Figure: Ground Implementation
Roadmap.
Figure: Airborne
Implementation Roadmap.
The scope was
refined by re-examining the main assumptions made at the development of
the proposal about phases of flight, airspace, level of intent information,
location of the separation process and traffic samples. The terms were
more precisely defined and the consequences of the choices on the relation
between intent and capacity, the avionics and ground systems were highlighted.
Moreover, special attention was given to trajectory uncertainty that will
be a key element in the assessment. The refinement of the scope lead to
operational concepts for the systems to be assessed, a precision of system
and assessment objectives and the basic hypothesis to be used in the project.
Air traffic above a floor level would be simulated considering a unique
separation standard. Part-task simulations would consider small areas while
fast time and full scale simulations would consider a larger area. Those
areas would correspond to a simplification of a congested part of the Western
European Airspace and ratios of 1, 2 and 4 would be applied to the demand.
Both structured and unstructured airspace would be considered for ground
control while only unstructured airspace will be considered for airborne
separation assurance.
Details of the results
on the refinement of the project scope can be found in D1-1.
The main objective
of the survey was to ensure that the INTENT project uses up-to-date information,
and that activities are not duplicated. A selection (not exhaustive) of
past and on-going Air Traffic Management projects in Europe and the USA
was chosen from areas of particular interest to the INTENT project, namely:
airborne and/or ground based separation assurance, capacity metrics and
human modelling. An agreed set of review criteria was established under
the headings: technological, ATM performance, human factors, economic and
institutional aspects. The projects were reviewed over the period January
to March 2001 and the results are presented by alphabetical order of name
of project.
The results of the
scope refinement and survey of past and ongoing activities are available
in D1-1.
An addendum to this deliverable giving an overview of the Eurocontrol Free
Routes Airspace Project (FRAP) can also be downloaded: D1-1
addendum.
Within this work
package an analysis was made of the stakeholders in the air traffic separation
process, more generally the air traffic management system as a whole as
well as the desired benefits that each of these stakeholders will look
forward to when considering investments. The modification of ground and
airborne systems to exchange and utilise intention information, as being
explored within the INTENT project, will require some investment. Within
this part of the work, the issues that drive business decisions to invest
in new technology such as that needed for an intention-based air traffic
management (ATM) system are described.
In carrying out the work, the full range of actors with an interest in the ATM system has been identified. From this “all-actors set”, the criterion is developed that differentiates the stakeholder from any other actor. This criterion is in fact that the stakeholder is interested in the system from a business perspective and therefore inclined to make investments, along the same lines as any other replacement or expansion investment. From this set, the key stakeholders have been identified being the principal system manufacturers (aircraft, air and ground equipment), the service providers, the airlines (the system users) and the governmental organisations.
When looking closer at each stakeholder from a business system perspective, it becomes clear that 6 key investors remain from the original list of over 20 actors: National governments and European Commission, air traffic service providers (including Eurocontrol), the airlines, airports, the avionics and the airframe manufacturers. An investment dilemma has been identified as the business systems of these key stakeholders are interlinked and competing. As a consequence the simultaneous investment needed to bring significant changes appear to be unlikely, and, even worse, strategic considerations may in fact even preclude anyone taking the lead.
This is the case even though the required and forecast growth figures appear to indicate ample opportunity for the various players to successfully establish investment opportunities. Considering only the quantities of new aircraft in the air transport sector shows that 7,610 passenger and cargo aircraft are expected to be delivered, during the next ten years (through 2009), worth approximately $560 billion (2000 catalogue prices). The 7,750 aircraft delivered during the following decade will be worth another $750 billion, giving a total twenty-year business volume of $1.31 trillion.
If one also considers that more than 1 in 4 flights within Europe were delayed more than 15 minutes in 2000, the collective cost of these delays to airlines is huge (about 4 billion Euro in 2000). Governments estimate the financial cost to passengers is at least the same, bringing the estimated total annual cost to airlines and passengers to over 8 billion Euro.
This business climate combined with the developments in environmental requirements and the investment dilemma that the key investors are experiencing, appear to give rise to the conclusion that governmental action will be key to launching any co-ordinated activities required to bring the various stakeholders together in joint and synchronised investments. Return of these investments would have to proceed along the lines of taxation, royalties, incentive schemes or the like.
