The ComplexityCosts project belongs to the 2nd WPE Call for Projects.
Consortium: University of Westminster ,Innaxis (Fundación Instituto de Investigación Innaxis),DLR (Deutsches Zentrum für Luft-und Raumfahrt)
Coordinator: University of Westminster
Project Leader: Andrew Cook (cookaj @ westminster.ac.uk)
The primary objective of ComplexityCosts is to better understand ATM network performance trade-offs for different stakeholder investment mechanisms in the context of uncertainty. The modelled investments are intended to mitigate the impact of uncertainty on performance. Embracing the non-linearities of complexity, thus significantly advancing the state of the art, the project’s stochastic, layered network model will include interacting elements and feedback loops. The project will analyse how the stakeholders’ interacting tactical and strategic cost structures affect system performance. ComplexityCosts thus seeks to quantify, and improve the understanding of, complex interdependencies that are often overlooked in trade-off models.
Introduction and problem statement
For every complex problem there is an answer that is clear, simple, and wrong. H L Mencken (1880 – 1956) ComplexityCosts will investigate innovative ways to embrace complexity in exploring the relationship between ATM performance and stakeholder investment costs. The relationship between stakeholders within the network and the costs associated with their investments is of a complex nature. For example, airlines adding further buffers to schedules and increased investment in ANSP infrastructure may both be intended to ultimately mitigate tactical delay. What is often overlooked, however, are the complex interdependencies of, and trade-offs between, such actions. By embracing such complexity, this project will significantly progress the state of the art in performance assessment.
Currently, approaches to the minimisation of costs to airspace users rely on the computation of an optimal balance between capacity and delay. However, the complexities of such cost assessments need to be taken into account. The efficiency of operational processes, their robustness to tactical variability, and the interactions between the stakeholders, all contribute not only to local impacts, but also to effects elsewhere in the network – hence multiple temporal and spatial scales need to be examined. Only through understanding these effects, including the propagation of negative impacts (such as aircraft delay) through the network, can the full cost impacts be properly assessed. The complex spatial distribution of such costs and of the corresponding benefits among stakeholders, whereby many different elements interact, means that linear models are incapable of capturing the complexity of the cost structures across the stakeholders. Analytical tools have not kept up with the challenge of identifying and assessing such complexities as reflected through multiple interdependencies, over multiple temporal and spatial scales.
Project objectives and expected results
The primary objective of ComplexityCosts is to better understand ATM network performance trade-offs for different stakeholder investment mechanisms in the context of uncertainty. The modelled investments are intended to mitigate the impact of uncertainty on performance. The project will analyse how the stakeholders’ interacting tactical and strategic cost structures affect system performance. ComplexityCosts thus seeks to quantify, and improve the understanding of, complex interdependencies that are often overlooked in trade-off models.
ComplexityCosts will advance the state of the art, moving this forward from: • influence diagrams, which typically rely on expert judgement; • simplistic assessments, such as mapping medians onto populations; • assessing only the cost of fragmentation, which does not take account of complex interactions. Instead, ComplexityCosts will afford fuller insights into the costs of complexity, with key outcomes: • furthering the understanding of interacting elements and feedback loops, thus taking account of non-linearities; • exploring the propagation of negative impacts through the network, and the underlying cost-driven mechanisms; • evaluating different strategies for achieving the same outcomes and goal(s), such as delay reductions. In alignment with the complexity science theme, key characteristics of complex systems will be considered, as follows: • multiple scales – by examining the impacts of local mechanisms elsewhere in the network (in time and space); • non-determinism and uncertainty – by assigning likelihoods to different performance evolutions, whereas most previous work has been linear and ultimately deterministic; • response to disturbance, system robustness and resilience – by demonstrating the flexibility afforded to the stakeholders under different (initial) operating outcomes; • potential emergent behaviour – by modelling the stakeholder response to unexpected (positive or negative) outcomes accompanying expected change.
