Proactive Improvements of Civil Infrastructure Operations Based on Data-Driven Simulation and Optimization

Efficient and safe operations are critical to civil infrastructure systems (CIS) such as airports. Currently, the operation performance is highly dependent on experienced human operators. For example, the expertise of air traffic controllers in runway assignments greatly impacts on the safety and efficiency of airport operations, as it impacts the overall traffic flow on the surface of airports. The large amount of operation data generated by the recent monitoring technology provides the potential to use historical data to guide less-experienced human operators in achieving high performance. However, current research falls short in adequately accounting for the uncertainties in operations and automatically utilizing historical data to optimize operations for safety and efficiency. For example, in metropolitan airports, the efficiency and safety of ground traffic depend on the experiences of air traffic controllers (ATC). Improper handling of traffic flows could increase collision risks and flight delays, progressively impacting air traffic networks. ATC must be skillful in assigning a runway for each aircraft while observing the historical and real-time air traffic data (e.g., Airport Surface Detection Equipment, Model X (ASDE-X)). However, due to the uncertainty and contextual changes in operations, the lack of guidelines for adjusting runway assignments in different scenarios makes it hard for less experienced ATCs to make effective adjustments. Automatically planning and designing safe and efficient runway assignments with uncertain and changing traffic is a practical challenge.

This research aims to establish a framework for automatically interpreting the historical recorded operational data into optimal control strategies, focusing on the critical runway assignment strategies for assisting human operators (i.e., ATC) with air traffic monitoring data. Especially this research conducted a data-driven simulation augmented by multi-objective runway assignment algorithms in airports that can proactively guide proper uses and real-time improvements of runway assignments in various traffic scenarios for multi-objective air traffic management. Two specific objectives include: 1) establishing a data-driven method that uses historical data (e.g., ASDE-X) for constructing operation simulations (e.g., airport traffic simulations); 2) exploring the strategies for multi-objective runway assignment under changing and uncertain traffic scenarios. The proposed research consists of three steps, as shown in the following: 1) transfer the large-scale and noisy historical data into a meaningful operation process model with traffic parameters and events derived from the data. This transformation would help to build the foundation of the simulation models of aircraft ground traffic operation; 2) build and validate the data-driven trajectory-based simulation model for airport tower operations with the historical data from the ASDE-X data. The comparison between the aircraft taxiing performance predicted by the simulation and real-world records will help to identify the validity of the data-driven airport tower operation simulations; 3) examine a multi-objective optimization algorithm that explores runway assignments under changing traffic scenarios. An optimal runway assignment should achieve multi-objective airport operation (e.g., maximum throughput, minimum fuel consumption) while minimizing the operational risks (e.g., conflicts in taxiing) under uncertainties.  

The developed trajectory mapping approach can consistently achieve above 95% matching accuracy for moving aircraft on the ground. The average required computational time for each aircraft’s trajectory is less than 2.8 seconds. The developed simulation model can effectively simulate the traffic on the ground at airports. The correlation coefficient value between the simulated taxiing time of aircraft and real-world records is 0.90, and the average difference between the average simulated taxiing time of aircraft and real-world records is around 79 seconds. The developed multi-objective optimizations for runway assignments can efficiently reduce the number of conflicting taxiing operations and the total fuel consumption and increase the hourly airport throughput by 71%, 38%, and 44%, respectively, under the changing traffic settings with speed and schedule variations. This research is expected to contribute to the progress of technologies for automated civil infrastructure operations, particularly in airports. The data-driven simulations and optimization approach developed in this study can be readily customized to suit different airports.