Alternatively, a major industrial entity should be able to stimulate or undertake the development of the air traffic system. This would be on condition of being accepted by the air traffic community at large in terms of acting as a pre-financer, any competitive aspects of that, his responsibility in terms of integrating community interests and the associated conduct of business.
The results of the
stakeholder analysis and the required benefits are detailed in D1-2.
To define a new
ATM system, an objective measure was required to measure improvements of
the ATM system in terms of capacity. Therefore, “capacity” was studied
and defined within INTENT. The results were essential to the whole project
since one of the primary goals was to quantify how intent information can
be used to increase capacity of Western-European airspace. The results
from this study were directly needed for the following part of project:
the design of the experiments.
Work has been done on the definition of capacity that will be used in the INTENT project and on describing reasons for capacity limitations for both the present system as well as future systems. The limiting factors for increased capacity were found to be:
An extensive overview of literature on capacity assessment and metrics has been conducted. These metrics have been implemented in tools. With these tools, experiments have been done to get acquainted with some of metrics. At the end of this work, the metrics that were proposed to be used for capacity assessment in the various INTENT real-time and fast-time experiments have been listed, as follows:
The capacity
study also investigated theoretical maximum capacity. This study looked
at capacity from a geometrical (static) point of view but also a number
of experiments were conducted in order to simulate the dynamics of aircraft
in airspace and their effect on airspace capacity.
The statical maximum capacity appears to be about 60 times the current maximum capacity for a separation minimum of 5 nm. The dynamic theoretical maximum capacity appears to be around 6 times the current maximum capacity. In this study, the simplification was made that all aircraft were at the same flight level.
The details of the
capacity study can be found in D2-1.
The aim of the
ground part-task experiments was to derive a human operator model for the
fast-time simulations based on measurements during the simulations and
to get feedback on the use of a conflict detection and resolution tool
incorporating aircraft intent information for an en-route sector. The experiments
were successful in that a human operator model was derived from the measurements
made in the simulation trials and the controllers provided useful feedback.
It was observed that the controllers could not safely handle the high traffic samples (2 x today and 4 x today) in the experiments. Scenarios with traffic densities of 1.7 x today were tried, but also proved to be a little too high. Traffic load appears to be the most dominant factor for controller workload, determined both from the run questionnaires, debriefing questionnaires and the workload model development.
The ground part-task simulation results indicate that the CD&R tool could be a helpful tool to reduce controller workload, especially in unstructured airspace. It appears that the route structure has little or no effect on controller workload if the CD&R tool is not present, but the CD&R tool is significantly reducing controller workload and increasing acceptability in unstructured airspace.
The planner controller is mostly using the CD&R tool; the CD&R tool is less efficient for the tactical controller. This is mainly caused by the detection function of the tool. The look-ahead time is 15 minutes, causing conflicts to be detected and presented that are not always of interest to the tactical controller. The resolution function is equally appreciated by the planner and tactical controller.
During the experiment and the debriefings, it became apparent that the human machine interface needs further improvement to reduce clutter in high traffic load conditions. Furthermore, familiarity with the simulated airspace and the new systems is an important issue for controller workload. Training of controllers is therefore very important and might need more attention and time in future experiments.
The results of this
study are detailed in D2-3.
The airborne
part-task experiments have been successfully completed. The aim of the
experiments was to derive human operator models for the fast-time simulations
based on measurements during the simulations and to get feedback on the
use of an Airborne Separation Assurance System incorporating aircraft intent
information. The experiments were successful in that human operator models
were derived from the measurements made in the simulation trials and the
pilots provided useful feedback on the use of intent information in an
ASAS system.
Six professional crews participated in the experiment and each of them experienced all four ASAS concepts (5 minutes state only, 5 minutes intent, 10 minutes intent, 20 minutes intent) with three traffic loads, up to 3 x today’s traffic load over core Europe, en-route airspace.
The following main conclusions can be drawn from the results of the experiment:
The fast-time
simulations have tested the ground and airborne concepts in a larger area
(9 sectors compared to 1 for the part-task simulations) and for a longer
time with more variations in traffic load and complexity. The workload
models derived from the part-task simulations were used in these simulations.
