Science
ACTUAL PROBLEMS OF AVIATION AND AEROSPACE SYSTEMS
Kazan Daytona Beach


Innovations in aviation and astronautics-2010

Scientific practical Conference of young scientists and students of the Moscow Aviation Institute

(April 26-30, 2010, Moscow, Russia)

Section «Applied mathematics and physics»

A.A.Puntus

Moscow Aviation Institute (State Technical University)

Russia

The scientific practical conference of young scientists, graduate students and students of the Moscow Aviation Institute, “Innovations in aviation and astronautics 2010,” was held on April 26-30, 2010. At this conference, the section “applied mathematics and physics”, where informative reports of students and graduate students of the department of applied mathematics and physics were presented, had a great popularity. The theses of given reports are presented below in this publication. Conference was dedicated to the 80th anniversary of the Moscow Aviation Institute. In the late twenties of the last century, because of the rapid industrialization of the national economy of the country and, in particular, with creation of domestic aviation, the issue about the need for the significant expansion of training engineering personnel was raised. As a result, by the order of the Supreme Economic Council of the USSR of March 20, 1930, on the base of the Aeromechanical Department of MVTU (the Bauman Moscow Higher Technical School), the Aeromehanical College (BAMY) was established. The college was renamed into Moscow Aeromehanical Institute, and later (in August of the same year), into the Moscow Aviation Institute (MAI). In MAI as well as in ВАМУ, there were three branches: aircraft construction, engine building and aeronautics. Since May 10, 1931, the institute moved to the department system of educational process.

On November 26, 2009, by the order of the Ministry of Education and Science of the Russian Federation, the Moscow Aviation Institute was established the category “National research university” for training specialists in all areas of design and testing of aviation, space and rocket engineering. This category indicates the priority of science, education, high technologies and culture. The new status of MAI is not only a recognition of the undoubted university merits, but also a great responsibility of the university staff and motivation for hard and creative work. The presence of qualified personnel and internationally recognized  scientific schools, the long-standing traditions of MAI, which guarantee training of highly qualified specialists, preservation and development of teaching and laboratory facilities make it possible to look with optimism into the future of the leading aerospace university of Russia. MAI today is a leading aviation university in Russia, a multidisciplinary educational and scientific center, which provides training of specialists of broad qualification for all branches of aviation and rocket-space science and industry. The objective facing the members of the institute for implementation of the program of further development of MAI, consists in transition to a qualitatively new level of training specialists, which is required by hi-tech industries in the conditions of the modern innovative economy, in other words, in creation of a unified educational environment “university-research centers-industry”.

As a result of the conference on April 26-30, the authors of the best reports are recommended to participate in the second International Scientific and Practical Conference “Scientific-technical creativity of youth is a way to a society based on knowledge.” This conference will be held on June 29 - July 2, 2010 in Moscow at the All-Russian Exhibition Center. The purpose of this conference is to attract students, graduate students, and young scientists of Russia and foreign countries to solve urgent problems of modern science, to promote integration of science, industry and education, to stimulate obtaining new knowledge, forming the basis of technological innovation, and to promote the development of innovative activities and professional growth of young specialists.

So, this publication contains the theses of reports of students and young specialists presented in the section "Applied Mathematics and Physics" at the conference “Innovations in aviation and astronautics 2010,” held on April 26 – 30, 2010 in MAI.

 

Member of the Conference Organizing Committee,

the Chair of the Section “Applied mathematics and physics”

Professor A.A.Puntus

A.I.Emelkin, V.I.Polezhaev, A.A.Puntus.

(Moscow Aviation Institute, State Technical University, Moscow, Russia).

Numerical modeling and research of structure of the current, local and average heat exchange at free convection in inclined layers.

Rapid development of numerical methods and constant increase in capacity of computers give more and more ample opportunities for numerical research of problems of mechanics of a liquid and gas. The great value has computing experiment. Computing experiment is intermediate between natural experiment and analytical research.

