DFDRs, FDAUs, FDIMUs, AIMS, QARs and FDA/FOQA data

[PJ2]

I would like to know from anyone with knowledge, expertise, experience or who may have or know of documentation, as to whether QARs which are typically designed and installed to serve FDA - FOQA - FDM Programs and are not crash-proof, obtain flight data from the same data sources as the crash recorders and the AIMS.

These sources are typically the ARINC 429 and 717 data busses.

I realize that dataframes are typically different, QARs being more robust in terms of parameters collected and sample rates employed. I am only asking if the data comes from the same data-busses and that it is the same raw binary data being translated into engineering data for use in FOQA - FDM analysis software.

My reading tells me that this is indeed the case, but I would like some input on the question. There is a sense within the industry that FOQA data which is employed in FDM/FOQA Programs and provided to airlines is somehow "suspect" in the sense that it is "not the same" as the DFDR recorded information and that it therefore can "be dismissed" as "advisory" rather than be taken as legitimate as data from the DFDR whereas DFDR data somehow as "legal" recognition and must be taken as accurate and upon which operational decisions can be made.

If anyone has any thoughts based upon the above qualifiers regarding knowledge etc, I would be very pleased to hear.

Thanks - PJ2

[Piper19]

My reading tells me the same. Digital data from ADC, aircraft systems, etc...goes to flight recorders and FDAU, but FDAU only receives the needed data, which can vary from airline to airline and can be selected by them. Flight recorders get full data. But FDAU and FDR get data from the same sources. FDAU data goes to WQAR to download.

[PJ2]

Piper19;

Thanks for the response. The essential issue is, the data is the same, regardless of which device it is going to. There is no "special" data targeted for the QAR - it is the same information that goes to the FDR. The key to understanding this is the notion of the dataframe and what the installed equipment and associated software is designed to capture. I needed to ask the question of other professionals because the notion of the validity of FDA data is being challenged by some, in that, while the DFDR data has legal standing and is regulated, FDA/FOQA data is not, and can be "dismissed" when "inconvenient".

Thanks for your input.

IGh;

Thanks - I know Mike - we've spoken a number of times on FDA issues. I'll check the ISASI site for his paper - it may assist.

I'm a bit surprised at the suggestion that airlines use the flight data recorder data for their everyday FDM - that means pulling the crash recorder every day (or downloading it from the FDR, which is not nearly as easy or cheap as downloading a QAR, especially if wireless technology, now quite cheap, is being used) - in fact, given the pressures of line operations, downloading from the FDR is, for all practical purposes, not possible, (it can be done, but the time and the mtc resources aren't available in most airline ops to make this a doable solution). The view I'm trying to fully understand and currently believe in, is, FDR data isn't any more "accurate" or robust than QAR data because it is from the same source. I've done extensive reviews of available literature on the net and all confirm this but I was wanting to know if anyone here had had experience with their data being challenged as inaccurate or wrong.

I know that there are extensive procedures to establish data accuracy and sensor validity - those regs are written out, but given today's recording technologies, (the FDAU is not an FDR - see above post), it would seem to me that regardless of how it is captured and presented, ("FDR" or QAR), the data has the same validity but is merely captured by different devices. A QAR can be programmed to capture far more parameters and at higher sample rates than an FDR which must go through rigourous certification processes before such changes can be executed. That doesn't mean the data is less accurate - it means that the processes governing FDRs (not FDAU though) is regulated, but FDA data from QARs isn't.

I don't mean to answer my own question here! I am very curious what others think of this notion and if they have had experience in dealing with airlines which do not accept FDA data from QARs as legitimate data upon which decisions may (or sometimes, must) be made.

Thanks,
PJ2

[john_tullamarine]

I'm not an electronics man (electrons, wave motion, Fermi levels and exclusion principles are all a bit esoteric for greasers such as me ..) so the following only relates to my end use of the gadgets ..

As I understand it

(a) the QAR is just a DFDR without the crashworthiness capabilities ie, a bit like comparing a Goggomobile with an armoured truck ..

(b) in any case, the data can be a bit suspect/difficult in part and interpretation needs to involve sensible engineering and flight mechanics considerations. As a simple example, temperature calibration is a right pain in the neck and, unless the box is flight tested subsequently to qualify the calibration, I wouldn't put any great credence on that particular output parameter.

given the pressures of line operations, downloading from the FDR is, for all practical purposes, not possible

This probably depends on the boxes but, for example, one fleet with which I am involved is set up so that the DFDR can be downloaded quite simply by plugging in a cable externally and dumping the memory .. and I generally do this myself .. easy peasy. The protocols have been agreed with the Regulator and the Regulator is most happy that this fleet's DFDR serviceability is regularly and routinely monitored. Wireless technology would be neater, but having to cart a laptop into the hangar and plug into a socket is not really a major imposition. The particular box holds enough data so that the downloads can be made to suit minor hangar visits or during low activity periods on line.

