Lets bore the audience with some history and clarity:-
If you believe the Airservices "risk analysis" for the windback of 2B was kosher, you obviously haven't read the three analysis by eminent risk management experts, including one paid for by CASA. Or maybe you really believe the error rate for pilots dealing with ATC instructions really is 50 to 100%, but for ATC person, only one in a million.
If one were to accept what you say as being the case, and apply a correction factor, the net effect [difference in outcome] of the collision modelling [ARM] [and ALARP] would amount to an amplitude difference of how much? I would suggest ‘nil’ in terms of scaled collision mitigation options that are available, particularly where surveillance is not available to offset the lack of an Air Traffic Control Service to both IFR [on conflicting but unknown VFR] and other conflicting VFR with VFR in Class E!
Whilst we are at it, we need to clarify the use of the word ‘analysis’. In the context of these ‘safety’ related discussions, the two main groupings within the ‘defined’ aviation processes are ‘HazID’, and of course separately ‘Aeronautical Studies’.
HazID’s
I have seen some of those for NAS2b, and know some of those directly involved. The HazID’s did not support the retention of a specific class of airspace above class D zones, rather, identified the hazards that needed mitigation. We should discuss how these were then applied [or not] to the non-existent volume specific assessments and ALARP? Which of course brings us on to the ‘governance’ trigger for the return to class C?
Mr Jeff Griffith [whom I have met] from the US, was first involved in Airspace in Australia in 1996, through 2004 [NAS2b]. In 1996, he was contracted by the then Australian NAS Implementation Group to provide technical input into each of the NAS characteristics. Previous to NAS2b commencement, industry raised concerns regarding the lack of volume specific safety assessments. None the less, NAS2b went ahead.
After several safety issues, and subsequent BASI/ATSB reports, moves were made by industry to address the inherent inadequacy of Australian NAS2b Class E.
The reports you mention
In 2004, Mr Griffith was again called on this time by CASA to assess the Airservices case for return to Class C over D. In August of 2004 Mr Griffith submitted the following:-
http://www.infrastructure.gov.au/avi...ith_Report.pdf
- He suggests that the US NAS is safe and efficient [see the NTSB data below]
- He makes no mention of the difference between Australia and the US in terms of surveillance, ATC sector size, traffic density comparisons.
- He makes no mention of the disparity between US D/E and Australian D/C Air Carrier movements.
- He seems unaware [or omits] that the US NAS operates Class C where similar levels of RPT/Air Carrier aircraft operate in Australia.
It was put to Mr Griffith at the time, that the absence of commonality between the two systems made transplanting one into the other necessitous of volume specific safety cases, which before rollout, had been rejected by the NAS people he had been working with.
In addition to the Griffith report commissioned by Mr Dolan at CASA, was of course Professor Terry O’Neill’s report, whom CASA also retained to assess the Airservices safety case for return to class C
http://www.infrastructure.gov.au/avi...AAs_NAS_2b.pdf
With regard to method, Professor O’Neill said:-
The other approach is to recognize that there is a wealth of relevant data available in other countries, for example the United States. Nevertheless there is considerable disagreement about whether the use of this data would be appropriate [8].
[8] It is noted that in previous modelling exercises, the accuracy of the ARM was tested when its risk profiles were compared with risk levels predicted by the US FAA’s collision probability formulae for uncontrolled and controlled non radar terminal areas – the formula used in air traffic control tower cost-benefit reviews in the US, Canada, Australia and New Zealand.
Here are some other extracts relevant to our discussion:-
The argument revolves around the comparability of the two countries. For example, Airservices argues that a comparison with the US is not appropriate due to differences in weather, topography, pilot behavior and culture and ATC systems. Some of these factors may increase while others may lower the rate of collisions. The following two points can be made about this controversy:
• It may be possible to estimate and formally characterize the size of any such differences and control for them statistically in the modeling process
• No two situations will ever be exactly similar. What is important is the relative importance of these differences compared with the size of the effect being estimated. The larger the apparent effects noted, the more confidence there would be that these were not entirely due to extraneous factors.
Correct. Which in hindsight is exactly why industry at the time were vigorously asking for
volume specific assessments prior to the NAS2b roll out. They did not happen.
He goes on to say:-
Reasonable agreement of the model with the US data would be reassuring. On the other hand, dramatic differences would suggest that all is not well with the modeling exercise. The comparison would need to allow for the differences, but the differences should not prevent the comparison being made.
Again correct. Why was it not done?
Alternatively, rather than checking the Australian estimated risks against the United States, the whole modeling process could be tested directly on United States system. United States traffic data from comparable locations could be fed into the TAAM software and a United States expert panel could be convened to populate a fault tree with probabilities. If this exercise produced estimated risks in reasonable agreement with observed empirical risks in the United States, then it would supportive of the modeling exercise in Australia.