Researches of last years are concentrated to the decision of convection problems at various orientation of a layer with various complicating factors, for example, internal sources of heat, and also radiation, vibrations, a magnetic field, porous and abnormal environments, change of an angle of slope in time etc. But, in spite of the fact that last decade on a first line works on experimental and theoretical research of spatial convection effects in inclined layers are put forward, complexity of structure of a current, possibility of management and requirement of new technologies force to come back to insufficiently studied question on two-dimensional effects of convection interactions in inclined layers.

The most general mathematical model for the description of currents of a continuous viscous liquid are the equations of Navier-Stokes, which in itself are difficult enough and differ nonlinearity, non-stationary, multiscale and presence of a great number of defining parameters, strongly depend on initial and boundary conditions.

In this work modes of interaction of convection mechanisms, caused by loss of stability of mechanical balance (having warmed up from below, lift-lowering current) and absence of balance (having warmed up sideways, cell structure) are studied.

In intermediate cases which are observed at change of an angle of slope, occurs convection interaction between these two mechanisms, leading to essential reorganization of structure of a current,  local and average characteristics of heat exchange. Development of convection research includes studying of following questions: definition of critical conditions of loss of stability, the analysis of structure stationary convection currents and its influence on characteristics of heat-weight transfer. To studying of a critical angle of such reorganization depending on parameters of lengthening, also this work (including trivial, but from it not less interesting case of square area) is devoted.

All data is received during computing experiment, i.e. a method of studying of devices or physical processes by means of mathematical modeling. He assumes, what after construction of mathematical model its numerical research allowing is carried out to simulate behavior of investigated object in various conditions or in different variants. Calculations were spent with the help of researching program COMGA (Cоnvection in Micro Gravity and Applications), developed at Institute of Problems of Mechanics of the Russian Academy of Sciences still in the late eighties, whence follows that the problem of creation of new programs is actual.

A.O.Simonenko.

(Moscow Aviation Institute, State Technical University, Moscow, Russia).

The decision of the nonlinear equations by means of the modified Newton interval method

The interval and interval uncertainty are initial concepts of the interval analysis. Interval uncertainty is a condition of incomplete knowledge of size for which the accessory to some interval, that is the closed numerical interval is known only. The mathematical discipline studying problems with interval неопределённостями and methods of their decision, is called as the interval analysis.

In the present work the problem of search of roots of the nonlinear equation on the set initial interval is considered.

One of methods of the decision of the given problem is Newton interval method. It is based that by means of operator Newton the width of an initial interval with each step decreases, and the final interval with the set accuracy turns out. Newton interval method allows to find only one (multiple) root of the equation and only on an interval which is not containing zero.

The work purpose is creation of the algorithm, allowing to find all roots of the equation on any set interval (including on an interval containing zero or infinity).

As a basis the algorithm of an interval method of Newton in which operation of division into an interval containing zero has been introduced is taken. Further the method has been changed and adapted for simultaneous search of all roots of the equation. The essence of the modified method consists in step-by-step reduction, splitting of an initial interval and localization of roots, namely, receptions of a final interval (or several) with demanded accuracy.

Further it has been generated two algorithms of the modified interval method of Newton: the first – with an exit on the set width of an interval, and the second – with search of the dot decision with the specified accuracy.

The software realizing the above described algorithms in which on base examples it is possible to check up method work is created. As a base example the nonlinear equation in which left part there is a polynomial which degree is set by the user is taken. Also entrance parameters are polynomial factors, an initial interval and accuracy of the decision.

In the program the tree of step-by-step results, the schedule of function is under construction and the decision is deduced by the specified method that allows to consider a decision course visually and most full. Efficiency of the developed software has been checked up on set of examples.

A.S.Kozhevnikov.

(Moscow Aviation Institute, State Technical University, Moscow, Russia).

The application of spectral method analysis for systems with a random period of quantinization to models of asset price dynamics.

At present the stochastic processes with jumps have become more popular than diffusion processes for modeling fluctuations of the market to manage risks and to rate the fair value of an option. In particular, when the process of price dynamics includes Brownian motion the asset price can be changed by a small value in a short period of time, while real prices are moving jumps (have big changes in small time intervals).

Financial models with jumps divided into two categories. The first category includes jump-diffusion models. The price dynamics is defined by the diffusion process, which is experiencing jumps at random times. The second category includes models with an infinite number of jumps on a small time interval. There is no need to introduce Brownian motion in models of the second category, because the dynamics of jumps already rich enough to simulate a nontrivial behavior at small time interval.

