ISSN 1813-1166. NAU Proceedings. 2008. №2 © Oleksandr I. Zaporozhets, Oleg O. Kartyshev, Gregory G. Golembievskiy, 2008 52 ENVIRONMENTAL PROTECTION UDC 626.517 Oleksandr I. Zaporozhets, D. E. Prof. Oleg O. Kartyshev, Candidate of Engineering (Russia) Gregory G. Golembievskiy, assoc. Prof. ACCURACY AND UNCERTAINTY OF AIRCRAFT NOISE MODELLING Various models for predicting aircraft noise around airports are described. Improved acoustic models of aircraft are formed by summing models for the noise sources peculiar to each of the aircraft types. Their accuracy and uncertainty are assessed by means of comparison with flight trials measurements. Описано різні моделі для прогнозу авіаційного шуму навколо аеропортів. Поліпшені акустичні моделі літаків створено за допомогою складання моделей для окремих джерел шуму, особливих для кожного з типів літаків. Їх точність і достовірність оцінено порівнянням з вимірюваннями, виконаними під час випробувальних польотів. Introduction The ability to assess and predict noise exposure accurately is an increasingly important factor in the design and implementation of any airport improvements [1]. Possible methods for modelling noise radiation, propagation and attenuation, include both analytical and semi-empirical results. The current tendency is towards less empirical and more analytical and numerical techniques. It should he noted that ICAO is carrying out analyses of existing models and methods for assessing the acoustical characteristics of the various sources associated with aircraft noise events and is making proposals for their use [2; 3]. Two approaches to analysis of aircraft noise phenomena have been defined and implemented in computer programs. The first approach is based on l/3-octave band spectra noise analysis of any type of aircraft in any mode of flight or during maintenance activities in the vicinity of an airport. It provides estimation of any type of aircraft noise criteria by means of set of noise spectra varied during the particular noise event or for any kind of noise exposure. The approach is implemented in a model and appropriate software NoBel. The second approach is based on the concept of "noise radius" and provides calculations of aircraft noise exposure units around the airports or at any noise monitoring point. The basic "noise radius"- relationships may be obtained from experimental data as well as by calculation (for example, by using the NoBel program). The task of deriving an acoustic model for each type of the aircraft under consideration has been proposed and solved in a manner that reconciles experimental data with calculation. Thus, the aircraft noise models, used in BELTRA solutions, are of sufficient reliability and accuracy. The second modelling approach has been utilized in software IsoBell'a. Here, the basic acoustic models for aircraft of any type will be examined on its accuracy and uncertainty. The acoustic model of an aircraft An aircraft is represented by a set of noise matrices, each dependent on flight mode and consisting of sound pressure level (SPL) spectra (in a l/3-oclave band form) for a defined number of directions of sound propagation from the acoustic source. In some cases the noise matrices are obtained experimentally, in others they are obtained by means of calculations based on the models for the particular acoustic sources [4-10] of interest for the aircraft under consideration. It is impossible to define the characteristics of all phenomena by means of analytical and semiempirical models only. The most common phenomena determining or influencing the accuracy of noise matrices are the engine installation effects and noise abatement treatments. Both experiments and calculations have some disadvantages and the derivation has been formulated to overcome them [1]. The sound pressure level spectrum {SPL jk ) of aircraft noise of any type in spectral bands N j , j=1,N j , and in some k-th direction of sound propagation, where k = 1, N k , with reference to previous considerations, can be defined by: jk jkp jk SPL SPL SPL    , (1) where SPL jkp is the predicted value of SPL jk resulting from a sum of particular models SPL jki for characteristic (or dominant) noise sources, i = 1,..,N s ;