This fleet has the FDR downloaded as the aircraft are not fitted with QAR. I did investigate the practicalities of installing QAR but the cost didn't warrant the utility. Had such an installation been pursued, the data would have been picked off in parallel with the DFDR.

the data has the same validity but is merely captured by different devices

Sounds about right to me. I, for one, could see little point in quibbling about whether the data came from one or the other. However, the problem I see is how one sets about interpreting some of the data, especially if one is interested in looking at short time histories ...

[PJ2]

John, thanks.

I appreciate that the DFDR can be downloaded by cable. We used to have the FDR replayed in our replay center once every six months (late '80's) before we stopped doing FDA completely for some years, but with a large fleet and turn-times of less than an hour these days, downloading from the DFDR isn't practical.

Flight-testing every parameter that the QAR is provided by the FDAU isn't practical either, and in today's extreme cost-restrictive airline environment, isn't possible. The QAR being used downloads over 1800 parameters from the narrow-body Airbus fleet type many of them system-oriented, (valve positions, flow rates, discretes) as well as FCU and FMA text annunications. We can't download ECAM messages. We download some, but not all FMGC entries. We do lat/long but it's only as accurate as a non-GPS FMC-calculated position.

Re, the DFDR can be downloaded quite simply by plugging in a cable externally and dumping the memory .. and I generally do this myself .. easy peasy. . .
Neat - good operation and clean. We have our mtc people replay the QAR cards when the fleet transits a main station. We couldn't possibly do it ourselves of course but it's great your system is working. Ours is not at all reliable in terms of getting all airplanes before the QAR card is full and data is missed.

The protocols have been agreed with the Regulator .
The Regulator's involved!!?? Interested even?! Wow, how'd you manage that? Where we are the regulator won't even support data security let alone FDA Programs. They say "do FDA" and that's it. We tried to get them interested in the above question (data validity, data protection under SMS) and it was an impossible task. There are no regulations regarding FDA here whatsoever - just, it's "nice" to have, under SMS. That changes next year I believe in terms of requirements to pass the IOSA audit process.

This fleet has the FDR downloaded as the aircraft are not fitted with QAR. I did investigate the practicalities of installing QAR but the cost didn't warrant the utility. Had such an installation been pursued, the data would have been picked off in parallel with the DFDR.
Thanks. We're considering WGL (wireless ground link) using cell-phone technology. We download about 4 to 6Gb of data daily.

Data interpretation is absolutely key in any and all of this and, unless one has years of experience and has flown, requires some line pilot input of some kind first to have credibility with the line guys and second to be able to interpret the data effectively. I'm speaking of a large fleet here and a large number of terminals/ATC and weather environments. I realize that not all Programs would necessarily require a pilot or that it can't be done effectively but keeping the above in mind, in the end, having an experienced crew member's eyes examining the data is invaluable.

Another thing - we keep all the data. We have about 4Tb (terabytes) of 320 data now and have a means to mine the data with new operational or commercial questions instead of using standard FDA analysis tools.

You've helped me with the key question though - why even consider the data "different"? But - we have that. Depending upon the event, sometimes the FDA data is dismissed. That's what's causing me to ask the question here.

[Blacksheep]

Just to clear up some things that are mentioned above which indicate a slight misunderstanding of data acquisition. The Flight Data Acquisition Unit (FDAU) captures data from the aircraft sensors. ICAO Annexe 10 is the source book for the rules, but different regulatory bodies may interpret these rules differently and several possible data frames are possible while remaining within ICAO Annexe 10 (E.g. FAA and UK CAA dataframes are significantly different with respect to capture rates for certain parameters.). Also, the FDAU monitors the mandated data to a dataframe setup that depends on when the aircraft was originally certified and what regulations are in place. Older aircraft are not usually required to upgrade the Flight Data monitoring system when new rules are brought in.

After signal processing and verification, FDAU output is passed to the DFDR for recording. (Verification simply means that the output data is checked for "reasonableness".)

Flight Data Recorder (DFDR) data from a representative flight sector is downloaded from each aircraft regularly by engineering and checked for capture error rate (must be <2% generally) The data is analysed and stored for reference - should there be an accident, the stored data is used as a data reference, as each aircraft will have slight variations in the parameter reference levels. In addition, the aircraft maintenance schedule calls for regular checks of the DFDR system and each of the recorded parameters is checked right across its range. (e.g. each aileron position measured at centre, full up and full down) using a data monitoring test set. Discretes are checked in both possible states (e.g. "On" or "Off") Any irregularities are corrected before a Certificate of Release to Service (CRS) is issued by the certifying engineer.