We consider Merton model and Bates model in which the behavior of asset price is described a process generated by a mixture of diffusion and discontinuous processes. It is assumed that asset price shocks are independent and identically distributed and form a Poisson flow of events with constant intensity, which equal the average number of jumps per unit time. Such models can be considered as a special case of systems with a random period of quantization.

The problem of finding asset price probability density function, expected asset price, and its dispersion as well as the option price in models of Merton and Bates is studied in this work.

To solve this problem we have developed the algorithm based on the spectral form of mathematical description (spectral method).The main advantage of this approach is the universality of application and ease of implementation. The spectral method allows us to reduce the Kolmogorov-Feller equation for probability density of asset price in Merton and Bates models to linear algebraic equations.

We offer to use the logarithm of asset prices as the state of a system and to convert equations of the Bates model so that the asset price (or its logarithm) and the variation will be measured in the same scale. These offers allow to simplify the solution of the problem by the spectral method and to get more exact results at small truncation of spectral characteristics.

Simulated results for different variants of the behavior of asset prices are analyzed, and made calculations to estimate the influence of the intensity of jumps and their distribution. The calculation results for models of Merton and Bates are compared.

V.N.Panovskiy.

(Moscow Aviation Institute, State Technical University, Moscow, Russia).

Formation of modified interval arithmetic and its implementation as a complex of programs of interval analysis.

Modern computational mathematics considers many practical problems, among which there are problems of solving systems of equations, finding the minima and maxima of functions of several variables and others. If for the decision of these questions we apply the interval analysis, creation of so-called interval arithmetic will be necessary. Interval arithmetic is an algebraic system, formalizing arithmetic operations on intervals as single objects. After describing such a system, we gain a possibility to use all main theorems, corollaries and algorithms of interval analysis.

We consider the following problem: to create and describe interval arithmetic, which can be used to solve various mathematical problems, and implement it as a complex of programs of interval analysis.

It is obvious, that except for the definition of simple binary operations (addition, subtraction, multiplication and division) it is necessary to identify all unary operations (taking the logarithm, module, power, etc.). It is also important to take into account two facts. First, while describing the new algebraic system we should use the extended real arithmetic to avoid questions of uncertainty of an operation. In [1] it is proposed to supply the set of real numbers with two ambiguities (plus and minus infinity) and realize all the known operations above this new set. Secondly, instead of the set of intervals we should refer to their associations, i.e. multiinterval, as many of the functions of the asymptotic behavior or even having a gap have a special domain, to describe which multiintervals fit better. The existence of interval arithmetic of this objects is described in [2].

In order to optimize the execution of an operation over multiintervals we should implement the operation of merging intervals, reducing the already existing intervals and combining overlapping.

The main result is a new algebraic system which is built on the basis of already existing interval arithmetic: multiinterval and arithmetic of William Kahan. There is a program written on Visual C#. NET 3.5, which visually shows the operations realized over multiintervals and solves the equations with several unknown persons.

1. E.Hansen, G.William Walster. Global optimization using interval analysis. Marcel Dekker, Inc, 2004.

2. S.P.Shariy. Finite interval analysis. Publishing «XYZ», 2009.

V.O.Kalas, P.S.Krasilnikov.

(Moscow Aviation Institute, State Technical University, Moscow, Russia).

Equilibrium position stability of the Sitnikov problem.

Sitnikov problem is one of the non-integrable problems in celestial mechanics that represents a rectilinear motion in the restricted problem of three bodies. Two equal primary masses move on two coplanar elliptic orbits, around their barycentre while a third infinitesimal moves on a line perpendicular to the motion plane of the first two masses and going throw the center of mass. Position stability in the first approximation has been studied and it was shown that equation in the first approximation is the linear differential equation of second order with periodic coefficient (Hill equation). Also it was shown that numerical investigation of stability using Lagrange series, which introduce the eccentric anomaly as the function of the time, is invalid when the eccentricity is “large” because any discarded term, which has more than first power of the eccentricity, affects on the edge of the stability very much and results are erroneous for all degrees of approximation. So that investigation of stability using series expansion parameter is reliable when eccentricity is sufficiently small.

