An Open Synthetic Operational Aircraft Performance Model (SOAP)

Modern aircraft rely on cyber-physical systems to monitor flight performance and ensure safe and efficient operations. Specialized Aircraft Performance Monitoring (APM) systems identify airplanes with deviant fuel consumption due to engine deterioration and increased drag. Developers of these systems face several data-related obstacles. First, access to operational data is often restricted due to proprietary concerns. Second, available datasets may lack the volume and variety of conditions needed for adequate training and testing. Finally, thrust and drag are not directly measurable in flight and must be inferred, making direct verification of APM predictions impossible. To address these combined challenges of availability, volume, variety, and verifiability, this paper introduces SOAP, a synthetic operational aircraft performance model that generates physics-based datasets for APM development. The model integrates industry-standard aerodynamic and engine performance formulations with stochastic parameterization to represent variability in aircraft condition and operational environment. Explicit variations in drag and per-engine fuel consumption embed ground truth for effects that are latent in realworld data. Operational factors, including altitude, ambient temperature, and speed, vary stochastically based on realistic operational patterns observed from ADS-B tracking. An example implementation demonstrates the model for a fleet of 35 Boeing 737-800 aircraft. The model is computationally efficient, enabling rapid generation of large, diverse datasets for developing advanced APM systems, including machine learning and Bayesian methods. By providing ground truth and scalable data volume, SOAP facilitates research that could detect subtle performance degradations that cost airlines billions annually.