This way, the conclusions from the fast-time simulations will "extend"
the conclusions of the part-task simulations to a larger scale.
The ground experiment was built around the premise that, the way in which shared aircraft intent information is most likely to impact on the capacity of ground-based control systems is by making feasible the use of automated CD&R tools for assisting controllers. Within this context, the following conclusions were drawn from the fast-time simulation ground experiment:
The first conclusion,
that automated CD&R enabled by shared aircraft intent information has
no significant effect on airspace capacity, is unexpected. Potential causes
for this unexpected result might be found in the workload models obtained
from the part-task simulations. However, if the gain in airspace capacity
from shared aircraft intent information was a sufficiently large one, it
might be reduced but would not be eliminated entirely by potential errors
in the workload models. Therefore, conclusion 1 might very well be true.
The airborne experiment was built around the premise that aircrews assisted by intent-enabled automated CD&R tools (rather than ground-based controllers) could have the primary responsibility for maintaining safe separation. Within this context, the following conclusions were drawn from the fast-time simulation airborne experiment:
Another interesting result was found regarding flight efficiency. Manoeuvres to resolve conflicts impact on flight efficiency by causing aircraft to fly extra distance or to fly at non-optimal flight levels. They increase the distance flown, the time airborne and the fuel consumed, and both ground and air fast-time simulation experiments collected statistics on these increases. The main conclusions are as follows:
The results of this
study are detailed in D2-3.
The aim of the
ground full simulation experiments was to validate the outcome from the
fast-time simulations and the part-task simulations. In order to do this
it was decided to test the CD&R tool in unstructured airspace only,
on a sector different from the ground part-task sector. An analysis was
performed on the subjective questionnaires and the controller workload
models have been checked.
From the results, it can be concluded that traffic load is the dominant factor for controller workload. The CD&R tool appeared to have no significant effect on controller workload. This is surprising and not in line with findings in the ground part-task simulations. Given the unbalanced experiment matrix in favour to the sessions with CD&R tool, it was to be expected that the runs with CD&R tool would be more acceptable, with less workload. The fact that this effect can not be confirmed from the ground full-scale simulations might be explained as follows:
Debriefing results show that CD&R tool is more useful to the planning controller, however the conflict detection part of the tool is more appreciated than the resolution part. It is suggested to improve the conflict detection part and suspend the work on the resolution part. Furthermore, the human-machine interface should be improved, especially the aircraft label cluttering should be solved.
Familiarity with the simulated airspace and the new systems is an important issue for controller workload, as indicated by the controllers in the ground full-scale simulations. Training of controllers on a specific sector is therefore very important and needs more attention and time in future experiments.
The workload models developed during the ground part-task simulations appear to be correct, except for the CD&R tool effect in the planner controller workload model. Since this effect was already very small in the controller workload model used in fast-time simulations, it is judged that this does not significantly change the overall fast-time simulation conclusions.
Although the results of the ground full simulations are not completely in line with the results of the ground part-task simulations regarding CD&R tool effect on controller workload, they are in line with the results of the fast-time simulations. Fast-time simulations found that the use of automated CD&R tools enabled by shared aircraft intent information has no significant effect on workload for either planning controller or the tactical controller, and hence cannot be expected to increase the airspace capacity of systems built upon ground-based control concepts. In this respect, the ground full simulations have validated the fast-time simulation results. Moreover, this gives reasons to believe that the difference between ground part-task simulations and ground full simulations is indeed caused by the training effect in the part-task simulations, which was less the case in the full simulations.
The ground full simulations have found that controllers are operating at their limits at traffic density 1.5. For this reason, the fast-time simulation conclusion that controllers can safely handle traffic densities of 1.7 in unstructured airspace with CD&R tool should be weakened to traffic density 1.5.
The results of this
study are detailed in D2-3.
The aim of the
airborne full experiments was to validate the outcome from the fast-time
simulations and the part-task simulations. In order to do this it was decided
to test the ASAS concepts with 10 minutes intent information and 5 minutes
state information in the climb and descend phases of flight. Five professional
crews participated in the experiment and each of them experienced all combinations
of ASAS concept, flight phase and traffic load. An analysis was performed
on the subjective questionnaires and the pilot workload models have been
checked.