Stability for eccentricity over the range zero to one was investigated by regularization original equation set. Regularization was made because equal unity value of eccentricity is a critical point of equation. The eccentricity-dependent magnitude of the trace of monodromy matrix was plotted. By using certain stability criterion (stability is the case if the magnitude of the trace of monodromy matrix is strictly less two) it has drown a conclusion about equilibrium position stability in the first approximation for most eccentricity values. Instability is the case on discrete set of eccentricity values, which satisfy the equality two for magnitude of monodromy matrix. Equal unity value of the eccentricity is an accumulation point for this set. The first ten eccentricity values of instability were calculated.

Research of stability for nonlinear approximation is needed serious consideration.

G.S.Filippov, V.S.Nikolaenko.

(Moscow Aviation Institute, State Technical University, Moscow, Russia).

The mathematical modeling radiation process of the complicated radiator radiation process by Monte–Carlo method.

In addition to the heat radiation of exhaust gas stream influence, the indicatrix radiation which goes from airship  to the conventional back half-sphere, is also influenced by the radiation from the inner surface of the motor aggregate.

We tale up the compound radiator that is a system of geometric adjustable surfaces kind of cone, cylinder, hyperbola. Inside arranged the radiant surfaces in the form of adjustable ellipsoid, cone, ring. The inner parts are the source of radiation. Accordingly the heat radiation of the heated components has the complicated form, taking into account the difference of temperatures, reflection coefficients, complicated geometry. Moreover, the radiation inside the surface can be repeatedly reflected.

To make the calculation of such a complicated radiator, there are several methods, for example the method of directing cosine. But the calculation algorithms in this case, can be rather comprehensive, recursive. Realization of this algorithm on computer, even on a very powerful one, takes too much time (hundreds hours). The “Monte Carlo” method usage is more reasonable, than other methods existent.

To make the calculation a complicated mathematical model of the radiation surface was built. The method main point is that for the microelement of the radiation surface, direction of the ray, going from the centre of that microelement, is given accidentally. It is considered, that all the radiation energy of the microelement or the cell is spread in the mentioned accidental direction. After that we examine the ray expansion, before it goes out of the nozzle or crosses other surfaces.

When the ray crosses other surface, the coordinate of the cross point is calculated and the new accidental direction of the reflected ray is defined (as the mirror reflection is missing). The energy of the reflected ray falls depending on the surface reflection coefficient. When the ray goes out of the nozzle we define the coordinate of it’s cross point with the conventional half-sphere, parted in areas. The radiant energy stream, falling on the mentioned half-sphere parts, is kept and sum up, in case the rays fall in one and the same part. In such a way the spatial distribution of the radiation or radiation indicatrix is defined.

The algorithm uses the probabilistic approach, that approximates the simulated process of emitting and reflecting the ray to the real physical process. This method is precise enough and saves the machine time. The developed algorithm goes with any other complicated radiator, and can be also improved for adding other factors’ investment.

For the derived result analysis the comparison with other methods’ calculation (directing cosine) was made. There was also made a comparison with the results, got by the ANSYS program. The analysis confirmed that the correct method in whole, and algorithm in particular were chosen.

Besides, the calculation of the radiation of gas flow by going out of the complicated radiator surface was made. At that we made the calculation of the radiation on the collecting sphere in different radiation conditions.

We made the calculation of the radiation of gas flow by going out of the complicated radiator surface, on conditions that a protective screen made of special sprayed material is put around it. We also made the calculation of this method effectiveness and practicability.

D.V.Metlitskaya, A.V.Panteleev.

(Moscow Aviation Institute, State Technical University, Moscow, Russia).

Creation of a complex of software «Genetic algorithms of conditional optimization with binary and real coding».

This work presents the genetic algorithms (GA) which are representatives of evolutionary methods of search. Genetic algorithms are based on modeling of processes of natural evolution. One of the most popular areas of the application of genetic algorithms is optimization of multiple parameter functions. As well as other methods of evolutionary calculations, genetic algorithms do not guarantee detection of the global optimum, but successfully work, when it is required to find "good enough" result for comprehensible time. Genetic algorithms can be applied, when the information on character and properties of investigated function almost completely is absent.

