Finite Element Methods

Exercise Sheets

Lecturer: Prof. Kriegesmann

1 Discrete Systems
Problem 1.1: 1D Discrete Spring System

You are given the above shown discrete 1D spring system consisting of 4 nodes and 6 springs.

Given quantities:

k1 = k2 = 1, k3 = k4 = 2, k5 = k6 = 4, F = 16

1.1.a: Assemble the connectivity matrix for the system.

1.1.b: Assemble the global stiffness matrix K for the system.
1.1.c: Assemble the global force vector for the system.
1.1.d: Solve the system for displacements. Take into account the boundary conditions.
1.1.e: Calculate the reaction forces in the constrained degrees of freedom.

1
1.1.a: (Result)

dofs
 
1 2
1 2

element
 
2 3
 
2 3
 
1 3
3 4

1.1.b: (Result)

 
+6 −2 −4 0
−2 +6 −4 0
K= 
−4 −4 +12 −4
0 0 −4 +4

1.1.c: (Result)

 T
f = 0 0 0 16

1.1.d: (Result)

 
0
2
u=
3


1.1.e: (Result)

   
f1 = −16

2
Problem 1.2: 2D Discrete Spring System

You are given the above shown discrete 2D spring system consisting of 4 nodes and 3 springs.

Given quantities:

k1 = k3 = 2, k2 = 1
F =2
 
1 1 −1 0 0
T1 = √
2 0 0 1 −1
 
0 −1 0 0
T2 =
0 0 0 −1
 
1 1 1 0 0
T3 = √
2 0 0 1 1

1.2.a: Assemble the connectivity matrix for the system. Assume that the first node of an
element is the one with the lower global node number and the second node of an element is the
one with the higher global load number.
1.2.b: Assemble the element stiffness matrices KG,1 , KG,2 and KG,3 , which are given in the
global coordinate system.
1.2.c: Assemble the global stiffness matrix K for the system.
1.2.d: Assemble the global force vector for the system.
1.2.e: Solve the system for displacements. Take into account the boundary conditions.
1.2.f: Calculate the reaction forces in the constrained degrees of freedom.

3
1.2.a: (Result)

dofs 

element

1 2 3 4
3 4 5 6
3 4 7 8

1.2.b: (Result)

 
+1 −1 −1 +1
−1 +1 +1 −1
K1,G =
−1 +1 +1 −1


+1 −1 −1 +1
 
0 0 0 0
0 +1 0 −1
K2,G =
0 0 0 0 


0 −1 0 +1
 
+1 +1 −1 −1
+1 +1 −1 −1
K3,G =
−1 −1 +1 +1


−1 −1 +1 +1

1.2.c: (Result)

−1 −1
 
+1 +1 0 0 0 0
−1 +1 +1 −1 0 0 0 0
 
−1 +1 +2 0 0 0 −1 −1
 
+1 −1 0 +3 0 −1 −1 −1
K=
0

 0 0 0 0 0 0 0 
0
 0 0 −1 0 +1 0 0
0 0 −1 −1 0 0 +1 +1
0 0 −1 −1 0 0 +1 +1

1.2.d: (Result)

 T
f = 0 0 −2 0 0 0 0 0

1.2.e: (Result)

 T
u = 0 0 −1 0 0 0 0 0

4
1.2.f: (Result)

   
f1 1
f2  −1
   
f5   0 
 = 
f6   0 
   
f7   1 
f8 1

5
2 Numerical Integration
Problem 2.1: Newton-Cotes Quadrature in 1D
You are given the function

f (x) = x3 + x2

which you should integrate over the interval

Z 3
f (x)dx
1

2.1.a: Determine the number of required steps n to calculate the integral exactly.
2.1.b: Calculate the integration point coordinates and get the integration point weights.
2.1.c: Evaluate the function at the integration points.
2.1.d: Calculate the integral using the Newton-Cotes method.

6
 2.1.a: (Result)
n = 2 steps. 3 points.

2.1.b: (Result)

x1 = 1, x2 = 2, x3 = 3

1 1 1
w1 = h · 1, w2 = h · 4, w3 = h · 1
3 3 3

2.1.c: (Result)

f (x1 ) = 13 + 12 = 2
f (x2 ) = 23 + 22 = 12
f (x3 ) = 33 + 32 = 36

2.1.d: (Result)

Z 3
86
f (x)dx ≈
1 3

7
Problem 2.2: Newton-Cotes Quadrature in 2D
You are given the function

f (x, y) = x3 y 2

which you should integrate over the interval

Z 2Z 1
f (x, y)dydx
0 −1

2.2.a: Determine the number of required steps nx in x-direction and ny in y-direction to

calculate the integral exactly.
2.2.b: Calculate the integration point coordinates (xi , yj ) and get the integration point
weights wx,i and wy,j .
2.2.c: Evaluate the function at the integration points (xi , yj ).
2.2.d: Calculate the integral using the Newton-Cotes method.

8
 2.2.a: (Result)
nx = 2 steps in x. ny = 2 steps in y.