In most Quick Access Recorder systems - but not all! - the FDAU data is captured in parallel with the DFDR data and is therefore exactly the same data.

In interpreting data, experience and objectivity are required. The data doesn't tell you the reasons for any particualr measurement or change in value. It is important to bear in mind that the raw data is providing information on what the sensor is measuring, not what was really happening. The sensor may be faulty or completely detached!

I have personally listened to inexperienced personnel looking at data that I had downloaded for investigation say things like "Look, he put in excessive nose down pitch and the speed began to run away within ten seconds."
No!
The data says that "pitch increased to 9 degrees nose down and the speed increased steadily over the next ten seconds to reach 275 knots". The data doesn't say why - it is just dumb data.

One reason why there may be a perception that DFDR data is more reliable for legal purposes than QAR data is that it is subjected to airworthiness engineering surveillance. Obviously that is true, but the same data is being recorded by the QAR. If there is any doubt about the veracity of information found on the QAR, the parameter in question can be verified against the DFDR recorded parameter. In the extreme an engineering check can be performed against the documented values.

But be aware that there is always a 2% possibility of data error for any individual parameter in any particular record.

[PJ2]

Blacksheep;

Many thanks - superb and very helpful post.

I think I have what I came looking for.

Checking the QAR data with the DFDR is not practical due to the fact that QARs typically have far larger dataframes and the parameters captured therein do not exist on the DFDR. That said, concurrence and reasonableness is achieved through daily detail analysis and on occasions when an incident occurs and both the QAR and the DFDR are available for examination. I think your 2% figure is spot on.

With increasing use of QARs in investigations and with critical safety and commercial decisions being increasingly made based upon FDA/FOQA data, many now believe that it is time for regulators to adopt the same approach to QARs and FDA Programs that have applied to DFDRs for decades. As it stands, some operators use the notion that the data is "unverified" as an excuse to ignore the data when it points to costs but seemingly take the commercial information at face value.

Again, thanks for your very helpful input.

PJ2

[john_tullamarine]

.. which is why the Regulator concerned with the fleet mentioned above is more than happy for us to review DFDR data routinely .. if we find the box or individual parameters a bit suspect .. we fix it and the base incident/accident investigation tool is there for the rare occasion on which it might be required for such purposes ...

concurrence and reasonableness is achieved through daily detail analysis

label me "an anal engineer" but it seems to me that, for the relevant parameters, a periodic inflight calibration validation is essential .. then one has some baseline real data upon which to base routine work ..

[PJ2]

IGh;

We have also several cases where the DFDR was, or some parameters were, inoperative but the QAR data was reliable (given the understandings in the thread) and used for investigation.

The Concorde accident was the most poignant example of using the QAR vice the FDR and it assisted greatly in the investigation.

In fact, we have more often had the case where the legally-recognized systems (AIMS, DFDR) have failed and the QAR/FDA system has been able to provide data and information on correcting/fixing the DFDR. Validation once per year does not always seem adequate.

Understand your comments on filtered data. I am making the assumption that this would not necessarily automatically apply to QAR data as such filtering is either downstream of, or on a parallel stream to, the FDR where such filtering occurred.

While I know this is somewhat rhetorical and doesnt' require a response, all this begs the question, why hasn't legislation caught up with technology in terms of accepting QAR information on the same basis as FDR information?

[PJ2]

Hello John;

Re, .. which is why the Regulator concerned with the fleet mentioned above is more than happy for us to review DFDR data routinely .. if we find the box or individual parameters a bit suspect .. we fix it and the base incident/accident investigation tool is there for the rare occasion on which it might be required for such purposes ...


A healthy and positive relationship to be sure. Not sure it can be done that way in a large fleet. We do inform our mtc people on such issues.

Re, concurrence and reasonableness is achieved through daily detail analysis

label me "an anal engineer" but it seems to me that, for the relevant parameters, a periodic inflight calibration validation is essential .. then one has some baseline real data upon which to base routine work ..
Well, I'm probably an anal pilot (or is that a redundancy?) and to the extent of my knowledge, require similar approaches only because critical safety and commercial decisions are made based upon this information and it has to be accurate. The difficulty is, the FDR carries only about 650 parameters and, as described above, our QAR provides over 1800 parameters. There just isn't any way to verify such data by employing the FDR.

However, "reasonableness" enters into this I think. Given that there are parameters which are discretes, rates, positions, and values, some are easier to "logically" validate than others. Sometimes it takes a "birds on a wire" approach - (when one can't see telephone wires, but can see birds sitting in a line, reasonable "evidence" of the presence of wire, etc) and even an experienced sixth sense, all of which go to the art of interpretation and none of which is appreciated or understood where we work, again the original reason for the question. The question is perhaps not one of acceptance of valid data so much as interest level...