There are two groups of genetic algorithms: genetic algorithms with binary coding [1-3] and genetic algorithms with real coding [3-5]. The first group uses binary alphabet for coding points on the set of possible solutions. The second group has resulted from refusal of idea of coding; the solution in a chromosome is represented in the form of a set of real numbers.

The objective function defined on set of possible solutions is considered in this work. It is required to find a global conditional maximum of function on this set. On the basis of the genetic algorithms was generated the program of search of a global conditional maximum. A working environment is Microsoft Visual Studio 2005, programming language is C #. The program works in a mode of dialogue with the user. The user interface includes the basic window with forms for data input and display of received results, the push-button panel for management of a course of computing process (it is possible to receive the solution at once or on steps). Program work begins with input of the initial data: type of optimized function, set of value of variables, and also from input of parameters of algorithm: population characteristics (the size, a maximum quantity of chromosomes (individuals), length of bit lines of genes), type of operators of selection, crossing, a mutation, and quantity of elite individuals (if elite strategy is applied). During work with the program it is possible to receive a population graphic representation, and also the schedule of change of the greatest value of objective function at transition from one population to another. Also it is possible to keep a report of the program and to keep the received results in memory of the computer.

On test examples (Rosenbrock saddle and sphere function) have been considered efficiency of algorithm. The analysis of the received results shows that the generated algorithms allow finding the comprehensible solution when function has difficult structure, and the decision of a problem of search of an optimum of sphere function does not cause difficulties.

1. J.N.Holland. Adaptation in Natural and Artificial Systems. Ann Arbor, Michigan: Univ. of Michigan Press, 1975.

2. D.Goldberg. Genetic Algorithms in Search, Optimization and Machine learning. Addison-Wesley, 1989.

3. A.V.Panteleev. Metaheuristic algorithms of search of a global extremum. – М: MAI-PRINT, 2009.

4. F.Herrera, M.Lozano, J.L.Verdegay. Tackling real-coded genetic algorithms: operators and tools for the behavior analysis// Artificial Intelligence Review, V. 12, № 4, 265–319, 1998.

5. Z.Michalewicz. Genetic algorithms, Numerical optimization and constraints// Proceedings of the 6th International conference on genetic algorithms, 151–158, 1995.

 

D.E.Pivovarov, V.I.Polezhaev, A.A.Puntus.

(Moscow Aviation Institute, State Technical University, Moscow, Russia).

Computational solution of third system of Navier–Stokes equations in Boussinesq approximation for convectional heat transfer problem.

Last time straight computational solution of Navier–Stokes equations is considered powerful and reliable tool for investigation of turbulence flows. The computational results of numerous examples correspond with experimental results. It impel us to apply this method for simulation convection processes of heat exchange.

There are a lot of algorithms of computational solutions today. All these ones differ in discretization scheme, accuracy and stability. Finite-differences scheme of N.Nikitin is interested in use of curvilinear coordinate system. In addition it uses semi-implicit third order accurate Runge-Kutta method, local error estimation and time-step control.

This paper describes above mentioned algorithm in application to convection heat exchange problem, realization it on the PC and check its applicability on the model problems such as Davis test, lifting-movable flow, Rayleigh-Benar problem, convection in inclined layers. Tecplot presents graphical realization of computational data. It is represented parametric investigations, maximal characteristics of heat exchange, comparison derived data with data of other authors and exposed third features.

I.F.Dmitrakov.

(Moscow Aviation Institute, State Technical University, Moscow, Russia).

The application of metaheuristic methods for the global extremum search of functions.

The problems of search of unconditional and conditional global extremum of multivariable functions by means of metaheuristic optimization methods are considered.

Detailed algorithms of application of the Simulated Annealing method, the adaptive Simulated Annealing method and the Differential Evolution method are generated.

The Differential Evolution method is based on the analysis of evolutionary processes. The Differential Evolution algorithms feature is use of differences between individuals (argument of objective function). It is realized by the linear operator, named "differentiation".

