Measurement-driven aircraft noise modelling

Abstract

Aviation continues to grow worldwide, bringing increasing challenges related to aircraft noise. Current noise assessments rely on best-practice models such as the European standard Document 29 (Doc 29) of the European Civil Aviation Conference, which is also used in the Netherlands. Given the strong dependence of airport policy and regulation on these models, their accuracy and reliability are essential. The objective of this thesis is therefore to improve the accuracy of the Doc 29 aircraft noise model using noise measurements collected around Amsterdam Airport Schiphol. The NOMOS monitoring network, consisting of 41 Noise Monitoring Terminals (NMTs), provides the measurement basis for several validation and model-improvement studies described in this thesis.

Best-practice noise models use Noise-Power-Distance (NPD) tables, which provide noise levels as a function of engine power (thrust) and distance between aircraft and observer. Additional correction factors allow predictions under non-standard conditions. While computationally efficient, these methods rely on simplifying assumptions about aircraft operations, source noise characteristics, and propagation effects.

The first step of this research is to identify the challenges and limitations of current noise predictions and measurements. A comparison between annual noise metrics (Lden and Lnight) from Doc 29 and NOMOS data reveals noteworthy trends: Doc 29 generally underestimates Lden, but overestimates Lnight. Background noise and undetected flights were quantified and found to predominantly affect low-noise regions at greater distances from flight paths. This analysis also highlights the strong dependence of model accuracy on the correctness of aircraft performance input parameters.

To address this, a second research stage develops a method for accurately determining engine thrust for historic flights. Using either acoustically derived N1 values obtained from blade-passing frequencies or directly from onboard flight data, estimates of delivered thrust are reconstructed. Since Doc 29 requires corrected net thrust as input, conversion methods were evaluated. Simulations using the Gas turbine Simulation Program (GSP) provided the most accurate results. Based on these simulations, coefficients were derived and validated to reliably estimate thrust from N1 for use in standard best-practice formulas. Using this improved input, a validation study showed strong agreement between modelled and measured Sound Exposure Level (SEL) and LA,max, though residual discrepancies remained.

To investigate the limitations of the NPD tables, back-propagated measurements were used to derive thrust-noise relations. Results indicate that for departures, the existing NPD tables overestimate the thrust-noise dependency. New NPD tables were therefore constructed directly from measurements, fully independent of the existing database. Applying these calibrated NPD tables improved departure noise prediction accuracy by approximately 25%, independently verified with measurements from Oslo Airport.

This thesis also assessed the use of standardised source spectra to calculate weather-dependent corrections. Due to the simplified atmospheric assumptions, the absorption of sound during propagation is limitedly affected by the source spectra. The use of these source spectra is thus sufficient for this purpose.

A final step is to benchmark the Doc 29 model to an independent empirical model, sonAIR. With more complex noise source modelling and propagation algorithms, sonAIR showed improvement compared to the measurements, specifically at lower elevation angles and in low-noise regions.

Overall, this thesis contributes to the validation and improvement of best-practice aircraft noise models through detailed analysis of noise measurements around Schiphol Airport. While Doc 29 performs well for estimating average noise exposure, its simplified assumptions about source noise and propagation limit its ability to represent the full complexity of aircraft noise. The improvements developed in this research contribute toward more reliable and transparent noise assessments.