The following main conclusions can be drawn from the results of the experiment:
Most results from the airborne part-task simulations have been confirmed, however the main conclusion from the airborne full simulations that the ASAS concept has no significant effect on pilot workload and acceptability is opposite to the airborne part-task findings, where intent-based CD&R was clearly showing better results than state-based CD&R. Debriefing results and mean ISA rating analysis confirm the airborne full-scale result on the ASAS concept.
This effect might be caused by the fact that the crews had to fall back to state-based CD&R while in intent-based CD&R runs. In several intent-based CD&R runs the crew had to fall back to state-based CD&R due to unavailability of intent-based resolutions. The traffic density and complexity did not leave room for intent-based resolutions.
The fact that intent-based CD&R did not always provide resolutions is not surprising, given:
As indicated in the debriefings, pilots argued that the intent-based CD&R should always provide a solution. However, this solution might have to include climbs during descend, or other illogical manoeuvres, which pilots find unacceptable as also found in the debriefings.
This aspect shows the principle differences between state-based CD&R and intent-based CD&R. It shows that intent-based CD&R as such might be preferred by pilots, but the physical limitations for finding intent-based resolutions is more limited with intent-based CD&R, compared to state-based CD&R. System complexity and physical limitations in finding intent-based resolutions might reduce the interest in airborne intent-based CD&R.
The workload model validation also confirms the airborne full-scale simulation results on the ASAS concept. The workload model for the pilot flying the aircraft (which did not include the concept effect) was validated, but the direct operational concept effect for the pilot not flying the aircraft is invalidated. The effect of the number of conflicts was clearly confirmed in the workload model analysis and also from the questionnaire data.
The significant flight phase effect is surprising, with lower workload in descend phase of flight and better acceptability in descent, compared to climb phase of flight. It was expected that descending, converging flights would generate more workload than climbing, diverging flights. This effect might be explained by the fact that in the climb, the crew had to reach a certain busy cruise flight level (FL330). Given the traffic densities, reaching this flight level was not always easy, since there was a large concentration of aircraft around this target cruise level. During descend, the crew only had to cross the busy cruise levels around FL330, descending from FL390. There were converging flights inbound Frankfurt, but they came from various directions. The concentration of aircraft inbound only took place mainly near the inbound fix to the Frankfurt TMA, where the experiment run was ended.
In fact, we could
say that climbing flights also have a convergence effect, namely a vertical
converging effect near the popular cruise levels. A descent has mainly
a horizontal convergence near the Initial Approach Fix. Apparently the
vertical convergence near cruise level can not be underestimated.
The fast-time simulations
concluded that workload scores for both flying and non-flying pilots increase
with traffic density. Furthermore, at any given traffic density, workload
scores for both flying and non-flying pilots are less with intent-based
CD&R than with state-based CD&R. Given the results from the full-scale
simulations, where the ASAS concept does not have a significant influence
on workload and acceptability, the fast-time simulation results with respect
to the differences in ASAS concepts should be weakened.
When taking into account the full simulation results on acceptability ratings with traffic density 3 and the peak workload analysis, the fast-time simulation conclusion that pilot workload is still below limits at traffic density 6 should be weakened to traffic density 3. The fast-time simulations have extrapolated the pilot workload models up to traffic density 6, but given the findings in the full simulations it is not certain if this is valid.
One of the interesting fast-time simulation results is the indirect reduction of pilot workload with increased look-ahead time, caused mainly by a better distribution of aircraft in the area. This result remains valid, even without the direct ASAS concept effect included in the workload models. For this reason, intent-based CD&R might be preferred over state-based CD&R with clear macro effects on the distribution of traffic over the area.
The results of this
study are detailed in D2-3.
The primary difference
between the current air traffic system and an ‘intent’-based airspace organisation
is the sharing of aircraft ‘intent’ information. The ‘intent’ information
is generated by aircraft in the system, or those just about to enter the
system, with other aircraft and any ground based traffic control function.
Who shares with who and level of information shared is a function of the
system architecture. The underlying goal is the maximisation of air traffic
capacity. To achieve this, optimisation must be made of the airspace available
and therefore aircraft tracks need to be conflict free. The performance
attributes associated with ensuring that aircraft tracks are conflict free
include the required precision/resolution of manoeuvres to remove conflicts,
availability and accuracy of ‘intent’ data etc.