The Simulated Annealing method is based on the analysis of process of freezing of liquids or recrystallization of metals in annealing process. Algorithms of Simulated Annealing with some probability suppose transition in a condition with higher value of objective function in the course of solution search, for that the point could leave a local minimum neighborhood.

The software environment of visualization of optimizing algorithms work process, where  above-listed methods are added, is developed.

This software environment allows to:

-         easily operate with parameters of a tested method;

-        analyze efficiency of its operation on test functions;

-         analyze method convergence;

-        "see" process of its operation;

-         effectively form a policy of method parameters choice  that has crucial decisive importance in  metaheuristic optimization methods;

-         see computation intermediate outcome;

-         efficiently make a comparison of the chosen method with other global optimization methods for the following result analysis.

The efficiency of the developed software is checked up on various base examples.

M.A.Zhmakin, A.O.Burmistrov.

(Moscow Aviation Institute, State Technical University, Moscow, Russia).

Minimax Identification of the kinematic model of the movement of aircraft in terms of a priori uncertainty.

In this paper we study the problem of minimax estimation of motion parameters of an aircraft (LA) with inaccurate given probability characteristics of observation errors. Movement measurements of an aircraft are made by measuring system, consisting of diverse radio instrumentation [1].

A major disadvantage of the optimal methods of estimation is the assumption that the probability characteristics of the model parameters of aircraft motion and observation errors are given a priori [2]. In addition, it is usually assumed that the model parameters relate to any particular type, which restrict the sound application of methods of statistical estimation. This paper describes a model of movement measuring aircraft, in which noises are partially unknown probability characteristics. The type of the distribution vector of random observation errors and element wise constraints imposed on the covariance matrix of errors are considered to be given. The very covariance of measurement errors is unknown.

The paper presents formulation and solution of the problem of minimax estimation with the use of some sufficient conditions that allow the operator to reduce the problem of finding minimax estimation to solve the dual problem, which is a standard problem of convex programming. The meaning of the dual problem is to define a set of probability characteristics of the parameters of the observation model, in which the optimum for the selected criteria of quality assessment is the least accurate (i.e., the distribution parameters and the noise model is the "worst" for a given class of admissible distributions). It is shown, that the minimax estimate the model parameters of motion coincides with the optimal in the mean integrated squared criterion of assessment, calculated under the assumption that the true covariance of errors is the same as the "worst".

The algorithm of searching "the worst" of the covariance matrix with the use of the software package SeDuMi system MatLab, designed to solve special problems of convex programming (e.g., SDP and SOCP), is shown in the report. Features of this package helped to find the "worst" covariance matrix taking into account the conditions of its symmetry and nonnegative definiteness. The report provides a method of constructing the set of uncertainty for the covariance matrix of observation errors in the form of multidimensional confidence region with the given level of reliability.

Obtained results allow to reduce significantly the demands on the level of a priori information about the characteristics of the observation model and to improve the reliability of the results of estimation in terms of a priori statistical uncertainty.

This work was supported by RFBR (grant № 09-08-00369), as well as in activities 1.1 FTP (state contract 30.09.2009 № 02.740.11.0471).

1. Zhdanyuk B.F. Fundamentals of statistical processing of trajectory measurements. M.: Soviet radio, 1978.

2. Malyshev V.V., Krasilshchikov M.N., Karlov V.I. Optimization of observation and control aircraft. Moscow: Mashinostroenie, 1989.

 

M.A.Lebedev, V.I Polezhaev, A.A.Puntus.

(Moscow Aviation Institute, State Technical University, Moscow, Russia).

The hysteresis of characteristics current and heat an exchange in angle of inclination.

In a problem about convection in inclined layers, depending on an angle of inclined, operates two mechanisms of thermal gravitational convection: a balance is possible, depending on a difference of temperatures between the top and bottom borders it can be steady or not steady, and the balance is impossible. The Critical angle where many-cells structure of current, which is characteristic for Benarovs cells mechanism, changes with one cell, which is characteristic for vertical slice with lateral heating, so we can’t define it unequivocally, because of appearing of gusterisis.

