Abstract : Finite Element Method (FEM) is based on the concept of building complex objects using simple elements or division in small parts of complex objects easily manipulable. This paper presents the modeling equations of fluid and thermal analysis, namely equations type: hyperbolic, elliptical parabolic. It will be noted that it is important to know every step of the method as the main stage is the development model for calculating the structure. Other important steps are the domain meshing and interpolation process of meshing with the effect sharing model structure in a certain number of pieces called finite elements. Here, we will present three types of equations modeling using Matlab. Finally, we notice that there are several programs that are based on MEF instead with a user-friendly interface. Even here it is necessary to understand the method steps in order to properly model the model chosen.

Key words : finite element method, parabolic, elliptical, hyperbolic, wave.

1. INTRODUCTION

In this paper we discuss solving problems using the finite element method further noted MEF. This method is a general method for solving partial differential equations approximate that describe physical phenomena.

MEF consists of portions of the study areas and recomposing field of study respecting certain mathematical requirements.

MEF is based on the concept of building complex objects using simple elements or division in small parts of complex objects easily manipulable. The applications of this simple concept can be found easily in real life, especially in engineering, it can be extended in any field, such as:

– Structural analysis (determining the state of tension or deformation of a structure requested);

– Thermal analysis (determination of temperature field and the flow of heat from a thermal requested structure);

– Fluid analysis (determination of current or potential function of speed);

– Analysis of electric / magnetic (electric or magnetic flux determination).

Basis of this method was first formulated in 1943 by German mathematician Richard Courant of (1888-1972), which, combining with numerical analysis method Ritz issues variational calculation and minimization achieved satisfactory solution for vibration analysis systems.

Since the 70s, the finite element method was used to solve the most complex problems in the field of elastic structures continue from civil, industrial or construction of dams to ships, these cosmic.

In this paper we chose to model the equations of fluid and thermal analysis, with three types of equations: hyperbolic, elliptical parabolic.

In the first part we’ll start with some basics about the finite element method analysis. It is important to know every step of the method as the main stage is the development model for calculating the structure. Other important steps are the domain meshing and interpolation process of meshing with the effect sharing model structure in a certain number of pieces called finite elements.

The following three types of equations will be modeling with MATLAB.

Finally, you will notice that there are several programs that are based on MEF instead with a user-friendly interface. Even here it is necessary to understand the method steps in order to properly model the model chosen.

2. FINITE ELEMENT METHOD (FEM) OVERVIEW

In general, physical phenomena are described in terms of differential equations mathematically, by whose integration, boundary conditions data give an exact solution of the problem. This has the disadvantage that it is analytically applicable only to relatively simple problems. The problems arising in practice are often complex in composition to the physical and geometric parts, loading conditions, boundary conditions, etc., so the integration of differential equations is difficult or even impossible.

The finite element method is used as a starting point, a full model of the studied phenomenon. It applies a series of separate small parts of a continuous structure obtained by the mesh, known as the finite element connected to each other at points called knots.

Figure 1 – Types of finite elements

2.1. STEPS TO SOLVING A PROBLEM USING THE FINITE ELEMENT METHOD

STEP 1. Dividing range finite element analysis.

In this step choose the type or types of finite elements suitable for the task, then divide finite element structure. This is called meshing and can be done by computer.

STEP 2. Establishment of finite element equations (basic equations).

Material behavior or the environment in the contents of a finite element is described by equations finite element equations called elementary. These form a system of equations of the item.

STEP 3. Assembling basic equations in structure system equations

The behavior of the entire structure is molded by assembling the system of equations of the finite elements in the system of equations of the structure, which in terms of physical means that the balance of the structure is conditioned by the finite element equilibrium. The assembly is necessary in the common node elements, function or unknown functions have the same value.

STEP 4. Implementation of boundary conditions and solving the system of equations of the structure. The system of equations obtained from implementing appropriate boundary conditions concrete problem is solved by one of the processes obÅinuite, for instance by eliminating or by digesting Gauss Choleski yield function values in knots. These are called and unknown primary or first order.

