FEM_Modeling_Project_Evaluation of a Steel Pulley and Shaft Design
2016 Fall ME 7210 Elasticity and Plasticity Final Project
1. 2016 Fall ME 7210 Elasticity and Plasticity Final Project
Strip stress over half space
Northeastern University Department of Mechanical Engineering
Mutian Fan
2. Abstract
This project discusses the theory of analysis of stress in a uniform loaded strip stress over a half
space. Based on textbook Elasticity 3rd
edition by Martin Sadd on page 196-198, the equations to
calculate stress field in x and y direction and xy shear stress are given. A MATLAB routine is
scripted to calculate and plot the theoretical stress field over 3 straight line under the loaded
area on the half space. Then Solidworks FEA analysis is employed to test out theory. Also,
correct boundary condition and mesh size are discussed in this report.
3. Theory
The principal theory behind the strip stress over a half space comes from superposition of single
point load solution. (pg 196-197)
The equations are shown in figure 1 and figure 2 below.
Figure 1
Figure 2
For practical, I only calculated and plotted 3 lines of stress that are directly under the
boundaries and at the middle of loaded area. I implemented a MATLAB routine and plot the
stress vs relative distance. Please refer to appendix A for MATLAB routine.
For all these 3 lines, I calculated xx normal ,yy normal, and xy shear stress. They are shown in
the appendix.
4. Finite element analysis
The part is created as a 100*100 mm square with 1mm thickness plate and a loaded area of 20
mm on the top. It is as shown in the figure 9.
For FEA, the boundary conditions have to be set up properly, I used different boundary
conditions to test and compare the theory plot to determine the correct boundary conditions.
I used only fixtures on bottom and sides without roller support, roller support with only bottom
and sides support with same dimension of the half space and mesh size to ensure the accuracy
of the effect of the boundary condition.
After mesh and simulation, X,Y normal stress and xy shear stress are showed and probe function
are used to plot the stress field along the lines.
Because there are 9 plot for each condition and we have 36 plots in total, I am not able to
compare them all here. In this paragraph, I will only use one set of plot to compare and explain
which is the correct boundary condition.
Figure 3 Theory x1 stress plot
Figure 4 x1 stress with fixtures only
Figure 5 x1 stress with bottom fixture and roller slide
support
5. It is quite obvious that in figure 5 and 6 the general trend of the plot is not right compare to the
theoretical plot. The major difference between plot 4 and 7 is that the plot 4 has a “wiggle” at
the end of the plot. It is been pointed out in the zoomed out version in the following.
Figure 4 x1 with fixtures only zoomed out
Other plots show same pattern, hence it is safe to conclude that the correct boundary
conditions are bottom and sides fixtures and roller slides on both sides just as shown in the
following SolidWork sketch.
Figure 6 x1 stress with side fixture and roller slide
support
Figure 7 x1 stress with full boundary condition include
all fixtures and roller slide support
Line 1
Line 2
6. Figure 9 SolidWorks Model with full boundary conditions
Next, the distance of loaded are to the edge must be determined to be far enough so the edge
end effect won’t affect the stress field in the center. In order to do that, the loaded area in the
FEA simulation remain 20 mm while size of the part(square) grows from 100 mm to 600 mm
with 100 mm increment. Each individual case has generated 9 plots so the results can be
compared to determine the end effects from the edge.
The following plots are from 500 mm and 600 mm size parts. It is quite obvious the shape of the
plots is nearly same and it is safe to conclude that 500 mm square is big enough for this
particular topic i.e. (ratio of the loaded area and actual size of the edge has to be more than
1:25). Also, for the demonstration purpose, the 100 mm size parts are also included to show the
effect of increase the size.
X normal stress along line 1
Figure 10 X normal stress along line 1 500*500 mm part Figure11X normal stress along line 1 600*600 mm part
Line 3
7. Figure 12 X normal stress along line 1 100*100 mm part
Y normal stress along line 2
Figure 14 Y normal stress along line 2 600*600 mm
part
Figure 13 Y normal stress along line 1 500*500 mm
part
Figure 15 Y normal stress along line 1 100*100 mm
part
8. XY shear stress along line 3
Figure 17 xy shear stress along line 3 100*100 mm part
Finally mesh size is refined as small as possible for 500*500 mm part so that the FEA results plot
are close to the plots from theory. The mesh size for the region under the loaded area is refined
and global mesh size remained relatively bigger (7.4mm) to reduce the time for processing. The
mesh size started from 5mm to 0.8 mm. Due to the limitation of the computing power, finer
results took very long time and system crushes after a few attempts.
Following plots are the best result and theory plots are added to compare.
Figure 16 x normal stress along line 1 vs theory
Figure 15 xy shear stress along line 3 500*500
mm part
Figure 16 xy shear stress along line 3 600*600
mm part
9. Figure 17 y normal stress along line 1 vs theory
Figure 18 xy shear stress along line 1 vs theory
Conclusion and Discussion
By comparing the theory and actual FEA simulation, it can be concluded that FEA generates
reasonable plots with boundary conditions to simulate the theory of a half space as plane stress.
The boundary conditions are fixtures on bottom and sides with roller support on both plane
surfaces. The x-normal stress have the best result in very refined mesh, the plots from FEA and
theory are almost same. The y normal and xy shear stress have some edge end effect, this could
also because the mesh is not refined enough.
It is worth mentioning that the stress field of xy shear stress in line 2 is almost 0 in the theory
and FEA plots generates very small stress in the magnitude of 1-2 pa. But in the plots they seem
very drastic because of the range of the axises.
The improvements for any future progress on this topic are refined mesh and study of the edge
end effect in the FEA.
10. Appendix A Matlab Code
%ME 7210 final project
%equation from text book sadd 3rd edition 8.4.50
%half space uniform loading
clear
clc
p=500; %uniform loading 500
%%
t11d=linspace(0.01,84)
t11=degtorad(t11d);
t12=degtorad(90);
sx1=-p/(2*pi())*(2*(t12-t11)+(sin(2*t12)-sin(2*t11)))
sy1=-p/(2*pi())*(2*(t12-t11)-(sin(2*t12)-sin(2*t11)))
sxy1=p/(2*pi())*(cos(2*t12)-cos(2*t11))
d=tan(t11)*20
plot(d,sxy1) %manualy change for different stress
title('xy shear stress along line1')
xlabel('distance in mm')
ylabel('stress in pa')
%%
clear
clc
p=500;
t21d=linspace(0.01,84);
t22d=180-t21d;
t21=degtorad(t21d);
t22=degtorad(t22d);
d=(tan(t21)*10)/100;
sx2=-p/(2*pi())*(2*(t22-t21)+(sin(2*t22)-sin(2*t21)))
11. sy2=-p/(2*pi())*(2*(t22-t21)-(sin(2*t22)-sin(2*t21)))
sxy2=p/(2*pi())*(cos(2*t22)-cos(2*t21))
plot(d,sy2) %manualy change for different stress
title('y normal stress along line2')
xlabel('distance in mm')
ylabel('stress in pa')
%%
clear
clc
p=500
t32d=linspace(179,100)
t31=degtorad(90);
t32=degtorad(t32d);
d=tan(3.14159-t32)*20/100
sx3=-p/(2*pi())*(2*(t32-t31)+(sin(2*t32)-sin(2*t31)))
sy3=-p/(2*pi())*(2*(t32-t31)-(sin(2*t32)-sin(2*t31)))
sxy3=p/(2*pi())*(cos(2*t32)-cos(2*t31))
plot(d,(sxy3)) %manualy change for different stress
title('xy shear stress along line3')
xlabel('distance in mm')
ylabel('stress in pa')