An expert panel from the US? great idea! Why was it not done before roll out?
Does any of this sound familiar?
It strikes me as foolhardy to remake similar errors of process.
"Low" and "vanishingly small" are far from the same thing.
The difference of course is that ‘low’ is part of the scaled outcome terms used when reporting considered risk/s. Where does ‘vanishingly small’ feature in defined aviation [airspace] risk assessment processes?
On that theme:-
Several Airservices "analysis" of US MAC/NMAC in E have been "interesting", one neglected the fact that much of the mid west is 6/7000 ft, and claims a number of en-route problems. When the geographical position and height was examined, it was clear that ALL to alleged en-route MAC/NMAC were, in fact, in circuit areas or the boundaries of circuit areas, they might have been 7/8000' AMSL, but they were only 1000/1500' AGL.
The class of airspace involved with these MAC and NMAC’s was? Remembering of course the practical differences between US D rules and ICAO D application such as Australia and the UK! Or were they plain old Class E accidents and incidents?
Have you a reference?
From the Broome
’Aeronautical study’, may 09’
In addition it must be stated that the ARM equates MBZs to CTAF(R) procedures and assumes that pilots are 99% procedurally compliant.
Clearly the ARM does not equate the error rate for pilots dealing with ATC instructions to be 50 to 100%. I’m not sure the Airservices case [2004 rollback] using TAAM came to that conclusion either.
And;
5. Modelling
The OAR used the Airspace Risk Model (ARM) to model the airspace affecting the Broome aerodrome. The Airspace Risk Model (ARM) and an F(N) curve was developed by the Civil Aviation Safety Authority (CASA) and utilised by the OAR.
CASA has developed ‘acceptable risk criteria’ with regard to the risk of midair conflicts within regional aerodrome terminal areas. The collision risk model developed by CASA, in 1996, is focused on a non-radar controlled terminal area model and no significant changes have been made since its development and presentation to the Review of the General Concept of Separation Panel (RGCSP), now the Separation and Airspace Safety Panel, of the International Civil Aviation Organization (ICAO).
This aviation risk assessment tool is internationally recognised, and has been used for many years. Funny thing though, where is the modelling calculations for the options within the Broome report, including the E over D, and C over D options?
Sound familiar?
In general terms regarding risk in class E:-
Aviation Accident Database Query There is plenty in there to sober up with!
More specifically, Class E, and the limitations of see-and avoid, I commend the following TSB Canada report to the interested reader:-
http://www.tsb.gc.ca/eng/rapports-re...6/a06o0206.pdf
The collision occurred in Class E airspace (see Appendix A) where there is no requirement for an air traffic control (ATC) clearance or radio contact with air traffic services. In this type of airspace, there is no requirement for position reports, traffic advisory calls, or for aircraft to be on a common very high frequency (VHF) radio frequency. Aircraft are not required to have a communication radio, a radar transponder, or collision-avoidance equipment on board. It is unlikely that there was any communication between the two aircraft.
There is considerable guidance and research material on this subject (see Appendix F for endnotes), salient aspects of which are as follows:
• Conspicuity characteristics
- Specific paint schemes and patterns may have an advantage in certain conditions but none has an overall advantage over another.
- Anti-collision/strobe lights do not have a significant effect in bright daylight.
- Landing lights are useful when the opposing aeroplane is in the direct beam.
• Pilot scanning technique
- The United States Federal Aviation Administration (FAA) recommends that pilots spend 75 per cent of the time scanning a 180° by 30° field of view outside the cockpit.
- Estimates vary from 54 seconds to 9 minutes to perform the scan.
- A total of 12.5 seconds is required after first detection for pilot recognition and reaction to avert a collision.
- In practice, pilots spend 33 per cent of the time scanning outside mainly within 10° of the direction of flight.
• Effect of traffic alerting
- Proportion of time spent scanning outside the cockpit tends to increase.
- Attention is focussed on known location of conflict.
- Probability and range of detection increase.
• Probability of detection for this collision geometry and closure rate
- Maximum discernable range for 6/6 visual acuity: 8.5 km.
- Earliest likely detection range and time: 3.2 km, 28 seconds.
- Probability of detection for 33 per cent outside scan time: 25 per cent.
