Engine Radiated Noise Prediction Modeling Using Noise Source Decomposition and Regression Analysis

An engine's radiated noise level is a very important attribute required for delivering customer satisfaction. Having an accurate radiated noise prediction capability during the planning, target setting, and initial design phases is critical to making the up-front decisions that enable the timely and cost efficient delivery of an engine that meets its radiated noise goals.

This paper describes a simple radiated noise model that is based on a combination of regression modeling and simplified analytical modeling. The regression model uses measured data from multiple tests that can be broken down to noise sources such as mechanical, combustion, and accessory components. The simple analytical models are used to determine the parameters that the decomposed noise data is regressed against.

The model developed in the paper is then compared to previous models suggested in the literature and to measured data from engines. The model is also applied to the task of radiated noise “futuring”, predicting what the competitive level of radiated noise will be at a future date when a new engine comes to the market.