From a functional analysis perspective, there are many combinations of solution architecture that could be put in place. There are two main implementations, namely a 'centralised' function responsible for resolving the conflicts (ground based or some other ATM style implementation), and a distributed architecture (aircraft based) where the aircraft have responsibility for the resolution of conflicts and achieving safe stable airspace. Further, it was identified during the function analysis that there may be the possibility of a hybrid system where some functions are performed centrally while others are performed in a distributed way.
However, there are a number of generic issues which will impact all of these options and which must first be considered and resolved before analysing the relative merits and difficulties of the potential implementation. Therefore the functions required to utilise ‘intent’ information and achieve conflict free airspace that is operating in a stable fashion were first analysed.
In the analysis, no assumptions about where the functions are performed or what systems are involved were made. However, one essential assumption is that the aircraft will maintain its predicted ‘intent’ path to the level of accuracy required by the air traffic system. The aircraft must also produce its own predicted trajectory/’intent’ message and broadcast it as no other element of the system would have the data to do this.
The function allocation process identified three potential ATM systems:
The CD&R functions have been analysed in all three potential ATM systems. From this analysis, it was concluded that there are two promising systems to look into further:
The results of the
function analysis and allocation task are detailed in D2-4.
The results of
the simulations and function analysis and allocation have been assessed
and technologies have been analysed required for the implementation of
these functions. An implementation roadmap is produced for both ground
based and airborne equipment.
In order to derive this roadmap, the existing, state-of-the-art and currently developed ground based equipment and processes were investigated to identify the ability to generate and process the aircraft intent information as identified to be required in the new ATM system. The following ground based equipment has been addressed among other CNS systems:
Similarly, the
following airborne equipment was addressed:
These descriptions
were used as input for the implementation roadmap and the conclusions and
recommendations.
In defining the implementation roadmap, the architectures of both interesting ATM systems (hybrid and airborne) have been assessed and separate implementation roadmaps for both systems have been defined. These implementation roadmaps are summarised in the figures below.
Figure: Ground Implementation
Roadmap.
Figure: Airborne
Implementation Roadmap.
As can be seen from the implementation roadmap, emphasis on the ground will be on the following systems for CD&R with intent information for the hybrid system:
For the airborne side in the airborne system, the required changes identified focus on:
The results of the
implementation roadmap are detailed in D3-1.
Workload models for controllers and pilots have been derived from the part-task simulations and used in fast-time simulations. The controller workload model is dominated by the number of aircraft in the sector, the pilot workload model is dominated by the number of conflicts.
Part-task and fast-time simulations have found that unstructured routes do not only provide better flight efficiency, as expected, but also lower controller workload, especially when the ground CD&R tool was available. This indicates that sector size and characteristics (military areas, route structure) may be a factor for the successful introduction of advanced CD&R tools for controllers.
Fast-time simulations have shown that systems based on intent information are more efficient in terms of time, distance and fuel than systems based on only state information, both in ground and airborne concepts. This suggests that exchanging aircraft intent information appears to have no benefit to airspace capacity although it might be very beneficial from a flight efficiency point of view. Moreover, exchanging aircraft intent information for traffic separation assurance might be very valuable from a safety perspective, both in ground and airborne concepts.
The comparison between the results for controller workload and those for pilot workload, as tested in the project, is interesting: whereas controllers become overloaded at about 1.5 times the summer 2000 traffic density, pilots are still not overloaded at 3 times this density. This comparison suggests that, in the long term, ATM systems based on concepts where aircrews have the primary responsibility for separation are likely to offer several times the capacity of those based on ground control concepts. Data on flight efficiency shows that additional fuel rather than pilot workload will be the factor that ultimately determines the traffic handling capacity of systems based on airborne separation concepts.
The figure below shows the overall airspace capacity figures for the various concepts tested.
Figure: Airspace capacity results.
The function allocation process has assessed three potential ATM systems:
The CD&R
functions have been analysed in all three potential ATM systems. From this
analysis, it was concluded that there are two promising systems to look
into further:
Finally, an implementation roadmap was derived for both the airborne and hybrid system. It was found that emphasis on the ground will be on the following systems for CD&R with intent information for the hybrid system:
For the airborne side in the airborne system, the required changes identified focus on: