Taşıt Hava Emiş Sisteminin Akustik Davranışının İncelenmesi

Abstract

Yüksek mühendislik bilgisi gerektiren ürünlerin tasarım sürecinde kullanılan analiz yöntemleri, bazı durumlarda yetersiz kalmakta ve tasarımdaki bilinmeyen sayısını arttırmaktadır. Bu nedenden ötürü, çeşitli analiz yöntemleri geliştirilmektedir. İstatistiksel Enerji Analizi yöntemi de, ürünün (genel olarak hava, uzay ve kara araçlarında uygulanmaktadır) modelinin kurulması, sistem ile ilgili parametrelerin belirlenmesi ve sonuç olarak da sistem cevabının oluşturulması prosesine “istatistiksel” bir bakış açısıyla yaklaşmaktadır. Alışılagelmiş yöntemler düşük frekans bandı için oldukça sağlıklı sonuçlar vermekte ve yapılan testler de bu analiz sonuçlarını doğruılar niteliktedir. Fakat çalışılan sistemlerin yüksek frekans bandındaki dinamik cevabı bilinmek istendiği zaman, bu alışılagelmiş analiz yöntemleri yetersiz kalmakta ve analiz sonuçları yüksek frekans bandında yanıltıcı olabilmektedir. İlk olarak uzayy araçlarında yapılan çalışmalarda ortaya çıkan “yüksek frekans bandı tahriği” ve incelenen sistemlerin bu frekans bandındaki hassasiyetleri, yeni bir yöntemin doğmasını sağlamıştır. Uzay araçlarının düşük kütleye sahip olması ve buna ragmen direngenliklerinin yüksek olması, yüksek frekans bandı tahriğine karşı bu araçları oldukça hassas kılmaktadır. Sistemlerin bu frekans bandındaki davranışı, birkaç bilim adamının dikkatini çekmiş, sonuç olarak yeni bir analiz yönteminin doğmasına neden olmuştur. 1960’ların başlarında, R.H. Lyon ve P.W.Smith Jr. isimli iki bilim adamı tarafından birbirlerinden habersiz bir şekilde, mekanik sistem modellerinin kurulması ve sistem cevabının hesaplanması amacıyla geliştirilen İstatistiksel Enerji Analizi yöntemi, zamanla diğer bilim adamlarının da katkısı ile diğer analiz yöntemlerine alternatif haline gelmiştir. Bugün uzay ve hava araçlarının tasarımından, otomobil gibi kara taşıtları kabin içi akustiği ve inşaat sektöründe bina içi akustiği tasarımına kadar çok çeşitli mühendislik uygulamalarında kullanılmaktadır. Yöntemin ismindeki İstatistiksel kelimesi, sistemin bazı parametre değerlerinin, istatistiksel olarak belirlenmesinden kaynaklanmaktadır. Enerji kelimesi ise, analizi yapılan yapının ivme, hız, deplasman gibi sistem parametrelerinin enerji ifadeleri kullanılarak oluşturulmasından kaynaklanmaktadır. Analiz kelimesi, bu yöntemin mühendislik problemlerinin çözümü amacıyla oluşturulmuş bir yol olduğunu belirtmektedir. Bu tez dahilinde, bir ağır vasıta hava emiş sisteminin dinamik davranışı, istatistiksel enerji analizi bakış açısıyla incelenmiştir. Genel olarak istatistiksel enerji analizinde, oda içi akustiği ve büyük yapıların dinamik davranışı, rahat bir şekilde modellenmekte ve analiz edilmektedir. İncelenen ağır vasıta hava emiş sistemi, küçük yapı elemanlarından oluşmaktadır, bu durumdan ötürü sistemin düşük frekans dinamik cevabı pek güvenilir değildir. Bu nedenle analiz 500 Hz frekans bandından itibaren yapılmıştır. İncelenen sistemin dinamik cevabı yüksek frekans bölgesinde daha güvenilir olduğu için, bu bölgede oluşabilecek gürültünün etkinliği daha rahat bir şekilde tespit edilebilmiştir. Örneğin turbo kaynaklı gürültülerin etkinliği ve yayılma yolları tespit edilmiş ve bu problemin çözümü anlamında bazı aksiyonlar alınmıştır.

The name SEA was coined in the early 1960’s to emphasize certain aspects of this new field of study. Statistical emphasizes that the systems being studied are presumed to be drawn from statistical populations having known distributions of their dynamical parameters. Energy denotes the primary variable of interest. Other dynamical variables such as displacement, pressure, etc., are found from the energy of vibration. The term Analysis is used to emphasize that the SEA is framework of study, rather than a particular technique. Statistical approaches in dynamical analysis have a long history. In mechanics, we are most familiar with their application to the vibration that is random in time of the deterministic system. It is useful to emphasize here that the important feature of the SEA is the description of the vibrating system as a member of a statistical population or ensemble, not whether or not the temporal behaviour is random. Traditional analysis of mechanical system vibration of machines and structures have been directed at the lower few resonant modes because these modes tend to have the greatest displacement response in many instances, and the frequencies of excitation were fairly low. Of course, the vibration of walls and floors and their high frequency sound radiation have been of interest for a long time, but mechanical and structural engineers have been generally unaware of or unconcerned with this work. The advent of fairly large and lightweight aerospace structures, and their attendant high frequency broadband loads, has meant much more attention to higher order modal analysis for purposes of predicting structural fatigue, equipment failure and noise production. A characteristic of higher order mode analysis, however, is basic uncertainty in modal parameters. The resonance frequencies and mode shapes of these modes show great sensitivity to small details of geometry and construction. In addition, the computer programs used to evaluate the mode shapes and frequencies are known to be rather inaccurate for this higher order modes, even for rather idealized systems. In light of these uncertainties, a statistical model of the modal parameters seems quite natural and appropriate. If there is cause for statistical approaches from the nature of dynamical problem, there is equal motivation from the viewpoint of application. Mechanical and structural designers are often faced with making environmental and response estimates at a stage in a project where structural detail is not known. These estimates are made for the qualification of equipment and the design of isolation, damping or structural configurations to protect equipment and protect the integrity of the structure. Highly detailed analyses requiring specific knowledge of shape, construction loading functions, etc., are not appropriate. Simpler statistical analytical estimates of response to environment that preserve parameter dependence are appropriate to the designer’s need at this stage. There is experience in dealing with dynamical systems described by random parameters. Two notable examples that have served as “touchstones” in early developments of SEA are the theory of room acoustics and statistical mechanics. Room acoustics deals with the excitation of systems of very many degrees of freedom (there may be over a million modes of oscillation of a good sized room in the audible frequency range) and the interactions between such systems (sound transmission through a wall is an example). The analyses are carried out using both modal and wave models. The very large number of degrees of freedom is an advantage from a statistical viewpoint; it tends to diminish the fluctuations in prediction of response. Statistical mechanics deals with the random motion of systems with either a few or very many degrees of freedom. However, it is random motion of very special type, which we may call “maximally disordered”. In this state of vibration, all modes, whether they resonate at frequencies near each other or are far apart, tend to have equal energy of vibration and to have incoherent motion. The energy of the modes is, aside from a universal constant, the system temperature. The state of equal modal energy is spoken of as “equipartition of energy”. In SEA, we sometimes make the equipartition assumption for modes that resonate in the same frequency band, but not for all modes. Statistical mechanics, and its related science, heat transfer, also teach us that thermal (random vibration) energy flows from hotter to cooler systems and that the rate of flow is proportional to temperature (modal energy) difference. One advantage of statistical analysis of systems may be seen from the practical aspects of room acoustics. If one truly has million modes to deal with, even the m.s pressure associated with each, changing with time as the flute gives way to tympani, would be a hopeless mass of information to assimilate. What one does instead is to describe the field by a few coherent features of the modal pattern (direct field and a few reflections) and incoherent energy (reverberant field) totalled into a few frequency bands. Thus instead of million measures of the sound field (which would be incomplete in any case without the coherence data), we are able to describe the sound field effectively by 10-20 measures. The statistical analysis also allows for much simpler description of system, whether one describes the field by modes or waves. In the former case, modal density, average modal damping, and certain averages of modal impedance to sound sources are required. In a wave description, such parameters as mean free path for waves, surface and volume absorption, and general geometric configuration are required. The number of input parameters is generally in balance with the number of measures (10-20) to be taken. The most obvious disadvantage of statistical approaches is that they give statistical answers, which are always subject to uncertainty. In very high order systems, this is not a great problem. Many of the systems we may wish to apply SEA to, however, may not have enough modes in certain frequency bands to allow predictions with an acceptable degree of certainty. To keep track of this, we may calculate the variance as well as the mean, and also calculate the confidence for prediction intervals. In addition to hard and fast computational problems, there are certain difficulties in the psychology of statistical methods. A designer is not dealing with a gas of complicated molecules in random collision; he is concerned about predicting the structural response of a wing, for which he has engineering drawings, to a loading environment, for which he has flight data. Instead of following deterministic calculation (probably computer based) it is suggested that he will get a “better” estimate if he represents the wing as a flat plate of a certain average thickness and total area. His incredulity may imagined. But he must remember that his knowledge of the wing at the 50th mode of vibration may be just as well represented by the flat plate as it is by his drawings. Also, the answers he gets by SEA will be in a form that is usable to him, generally retaining parameter dependence that will allow him to interpret the effect of certain simple design changes on response level. With this brief explanations about the method, we can understand that we can predict also the response of small sized structures, which have a few modes in lower frequency bands. In this thesis, a heavy truck air intake system is the subject to the analysis. This intake system has got large tubes according to normal passenger vehicle intake system. However it has got not much modes in the lower frequency band, because of that reason analysis have been done from 500 Hz frequency band. Below that frequency band, results have got lower confidence levels. With the help of other analysis methods, some good correlated models have been created regarding to this air intake system. Those methods used to analyse low frequency band response of air intake system parts. For example finite element method has been used, in order to predict air intake system panel modes which are really important for radiated noise analysis. Also air intake orifice noise predicted using 1D analysis tools. These 1D models have been correlated with the help of orifice noise test. Test has been done in Eskişehir, on the proving ground. During tests, orifice noise has been measured. With the help of this orifice noise test data, 1D orifice noise models have been correlated. Radiated noise analysis has been performed in order to predict the noise radiation from the air intake system part surfaces. With this method, radiated noise directivity from the air intake system part surfaces can be also predicted. This information gives us the ideas of noise reduction. Mostly, surface treatment applications have been used in order to reduce radiated noise from surfaces. All of those analysis methods can be used to predict dynamic responses of th system for low frequency band. There is a need of high frequency method. This need caused to born a new method called statistical energy analysis. At higher frequencies confidence levels regarding to dynamical response are high, so that leads us to predict high frequency problems. For example, turbo related high frequency noises can be predicted with this method. Noise patterns and paths can be predicted and take an action to attenuate those noise levels.