STEP 5. Perform additional calculations to determine the unknown side.

In some problems, after finding the primary unknowns, the analysis concludes. This is usually the case when heat conduction problems, the primary unknowns are the nodal temperatures. On other matters, however, only the unknown primary knowledge is not sufficient, the analysis must proceed with determination unknowns secondary or second order. These are higher order derivatives of the primary unknowns. Thus, for example, mechanical problems of elasticity, the primary unknowns are the nodal displacements. With their help at this stage, determine secondary unknowns that are specific strains and tensions. And if problems continue with thermal analysis can determine which side unknowns are the intensities of heat flows (thermal gradients).

3. MATHEMATICAL MODELING IN MATLAB BY FINITE ELEMENT ANALYSIS

It made an application that allows the user to choose the modeling results in Matlab using finite element method for each of the three types of equations. Figure 2 – Interface for three types of equations in Matlab

3.1. HYPERBOLIC EQUATION

The first model chosen is the hyperbolic equation modeling especially where waves. The application allows the achievement of a movie showing the movement of the waves.

Figure 3a) – Wave equation

In these images is seen as wave height and spreads begin to decline further.

Figure 3 b) â” Wave equation

Figure 3c) â” Wave equation

3.2. ELLIPTICAL EQUATION

For this equation to make an application to the MEF in 2D. The equation is selected elliptical shape

-âu=32(x-x^2+y-y^2 ): pe Î©,u(â,Î©)=0,

with Dirichlet boundary conditions.

Figure 4 â” Elliptic equation

In the figure 4, as a result of modeling, it is noted that the finite elements were chosen triangle type and the surface is uniform.

3.3. PARABOLIC EQUATION

For parabolic equation we chose a design that enables the modeling of the heat transfer body through the isotropic temperature-dependent heat transfer.

Figure 4 a) â” Heat transfer equation

The finite elements were all type triangle and in the following images is observed as the heat transfer from the body varies, the highest temperature being in the red and the blue color area below (Fig. 4a and Fig. 4b).

Figure 4 b) â” Heat transfer equation

Here, the user is able to see a movie to better understand how temperature varies and beginning to increase (as shown in Fig. 4c) as the body is subjected to certain tests.

Figure 4 c) â” Heat transfer equation

4. CALCULATED BY COMPUTER-AIDED FINITE ELEMENT

Using a computer and appropriate software are indispensable for the application MEF as simple a structure. Even the principle of the method results in a large amount of numerical calculations that can not be achieved only on new computers with specialized software. Accordingly, FEM analysis of Analytical acquires a fence automation, which can be a trap by loss of control over the operations they carry out FEM program. For the analysis of complicated spatial structures can become difficult due to automation preprocessing model, check its accuracy, and correct errors in them and render amendments to the initial model.

At present there are numerous programs in general specialized types of problems. Some of these include: ANSYS, MIKE, FLUENT, COMSOL, NASTRAN, MOSAIC, GffTS, etc. They allow through all the stages referred to in Chapter 1 and others such as geometric model construction, introduction of information related to material, mesh geometric model, the application loads, and limit conditions, solving itself and post-processing calculations. Strong development of graphical user interfaces of computers today allows a particularly effective treatment, suggestive and rapid calculation results in the form of graphical representations of bodies compared deformed body – the body unaltered, travel by model, representing the portion of the pattern, etc.

4.1. PREREQUISITES A USER M.E.F.

A user is forced to solving a particular problem. The program applied calculation method does not solve the problem but a model of it, which generally conceives user. The results can be confirmed or not, depending on how the model was chosen calculation.

Modeling is a task of simplifying the structure by including various portions of the structure in the category: bars, plates, massive bodies, by considering and bearing loads.

Correct modeling is a matter of experience, inspiration and not least the knowledge of the theoretical basis of the method.

Once established computing model needs to be prepared to solve the problem input. Each finite element program has certain peculiarities that must be learned, but there is a basic method that, once mastered, allows any program finite element approach.

THE STAGES OF ANALYSIS BY FEM

Large corporate programs are three important stages of solving a problem using MEF.