In this work, which continues researches was considered a flat convection substance movement in a horizontal layer, which was warmed up from below. At first, the angle of inclination changes from 0 degrees to 180, and then upside-down from 180 to 0 degrees with a step of 15. In the result, the middle numbers of Nusselt for different numbers of Prandtl, which are corresponded to such substances as air at a room temperature, water with a temperature 100°С or to transformer oil with a temperature 100°С were got and brought to the tables. A dependence of Nusselt number from angle of inclination was ambiguous. In the work were two modes: cellular and vortical, and unstable condition. The transition between two exists modes in changing process of angle of inclination has a gusterisis character.

The hysteresis is a characteristic effect convection, when a current and heat exchange in one of directions does not coincide with heat exchange in the opposite direction. This effect is known and in other sciences and has practical application in the physicist, chemistry, biology, but in convection heat exchange, despites on significant effects, it wasn’t investigated regularly. The detailed research on the basis of model with periodic boundary conditions is given in [4].

The work shows that both modes lead to increasing of heating a stream in comparison with molecular, corresponding to a plane-parallel current. The cellular mode provides much more intensive heat over carrying, than vortical. It with higher speed of a cross-section current in the cellular mode. As a result transitive character of this phenomenon gusterisis type which leads close in relation to initial, structure of convection and to the Nusselt number, which is different from opposite rotation from the initial is accurately observed. So for air at a room temperature this difference has made approximately 15 %, and for transformer oil at temperature 100°С more than 20 %. Researches are executed by means of the computer laboratory which description is given in [3].

1. D.Е.Pivovarov, V.I.Polezhaev. The structures of the flow and feature of heat exchange at free convection in inclined layers (The 17 School-seminar of young scientists and specialists under the direction of the academician А. I. Leontyev “The problem of gasdynamics and thermalmass exchange in aerospace technologies”, Zhukovskij, May 25-29, 2009).

2. V.I.Polezhaev, S.А.Nikitin. Feature of heat exchange at freely-convective interactions in closed volumes. Technical and technological annexes (The 17 School-seminar of young scientists and specialists under the direction of the academician А. I. Leontyev “The problem of gasdynamics and thermalmass exchange in aerospace technologies”, Zhukovskij, May 25-29, 2009).

3. M.K.Ermakov, S.A.Nikitin, V.I. Polezhaev. System and computer laboratory for simulation of processes of convective warm and mass exchange. Information RANS, МZhG  № 3, 1997, p. 22-38.

4. G.Z.Gershuni, E.M.Zhukhovitskij, A.A.Nepomnyaschiy. Stability of convective flows. М.: Science. The main editing of physical and mathematical literature, 1989, p. 53-55.

O.V.Ryazantseva.

(Moscow Aviation Institute, State Technical University, Moscow, Russia).

Global optimization by continuous tight-fisted (greedy) randomized adaptive search procedure.

Global optimization problems abound in many fields, including materials science, biology, chemistry, and genetics, military science, electrical engineering, robotics, and transportation science.

Now at the decision of problems of designing of space systems and space-rocket complexes metaheuristic methods of global optimization are widely enough applied. They allow to receive close enough to true result for comprehensible time and without considerable simplifications of mathematical statements of problems. An example of such method is the greedy randomized adaptive search procedure (GRASP) for global optimization.

The problem is: search of a global extremum of functions of many variables on set of admissible decisions where admissible values of corresponding variables are segments.

The method is uses a multi-start local search procedure, where each GRASP iteration consists of two phases, a construction phase and a local search phase.

The first phase (construction phase) combines greediness and randomization to produce a diverse set of good-quality starting solutions from which to start local search. Then the received on the second phase points take as initial for the first phase and procedure repeats. The best solution over all iterations is kept as the final solution.

On a construction phase a line search in the directions parallel to co-ordinate axes is made. After the line search a restricted candidate list (RCL) is formed. Then one direction in a random way gets out of the RCL. Found in result of search in this direction the point undertakes as initial for the decision continuation, the used direction is excluded from the list, corresponding value of solution coordinate is fixed. The described sequence of actions proceeds until all coordinates of a solution vector will be fixed.

On a local search phase the received point improves by search in the directions which quantity is defined by dimension of a problem. Search in directions depends on a step which can decrease in the course of search. The initial step gets out depending on the set of search.