A number of stakeholder investment mechanisms will be proposed. These mechanisms are ways of putting money into the ATM system, with a planned benefit for at least one stakeholder. Such benefits are then assessed in terms of the performance impact and corresponding trade-offs. The investment mechanisms explored will be realistic and selective. Some will challenge established conventions, representing more of a paradigm behavioural shift – including strategies that may appear politically unlikely. A (conventional) example investment for an AO would be building more buffer into a schedule. An example for an ANSP would be a technology investment allowing higher sector throughput. These cases both represent strategic cost investments with anticipated tactical (delay) cost savings. Some investments will be discrete for given stakeholders (e.g. cockpit solutions for three AOs), others might be explored as a package (e.g. cockpit solutions with improved slot management). Costs will be modelled according to differential characteristics of stakeholders, for example with adoption rates formulated as a function of airline business models and ANSP ownership structures. The impact assessments will be evaluated using a layered network model, which incorporates uncertainty. There is a close relationship between the types of uncertainty modelled (for example tactical delay) and the investments mechanisms (such as schedule buffer), since the investments are designed to cope with disturbance and offset the cost thereof. One investment mechanism may be designed to mitigate against multiple sources of disturbance (such as, indeed, the components of tactical delay). Different types of disturbance will be applied under various model scenarios. Care needs to be taken when defining the scenarios to ensure that they are internally consistent, also in the context of the investment mechanisms, and representative of potential future stresses placed on the ATM network (such as traffic density relative to capacity). The combinations of scenarios and investment mechanisms will be chosen so as to furnish the most insightful trade-offs. There will necessarily be a delicate balance between generating too many and too few model outcomes to explore. Whilst it is not possible to forecast the exact evolution of traffic or costs, ComplexityCosts will, for the first time, take into account both uncertainty and the major complexities of the ATM network, thus providing a far more useful and holistic description of the costs structure and corresponding trade-offs of European ATM than has hitherto been available through classical analytical (e.g. spreadsheet-based) tools. The project will begin with a review of the state of the art, to ensure that full account is taken of current R&D, with particular regard to: investment assessment and trade-off analyses; the developing SES context; and, available data and forecasts, including the recently revised long-term forecasts within EUROCONTROL (e.g. ‘Challenges of Growth’ series of studies, focusing on 2030 (2035) and 2050), and those external to EUROCONTROL. A vertical activity, maintained throughout the project, is the data management process, which will ensure quality (clean) data are delivered on-time to the required modelling and design processes. The three research workpackages may be summarised as follows.
WP1. Design of investment mechanisms
This workpackage will develop the cost models; design the investment mechanisms, and provide supporting data management to the project. Strategic cost allocations will draw on and update the consortium’s existing research (e.g. regarding the strategic cost of adding buffer to schedules) and also on published work (e.g. PRU’s ANS costs per service unit, by member state). These cost estimations will re-assess common assumptions of linear relationships at aggregate levels (e.g. provision of capacity and cost thereof, e.g. through marginal provision in low density sectors). The strategic costs will be deterministic. Tactical costs (typically manifested through delay impacts on stakeholders) will be assigned to the performance outcomes, ensuring that these function within the model at both the stakeholder and nodal level. These will be partly stochastic to reflect operational uncertainty. Such costs to the airline will include fleet, fuel and carbon, crew, maintenance, and passenger ‘hard’ and ‘soft’ costs and will published for independent use by industry and researchers.
WP2. Cost, network and resource modelling
This workpackage will develop and implement the core model in the context of uncertainty, by building a stochastic, layered network model with, but not limited to, cost structure layers and infrastructural layers (distinctly mapping flights and passengers, for example). Although independently definable, the layers will interact with each other and individually evolve through time; these processes will be modelled by making use of event-driven simulation techniques. The network modelling process can be described by three fundamental steps: (i) defining the general model and layers (e.g. stakeholder and agent classifications); (ii) defining the parameters and indicator types (e.g. deterministic, stochastic, logical) to be used to measure the investment mechanism impacts; and, (iii) designing the integration of the layers and the influence model (e.g. using complex network techniques to link parameters to metrics). With the influence model we will try to limit the number of parameters that affect each metric (probably by fixing some of them). Such dependencies are ultimately explored in WP3.
WP3. Holistic investment trade-off analyses
This workpackage designs the model’s scenarios (for example with respect to traffic growth and future airport capacities), in close alignment with the cost structures of WP1 and the model configuration in WP2. This workpackage also designs and performs the trade-off analyses. In the first phase, static properties, then dynamic aspects will be configured to produce a deterministic model. Then the stochastic properties of the ATM system will be incorporated, in readiness to run the baseline scenario. Disturbance will also be simulated as an input. In the second phase, the investment mechanisms will be used to co-define particular scenarios, and relevant KPIs will be selected and evaluated. For a given investment mechanism and scenario, simulations will be run to assess the KPI trade-offs. It will be necessary to align the description of the layered model (both the parameters and the metrics) and the scenarios that build on the model. It is also possible that a parameter in one scenario will act as a metric in another scenario.