Table 1 â” FEM solving phases

Input data

(Preprocessing) Processing Output

(Postprocessing)

Nodal coordinates;

Types of bearing;

Jams (boundary conditions);

Loading (mechanical, thermal, etc.);

Material properties;

The shape, type and size of F.E.

Tensions;

Movements; Temperatures;

Current function;

Electric / magnetic flux

Preprocessing step is preparation of input data needed to resolve a problem and save them in a data file.

Processing is effective in solving the problem numerically model. Data already prepared (in preprocessor) are taken from the data file and run the type of problem.

Postprocessing is the “viewing” stage of the results in tabular or graphical form. This phase allows evaluating and commenting on the results.

Factors that influence mesh

A number of elements which are meshing condition:

– Type Finite Element. They are chosen depending on the type and scope of analysis, required accuracy, of unknown size variation etc.

– Parabolic elements are preferable to linear, because the same number of nodes, element discretization parabolic solution is more accurate than the linear elements.

– If there are several types of finite elements, on the border between them should be mainstreamed.

– The size and number of finite elements influencing convergence solution. Note that a larger number of elements approaching the solution exact result, but an excessive increase may lead to a “collapse” due to its error by car to a large volume of calculations.

– Positioning nodes, which generally is uniform in structure. Discontinuity in geometry or in loading require the choice of intermediate nodes. Moving to an area with fine mesh one with coarse mesh should be done gradually, not suddenly.

– The degree of uniformity of the mesh. Avoid using items with elongated (very sharp triangles, rectangles with aspect ratio higher than 3). Preferably it would be like meshing with triangles contain only equilateral triangles, rectangles contain only mesh with square and type the space with elements still BRICK, especially elements still contain cubic.

5. CONCLUSIONS

This paper deals with the method MEF from mathematical point of view can be treated as a process for obtaining a numerical solutions approximate for solving a system of difference equations defined partly on a finite domain (D) with boundary conditions (boundary) data.

It was noted that the work (D) is decomposed in a finite number of simple subdomains (Finite Element), connected to each other on the borders of separation in a finite number of points called nodes. In general geometry of the domain (D) is approximated by simple subdomains meeting.

Unknown function (temperature, displacement, etc.) is approximated by locally every finite element interpolation functions defined in relation to acesteoa values in hubs located along finite elements. These functions have been termed as the basic functions.

Meeting interpolation functions for the entire domain (D) is a set of function approximation and their nodal values are generalized coordinates. Test functions are introduced into the system of differential equations and nodal values are determined by methods employed in calculating the variational (Ritz method, Galerkin).

All steps were taken to resolve the three types of equations: elliptic, parabolic and hyperbolic where he finally obtained a 2D / 3D modeling of this equation.

It was noted that there are a multitude of programs that work with finite element method, but not requiring full calculation of each user; they have a more friendly graphical interface that allow only the introduction of basic data such as the type of equation, the number of finite elements to be meshed area etc.

6. REFERENCES

[1] Strang, G., Fix, G.J., An Analysis of The Finite Element Method, Prentice Hall, 1973

[2] Stefan I. M., Bistrian D.A., Introducere Ã®n metoda elementelor finite, Cermi IaÈi, 2008

[3] Asadzadeh M., An introduction to the Finite element method (FEM) for differential equations, 2010

[4] Nikishkov, G.P., Finite Element equations for heat transfer, Hardcover, 2010

[5] Garth A. Baker, A Finite Element Method for First Order Hyperbolic Equations, American Mathematical Society, Volum 29 , No. 132 (Oct., 1975)

[6] Madden N., Numerical Solution to Differential equations using Matlab, 2012

[7] Xiang Du, Yuan Dong, John C. Bancroft, 2D wave-equation migration by joint finite element method and finite difference method, Tsinghua University, CREWES Research Report â” Volume 15 (2003)

[8] http://persson.berkeley.edu/distmesh

[9] http://www.particleincell.com/blog/2012/matlab-fem/

[10] http://cfdp.utcb.ro