پديد آورنده :
ايوبي، محسن
عنوان :
استفاده از شبكه هاي عصبي مصنوعي در مدل فليملت آرام جهت مدل سازي احتراق گستره اي قابل توجه از شعله هاي غير پيش مخلوط مغشوش
مقطع تحصيلي :
كارشناسي ارشد
گرايش تحصيلي :
تبديل انرژي
محل تحصيل :
اصفهان:دانشگاه صنعتي اصفهان،دانشكده مكانيك
صفحه شمار :
چهارده،124ص.:مصور،جدول،نمودار
يادداشت :
ص.ع.به فارسي و انگليسي
استاد راهنما :
محسن دوازده امامي
استاد مشاور :
احمد صابونچي
توصيفگر ها :
احتراق غير پيش مخلوط , سينتيك شيميايي , تابع دانسيته احتمال
تاريخ نمايه سازي :
25/2/89
استاد داور :
علي اكبر عالم رجبي،محمد دانش
تاريخ ورود اطلاعات :
1396/09/28
چكيده فارسي :
به فارسي و انگليسي: قابل رويت در نسخه ديجيتالي
چكيده انگليسي :
Usage of ANN Method in Laminar Flamelet Model for the Modeling of Turbulent Diffusion Flames Mohsen Ayoobi m ayoobi@me iut ac ir Date of Submission 2010 3 15 Department of Mechanical Engineering Isfahan University of Technology Isfahan 84156 83111 Iran Degree M Sc Language Farsi Mohsen Davazdah Emami mohsen@cc iut ac ir Abstract The complexity of combustion phenomena caused by the interaction between chemical kinetics and turbulence has been investigated in this thesis using Laminar Flamelet model along with the Artificial Neural Networks technique with the least possible simplifications The application of Flamelet model provides the possibility of the usage of every complicated chemical mechanism However in most of other models because of the simultaneous solution of kinetics and turbulence researchers have to implement reduced mechanisms to prevent the increase in requisite CPU time and memory resulting in more discrepancies Another point in this work is the application of ANNs as a shortcut between the solutions of chemical kinetics and turbulent fluid flow This has yielded a conspicuous reduction in the time consumption of solution without any defect on the preciseness of results In this work first of all a data bank is built using the solution of opposed diffusion flames in different strain rates Then in that data bank species mass fractions and temperatures are related to mixture fractions and scalar dissipation rates which are calculated from the information existing in the aforementioned data bank In the next step for the consideration of turbulence effects numerical integration has been done using probability density functions PDF As a result of numerical integration the favre average amounts of thermo chemical properties of flow in various amounts of mean mixture fractions mixture fraction variances and mean scalar dissipation rates can be achieved Now a Flamelet library has been built over which an ANN can be built Weights and biases of the already constructed ANN have been implemented in the CFD code to predict the values of species mass fractions and temperature in different locations of the solution domain Consequently in the CFD code there s no need for the solution of every single species transportation equation That s why a considerable reduction in CPU time has been observed It s noteworthy that important minor species including some pollutants has been analyzed as well and an acceptable agreement has been observed comparing with experimental results Moreover the effect of different probability density functions has been investigated in the numerical integration It has been perceived that the application of Log normal PDF as a function of scalar dissipation rate yields better agreement in comparison with the usage of Delta PDF as a function of scalar dissipation rate However those slight differences between already mentioned results can be neglected because of more time consumption in the former state Key Words Non Premixed Combustion Artificial Neural Network Laminar Flamelet Model Turbulence Chemical Kinetics Probability Density Function
استاد راهنما :
محسن دوازده امامي
استاد مشاور :
احمد صابونچي
استاد داور :
علي اكبر عالم رجبي،محمد دانش