TSB records indicate that 16 mid-air collisions occurred in Canada during the preceding 10-year period resulting in 27 fatalities and 5 serious injuries. Of these accidents, 4 involved some form of formation flight or gliders operating in thermals, and the remainder involved aircraft that were not associated with each other. None were within Class D or higher airspace, and none occurred under ATC or advisory service. A total of 6 accidents occurred within the traffic zone of an uncontrolled airport and 6 involved flight beneath controlled airspace associated with a major airport. One of these occurrences led to a Transport Canada safety review in 2001-2002 of VFR operations in the vicinity of Toronto, Ontario,1 significant aspects of which are as follows:
• Routing restrictions and confined vertical dimensions contribute to traffic congestion.
• Two system deficiencies were identified concerning availability and quality of aeronautical information and lack of standard operating procedures for VFR operations.
This is what concerns me [and apparently many others] with Class E and not properly utilising/applying A.L.A.R.P :-
• A number of risk scenarios were identified in which a mid-air collision was a potential outcome that was considered unlikely within the five-year time horizon of the study.
Yet ‘vanishingly small’ came home to roost!
• Several recommendations were made to address the deficiencies and risks.
There has been little progress in implementing the recommendations.
Another of the occurrences took place approximately 5 miles west of the location of this accident and in similar weather conditions. It involved a Cessna 172 aeroplane on an instructional flight and an ultralight aeroplane on a pleasure flight. The right main wheel of the Cessna 172 rolled along the top surface of the left wing of the ultralight. The ultralight was damaged but was able to land safely. The Cessna 172 was undamaged. Neither aeroplane saw the other before the collision.
Make that came home to roost twice!
A number of international studies (see Appendix F) have addressed the overall issue of risk of collision and effectiveness of the see-and-avoid principle. All acknowledged the underlying physiological limitations at play and that, when mid-air collisions occur, “failure to see-and-avoid is due almost entirely to the failure to see.”
One study stated that “our data suggest that the relatively low (though unacceptable) rate of mid-air collisions in general aviation aircraft not equipped with TCAS is as much a function of the ‘big sky’ as it is of effective visual scanning.” Specific results relevant to this occurrence are as follows:
• A French study of mid-air collisions in a 10-year period found that the see-and-avoid principle was becoming no longer adequate as the sole means of averting collisions.
• An Australian study concluded that the see-and-avoid principle, in the absence of traffic alerts, has serious limitations and that the historically small number of mid-air collisions is as much due to low traffic density and chance as it is to the successful operation of see-and-avoid. The most effective response to the many flaws of see-and-avoid is to minimize the reliance on it.
• A German study on the detection of gliders and small motorized aircraft found that passive conspicuity measures did not overcome the underlying limitations of the see-and-avoid principle. It recommended further development and promotion of GNSS and Mode S transponder technology, noting that there is already on the market a GNSS–based system that is ADS-B compatible, called FLARM. That system is in use on gliders and provides collision-avoidance information.
• Eurocontrol examined risk of collision scenarios between uncontrolled VFR general aviation aircraft and both other uncontrolled VFR traffic and IFR commercial air traffic and found that see-and-avoid was ineffective as the sole means of averting collisions. It preferred technological solutions, specifically increased use of Mode S transponders to function with secondary surveillance radars (SSRs), aircraft collision-avoidance systems (ACAS)/TCAS, TIS, and ADS-B, and to that end endorsed continued development of the Light Aviation SSR Transponder (LAST) including low-powered variants. It found that systems such as FLARM could reduce the risk of collision between VFR aircraft.
• A British Royal Air Force study into mid-air collisions deemed to be random found that the probability of conflict is proportional to the square of the traffic density and recommended avoiding altitude restrictions that concentrate traffic.
Common theme!
Analysis
There was no indication that crew performance played a role in this accident or that either aeroplane was ill-equipped as to conspicuity devices or that those devices were not used appropriately. A realistic probability of either aeroplane detecting the other was 25 per cent, and without detection, the collision was unavoidable. The two aeroplanes were on a constant collision course; therefore, there was no relative angular movement that could be detected by peripheral vision to aid in detection. There was no other means of alerting either aeroplane as to the presence of the other. ATC does not provide traffic advisories in that airspace.
This about sums it up for me:-
Measures such as improving aircraft conspicuity, pilot scanning technique, and pilot traffic awareness can reduce risk, but they do not overcome the underlying physiological limitations that create the residual risk associated with unalerted see-and-avoid. There is only limited potential to further reduce risk by fine-tuning the unalerted see-and-avoid concept, and such an approach does little to address the risk of collision between VFR light aircraft and IFR commercial traffic in congested areas.
A meaningful improvement to the ability to see-and-avoid between uncontrolled VFR aircraft requires a practicable, affordable method of alerting pilots to the proximity of conflicting traffic. Reduction of conflicts between VFR aircraft and IFR traffic depends on making aircraft that are presently not transponder-equipped visible to ATC or to the IFR traffic.
Learn from others who know from bitter experience, the bitter experience is world-wide!