Вы находитесь на странице: 1из 6

L3 Programming the Finite Element Method

Ben Richardson

L3 Programming the Finite Element Method matlab Coursework. This coursework looks at the use of the Finite Element Method (FEM) and compares the results obtained from FEM with that of theory. 1. Theoretical Calculations using Euler-Bernoulli Theory In order to determine the location of the neutral axis ( ) of the T-section use: [1]

In order to determine the second moment of area of the T-section: [1] Where and is the difference in the height between the centroid of the T-section and the section currently in consideration.

Using Euler-Bernoulli theory its possible to see that: [2] For a cantilever:

As:

Using a similar method its possible to see that for a simply supported beam:

As:

We now have the maximum deflection of the cantilever( ) and the simply supported beam ( ) alongside the location of the neutral axis on the T-Section( ).

1 19/01/2011

L3 Programming the Finite Element Method

Ben Richardson

2. Modified Version of inputdata.m function [E,v,coord,etopol,bc,forcedata]=inputdata E=70e9; v=0.2; node=0; nelx=2; nely=6; nelz=5; c=1; length=1.6; depth=0.06; height=0.1; coord=[]; bc=[]; if c==2 length=0.8; end for z=0:height/nelz:height for y=0:depth/nely:depth if(y>0.01)&&(z<0.08) break; end for x=0:length/nelx:length node=node+1; if(y==0)&&(x~=0) bc=[bc; 3*node-1 zeros]; end if(x==length)&&(z==height)&&(y==0) force=3*node; end if c==1 if x==0 bc=[bc; ((3*node-2):(3*node))' zeros(3,1)]; end end if c==2 if(x==0)&&(z==0.07) bc=[bc; ((3*node-2):(3*node))' zeros(3,1)]; end if(x==length) bc=[bc; 3*node-2 zeros]; end end coord=[coord; x y z]; end end end change=1; etopol=[]; h_change=nely/6; for z=1:nelz if((z/nelz)<=(4/5)) number=nely/6; else number=nely; end if((z-1)/nelz)>(4/5) h_change=(nely/6)+(((change*nely)-(change*(nely/6)))/(z-1)); change=change+1; end for y=1:number for x=1:nelx if((y/nely)>(1/6))&&((z/nelz)<(4/5)) break; end node=x+((y-1)*(nelx+1))+((z-1)*(nelx+1)*(h_change+1)); n1=node; n2=node+((nelx+1)*(number+1)); n3=n2+1; n5=n1+nelx+1; n6=n2+nelx+1; n7=n6+1; n8=n5+1; etopol=[etopol; n1 n2 n3 n4 n5 n6 n7 n8 ]; end end end forcedata=[force 5e+2];

n4=n1+1;

2 19/01/2011

L3 Programming the Finite Element Method

Ben Richardson

3. Convergence Graphs for Case I and Case II

Figure 1 Case I Deflection of the Cantilever

Figure 2 Case II Deflection of the Simply Supported Beam

4. Isometric Plotmesh Figures for Different Discretisations of Case I

Figure 3 Element Topology of 6-6-5 (x-y-z) 60 elements

3 19/01/2011

L3 Programming the Finite Element Method

Ben Richardson

Figure 4 Element Topology of 32-6-5 (x-y-z) 320 elements

5. Tabulated Vertical Displacement Results for all Discretisations Figure 5 Case I Displacements Number of elements x-y-z 2-6-5 6-6-5 12-6-5 24-6-5 32-6-5 50-6-5 100-6-5 2-6-10 6-6-10 12-6-10 24-6-10 32-6-10 50-6-10 2-12-10 6-12-10 12-12-10 24-12-10 12-6-5 12-12-5 12-18-5 12-24-5 12-6-5 12-6-10 12-6-15 Total Number of Elements 10 60 120 240 320 500 1000 20 120 240 480 640 1000 40 240 480 960 120 240 360 480 120 240 360 Displacement (mm) 0.21190 1.5000 3.4000 4.9000 5.3000 5.6000 5.8000 0.21190 1.5000 3.4000 4.9000 5.3000 5.6000 0.21190 1.5000 3.4000 4.9000 3.4000 3.4000 3.4000 3.4000 3.4000 3.4000 3.4000

4 19/01/2011

L3 Programming the Finite Element Method

Ben Richardson

Figure 6 Case II Number of elements x-y-z 2-6-10 6-6-10 12-6-10 24-6-10 28-6-10 50-6-10 60-6-10 2-12-10 6-12-10 12-12-10 24-12-10 6-6-10 6-12-10 6-18-10 6-24-10 6-6-10 6-6-20 6-6-30 6-6-40 Total Number of Elements 40 120 240 480 560 1000 1200 80 240 480 960 120 240 360 480 120 240 360 480 Displacement (mm) 0.096590 0.42690 0.63000 0.71550 0.72420 0.74150 0.74150 0.096590 0.42700 0.63000 0.71550 0.42690 0.42690 0.42690 0.42690 0.42690 0.42690 0.42690 0.42690

6. Stress Distributions predicted by FEA

Figure 7 Element Topology of 6-6-10 (x-y-z) 320 elements Case II

Unsurprisingly, figure 7 proves that as you move along the x direction the stress becomes larger and larger. However, the stress fields cross over when the height of the section is approximately equal to 0.07. This is an unsurprising result as the neutral axis was calculated earlier in section 1 as 0.07m, meaning at that point there should be no stress as its the crossover point between compression and tension of the beam.

5 19/01/2011

L3 Programming the Finite Element Method

Ben Richardson

7. Comment on Greater Accuracy is Obtained in a Deformation Analysis When Using a Large Number of Finite Elements. As the results obtained show, its possible to see that this comment is true to an extent. However, what is more important when considering a deformation analysis is the element topology. As the tabulated results show; there are times where two analyses with the same number of elements have vastly different results in displacement. This is due to, in the case of the Tsection considered, increasing the number of elements in the x direction is not only the best method; but as some of the results show, possibly the only method to obtain greater accuracy. The isometric plots in figures 3 and 4 highlight the vast difference when increasing the number of elements in the x direction. The results tables in figures 5 and 6, alongside the convergence graphs in figures 1 and 2 show particularly what little difference is made when increasing the number of elements in the y or z direction. In fact, one can see that when the x direction isnt changed at all, greater accuracy isnt obtained no matter the increase made in the y or z direction. The modified code of inputdata.m would only function when the number of elements in the y direction was a multiple of 6 and when the number of elements in the z direction was a multiple of 5. Due to this it meant results could not be obtained where nelz was below 5 or nely was below 6, so its not possible to see what difference would be made in decreasing the number of elements in the y or z directions. Overall, the graphs in particular show increasing the number of elements in the x direction creates an exponential curve that flattens out at approximately the answer previously calculated in line with euler-bernoulli theory. For both cases its possible to see that once the number of elements have gone above 500, although there is slight improvement, many would see the answers obtained as acceptable considering the length of time taken to calculate 1000 or more elements. Optimal Mesh Arrangement For Case II With 100 Elements

Here the displacement boundary conditions will be to: Hold all nodes in the y direction when y=0m Hold all nodes in x, y & z directions when z=0.07m and x=0m Hold all nodes in the x direction when x=0.8m

8. References [1] - Wu, D. Level 1 Mechanics Lecture Notes, Durham University, School of Engineering, 2008. [2] - Augarde, C. Level 1 Structures Lecture Notes, Durham, School of Engineering, 2009.

6 19/01/2011

Вам также может понравиться