ZeFMaP

ZeFMaP: Zero Failure Management at Maximum Productivity in Safety Critical Control Rooms.

SINTEF ICT

FREQUENTIS AG

amela.karahasanovic@sintef.no

Summary

The ZeFMaP project investigated if and how process improvement methods and tools coming from other domains can be used in the context of tower control rooms. We have proposed a four step productivity improvement process called ZeFMaP that includes the following:
• Step 1: Modelling the target process into a production workflow and dividing it into “production steps.”
• Step 2: Optimizing the “human machine symbiosis” for each step (outside the scope of the project).
• Step 3: Analyses of the decision points and decision content within each of the steps with the aim of optimization for each decision for the overall process and the improvement of each production step through a feedback loop.
• Step 4: Improvement of the target process through a feedback loop.

The ZeFMaP process was developed in an iterative manner. In the first iteration we surveyed process improvement approaches currently used in other industries and discussed the potential use of those approaches within an ATM setting. We then outlined the ZeFMaP process and applied Step 1 (modelling the target process) on the TWR control room process of the Hamburg airport. An improved version of the TWR workflow was proposed and implemented in the electronic flight strips tool.

In the next iteration we conducted a controlled experiment with five air traffic controllers using replay of recorded traffic samples from the Hamburg airport, and the NAVSIM air traffic simulator at the University of Salzburg. The results from the first experiment indicated that the Failure mode, effects and criticality method (FMECA) was of limited usefulness for improving productivity of TWR processes. The scenario that was used in the first experiment was a fairly easy task to perform for the experiment participants and the number of non-optimal decisions appeared to be pretty low, indicating that richer data sets are needed when conducting further investigations.

In the last iteration we conducted a second experiment, aimed at collecting richer input data to the ZeFMaP process. In similarity with the first experiment, the second experiment involved also a set of real-time simulation exercises where five air traffic controllers were subjected to realistic work scenarios that were played out in real-time using the NAVSIM simulator at the University of Salzburg. During the simulation exercises, the facilitators gathered qualitative and quantitative data, allowing for a detailed analysis and understanding of productivity and safety in the TWR control room. Compared with the first experiment, the second experiment involved a greater number of measured runs, and scenarios with higher air traffic loads. The data collected in the second experiment were used to evaluate the usefulness of holistic optimization and visualization within the third step of the ZeFMaP process.
The validity of the applicability of the overall approach (the ZeFMaP process itself) has been proven by implementing Step 1 and Step 3. The benefits achieved by applying Step 1 of the ZeFMaP process were:
- A better understanding of the process and sub-processes which could then be used to discuss alternatives directly with controllers and/or system developers
- The identification of value adding and non-value adding sub-processes which enables a concentration on value adding sub-processes
- Selection of key performance areas from the SESAR performance framework that was used for assessment of improvements
- The identification of responsibilities and hand over points showing improvement to potentially reduce the necessary communication and resulting in an optimized HMI
- The identification of decision points which are necessary input for Step 3 of ZeFMaP process
In addition based on the results gathered within the Step 3 we can recommend the application of workflow process modelling in the ATM domain.

In this regard we proposed and evaluated to tools to improve decision making within the tower control process (the overall process and the sub-processes). According to best of knowledge both of them introduce new concepts; the first one introduces workflow and performance visualization and the second one introduces holistic optimisation.

Our results indicate that workflow and performance visualization can be a useful aid to facilitate learning in TWR control rooms. We believe that the tool will be useful for analysis and improvement of the decision making process. It can also be used by the controllers for self-assessment of their performance at the end of a working day or for training of ATCO students, particularly in the sessions following training simulations. To increase the usefulness of the visualization tool, future versions should provide functionality for listening to the communication between ATCOs, and between ATCOs and pilots.

For further optimization we propose an integrated approach to departure management and surface routing in airports. Our results show that the integrated approach covers calculations and trade-offs probably outside of human capability when handling the Hamburg airport in simulated scenarios. The decrease in average taxi time was between 33% and 36% while punctuality improved with 57% to 67%. Still we need to make the model even more complete with respect to detailed real-world constraints. We need to incorporate the ability to maintain stable solutions in a dynamic environment. We also want to extend our model to fully include the arrival management. Such development would lead to a decision support tool covering both planning and real time support. Finally, this optimization technology can also be developed to become a part of learning tool for controllers that will provide a comparison with their decisions and the most optimal ones.

Furthermore, our results show that active operational use of optimization tools provides a number of direct economic savings together with greater flexibility, efficiency, overview and safety of the planning effort. The ZeFMaP project's experience with the overall process and the associated tools form a solid basis for our conclusions. However, there are several issues that were not addressed within the project. Future work includes investigating if additional tools can be useful within the ZeFMaP process. An investigation on how other ATM processes can benefit from the ZeFMaP approach is also highly relevant.

The interaction between Step 2 and Step 3 in the ZeFMaP process can be further investigated. Some dedicated variables or parameters in a heuristic algorithm can be tuned by input from human experience, analysis and judgment. On the other hand, the performance of ATCOs might be improved when an automated decision support tool is used.

ZeFMaP publications:

SID 2012 paper

SID 2011 paper

Close-up meeting presentation

Projects: