusts, engine failures, technical failures on the take off roll. The performance of the aeroplane has already been calculated. All you need to do is ensure you have the thrust you need, light the blue touchpaper and go. The acceleration will take care of itself. Fancy acceleration monitors will just add yet another dangerous, distracting complication. There are so many safety factors built into the performance (ie 50% or 150% wind component). If you start your takeoff roll with a 40kt headwind, you're already well on the way there. The big drawback to me is that there will be a greatly increased incidence of abandoned takeoffs due to 'acceleration problems'- these will inevitably arise at high speeds, and stopping may well be more dangerous than going- an abandoned take-off should not be viewed as a 'safe event'.

Notso - maybe you've already forgetten MK's recent disaster in Halifax!!! A stopwatch with an approx. time to 100 kts might have helped them realise things weren't right.
Also remember "Palm 90" that crashed in the Potomac! Same thing acceleration wise.

And really it's NOT difficult to glance at the clock at the 100 kt call is it??

I am reluctant to discuss the Halifax incident and make any implication the crew were somehow to blame (as you appear to imply) until facts come out. It seems somehow wrong to point a finger at people who cannot defend themselves when we don't know yet what it was about.

I still stand by my judgement that the last thing you need on takeoff is an added distraction. This is from 23 years flying heavy jets on limiting runways. The performance has many safety factors in it. I believe there would be even more accidents in the form of unnecessary overruns with such a system. But I don't expect everyone to agree- just tossing an opinion in to be savaged by the wolves!

I don't seem to be "implying" that it's the crews fault at all, I just state that that they may have been able to pick up the fact that something wasn't right at an earlier stage.

Hey guys! An intelectual discussion... not often seen on this site!

I have a graph depicting acceleration versus time on a DC-8.. but the unfortunate thing is that it's all based on someone's old calculations on a -55. As you can imagine, these wont even get close with a -62, a -63, a -73 or anything above that, what can be one for those of us who deal with this in the real world, and don't have enough brain-power to deal with all the added worry of "will my heavy -8 get airborne out of this terribly short strip, or not? (completely disregarding the 747 YHZ site clever inputs!)

As I write this, my collegues advise me to keep away from discussions like this, as they don't concern me...

But I must ask (and I plead with 747Fecal and A-Floor not to respond), am I wrong in the gen I have been given, or should I just carry on as I do, pleading to get these antiquated bits of noise-generators into the air?

Although my previous post in this thread was more than a little based on reminiscences, the subject matter is something I have always felt strongly about. I'll go out on a limb here: I think performance graphs are nothing more than a starting point; the final calculations have to be based on skill attained over time.

A f'instance. I was being checked by my boss on a flight into a small airfield in Northern Yugoslavia. (BAC 1-11) He did the T/O on the return, and somewhere over a rocky beach, I called V1. My concern had been expressed verbally and by fire-walling the power around half way.

The overall performance had been degraded ( I have always assumed ) by the huge uneven concrete slabs thundering away under the wheels, presumably displaced by movement of the landmass. There was nothing in the tables about runways that are shaking the aircraft to pieces.

There were a couple of other times as a F/O, where I have been left stuffing the power through the panel and playing the flaps like some musical instrument. In all cases there was no reaction, or post flight analysis. Bad, bad days, but the graphs had all been ‘correct'.

There has to be a better way. As has been mentioned, all the data are there...the hardware is in place in a lot of cases. Let's have the software.