This stochastic search method is simple to implement, is widely applicable, and does not make use of derivative information, thus making it a well-suited approach for solving global optimization problems.

Multi-start procedure, that is repeated the algorithm of search for different initial random points, provides additional accuracy because total result is the best of received on all starts.

Program realization of the presented method is executed by means of Builder C ++. Its functions:

-         input parameters of a method;

-         choice of criterion function from the list;

-         decision the problem and conclusion of the detailed information, including step-by-step;

-         construction graphs: level lines with the found point of an extremum, values of criterion function on starts; evolution of the best value of criterion function following the results of the realized starts;

-         viewing of a detailed protocol of the decision;

-         granting of the help information.

The program possesses the intuitively-clear interface and allows to show efficiency of application of considered algorithm for the chosen functions by change of parameters of a method.

The program realization of method can be useful in the course of training as demonstration of work of global extremum search algorithm. It is also allows to receive set of the data necessary for the analysis of its efficiency.

T.E.Churkina.

(Moscow Aviation Institute, State Technical University, Moscow, Russia).

On satellite attitude motion stability at moon type resonance.

The attitude planar motion of a satellite (solid body) in the central Newtonian solar gravitational field in an elliptical orbit is considered. Moon type motion is under consideration (i. e. the satellite in time of one orbital revolution makes full turn in absolute space around the axis passing its center of mass).

Investigated is the nonlinear problem on the motion stability at both plain and spatial perturbations.

Curves of third and fourth order resonances are built in the space of problem parameters in stability regions of first approximation. For the parameters values corresponding resonance curves Hamilton function has been led to normal form and the conclusions about motion stability have been drawn with the help of well known criteria on the coefficients of this normal form. All the calculations have been made with the help of computer.

Yu.O.Khan, A.I.Fedyushkin, A.A.Puntus.

(Moscow Aviation Institute, State Technical University, Moscow, Russia).

Mathematical model of calculation of thermal convection at normal and lowered gravitation.

This work is devoted to mastering the hydrodynamic set of programs Fluent and geometry and mesh generation software Gambit through an example of solving a test problem of convection in a square cavity heated from one side. The purpose of the work is to construct a mathematical model for calculating thermal convection in dimensionless variables for different Rayleigh numbers by using the complex of programs Fluent, designed to solve problems of hydrodynamics in dimensional variables. This raises the need for testing the results of numerical calculation.

The problem of thermal convection of a viscous incompressible fluid in a square closed area with thermally insulated horizontal walls and specified boundary temperatures on vertical walls is examined. A mathematical model of this problem is two-dimensional Navier-Stokes equations in dimensionless variables in the Boussinesq approximation. For speed, no-slip condition is put at all boundaries, and for temperature, the following assumptions are made: the lower and upper boundaries are heatproof (adiabatic), the right and left boundaries are isothermal. As initial conditions, a stationary fluid with a linear temperature distribution is given.

This problem of natural convection in a square area was solved in three natural variables: speed, pressure, and temperature. To simulate this convection problem, the dimensionless equations were used, including dimensionless parameters: Prandtl number, which characterizes the measure of similarity of the temperature and velocity fields in the flow, as well as the Grashof number. In addition, the system of Navier-Stokes-Boussinesq equations was used, written in the following variables: current-temperature and vortex-function (which does not include pressure). Calculations were made using modern hydrodynamic program Fluent.

Results of calculation showed that the most accurate of all the solutions of this problem is the reference solution obtained by extrapolation on the grid with zero step, by using the finite-difference method on different grids. The calculation results also showed that for given boundary and initial conditions with increasing Rayleigh number, the maximal value of the absolute value of velocity is shifted more to the boundaries of the area, and the isotherms tend to adopt a horizontal position. As a result of the research using the program Gambit, a geometric model and a non-uniform computational grid are reproduced, and further by using the program Fluent, parametric calculations are conducted and comparison of the obtained results with the “standard” solution is given for different Rayleigh numbers, which showed good accuracy of calculations.



[Contents]

homeKazanUniversitywhat's newsearchlevel upfeedback

© 1995-2008 Kazan State University