2.2.b: (Result)

x1 = 0 x2 = 1 x3 = 2
y1 = −1 y2 = 0 y3 = 1

1 1 1 4 1 1
wx,1 = hx · 1 = wx,2 = hx · 4 = wx,3 = hx · 1 =
3 3 3 3 3 3
1 1 1 4 1 1
wy,1 = hy · 1 = wy,2 = hy · 4 = wy,3 = hy · 1 =
3 3 3 3 3 3

2.2.c: (Result)

f (x1 , y1 ) = f (0, −1) = 0

f (x1 , y2 ) = f (0, 0) = 0
f (x1 , y3 ) = f (0, 1) = 0
f (x2 , y1 ) = f (1, −1) = 1
f (x2 , y2 ) = f (1, 0) = 0
f (x2 , y3 ) = f (1, 1) = 1
f (x3 , y1 ) = f (2, −1) = 8
f (x3 , y2 ) = f (2, 0) = 0
f (x3 , y3 ) = f (2, 1) = 8

2.2.d: (Result)

Z 2Z 1
8
f (x)dx =
0 −1 3

9
Problem 2.3: Newton-Cotes Quadrature Order of Accuracy
You are given the functions

f1 (x) = 2x2 + 3x + 1
f2 (x) = 3x + x3
f3 (x) = 1
f4 (x, y) = x2 y 2 + x3 + y
f5 (x, y) = xy + y 4
f6 (x, y) = x2 y 2 + x3 + y 3

Rb Rb Rb
for which the integral a fi (x)dx for f1 to f3 and axx ayy fi (x, y)dydx for f4 to f6 should be
calculated using the Newton-Cotes quadrature scheme.

2.3.a: Determine the minimum number of steps and points required to calculate the integrals
for f1 to f6 exactly. For the multivariate functions f4 to f6 determine the number of steps and
points per variable dimension and in total.

10
2.3.a: (Result)

• f1 ⇒ 2 steps required (3 points)

• f2 ⇒ 2 steps required (3 points)

• f3 ⇒ 1 steps required (2 points)

• f4

– ⇒ 2 steps required in x-direction (3 points)

– ⇒ 2 step required in y-direction (3 points)
– [3 × 3] = 9 points in total

• f5

– ⇒ 1 steps required in x-direction (2 points)

– ⇒ 4 step required in y-direction (5 points)
– [2 × 5] = 10 points in total

• f6

– ⇒ 2 steps required in x-direction (3 points)

– ⇒ 2 step required in y-direction (3 points)
– [3 × 3] = 9 points in total

11
Problem 2.4: Gauss-Legendre Quadrature in 1D
You are given the function

f (x) = x3 + x2

which you should integrate over the interval

√
Z 1+ 3
√ f (x)dx
1− 3

2.4.a: Determine the number of required evaluation points n, the points ri and weights wi to
calculate the integral exactly.
2.4.b: Calculate the integration point coordinates in x.
2.4.c: Evaluate the function at the integration points.
2.4.d: Calculate the integral using the Gauss(-Legendre) quadrature.

12
 2.4.a: (Result)
2 points required.
1
r1 = − √ , w1 = 1
3
1
r2 = + √ , w2 = 1
3

2.4.b: (Result)

x(r) = 0
x(r2 ) = 2

2.4.c: (Result)

f (x(r1 )) = 0
f (x(r2 )) = 12

2.4.d: (Result)

Z b √
f (x)dx ≈ 12 3
a

13
Problem 2.5: Gauss-Legendre Quadrature in 2D
You are given the function

f (x, y) = x5 y 3

which you should integrate over the interval

q
5
√
Z + 3
Z 1+ 3
q √ f (x, y)dydx
− 53 1− 3

2.5.a: Determine the number of required integration points nx in x-direction and ny in y-

direction to calculate the integral exactly using Gauss(-Legendre) quadrature.
2.5.b: Calculate the integration point coordinates (xi , yj ) and get the integration point
weights wx,i and wy,j .
2.5.c: Evaluate the function at the integration points (xi , yj ).
2.5.d: Calculate the integral using the Gauss(-Legendre) quadrature.

14
2.5.a: (Result)

nx = 3
ny = 2

2.5.b: (Result)

r
3
rx,1 = −
5
rx,2 = 0
r
3
rx,3 = +
5

x(rx,1 ) = −1
x(rx,2 ) = 0
x(rx,3 ) = 1

y(ry,1 ) = 0
y(ry,2 ) = 2

wy,1 = 1
wy,2 = 1

2.5.c: (Result)

f (−1, 0) = 0
f (−1, 2) = −8
f (0, 0) = 0
f (0, 2) = 0
f (+1, 0) = 0
f (+1, 2) = 8

2.5.d: (Result)

Z bx Z by
f (x, y)dydx ≈ 0
ax ay

15
Problem 2.6: Gauss-Legendre Quadrature Order of Accuracy
You are given the functions

f1 (x) = 2x2 + 3x + 1
f2 (x) = 3x + x3
f3 (x) = 1
f4 (x, y) = x2 y 2 + x3 + y
f5 (x, y) = xy + y 4
f6 (x, y) = x2 y 2 + x3 + y 3

Rb Rb Rb
for which the integral a fi (x)dx for f1 to f3 and axx ayy fi (x, y)dydx for f4 to f6 should be
calculated using the Gauss(-Legendre) quadrature scheme.

2.6.a: Determine the minimum number of points required to calculate the integrals for f1 to
f6 exactly. For the multivariate functions f4 to f6 determine the number of points per variable
dimension and in total.

16
2.6.a: (Result)

• f1 ⇒ 2 points required

• f2 ⇒ 2 points required

• f3 ⇒ 1 points required

• f4

– ⇒ 2 points required in x-direction

– ⇒ 2 step required in y-direction
– [2 × 2] = 6 points in total

• f5

– ⇒ 1 points required in x-direction

– ⇒ 3 step required in y-direction
– [1 × 3] = 3 points in total

• f6

– ⇒ 2 points required in x-direction

– ⇒ 2 points required in y-direction
– [2 × 2] = 4 points in total

17
3 Isoparametric Elements
Problem 3.1: Isoparametric Mapping
You are given the quadrilateral linear element shown in Figure 1a in natural coordinates and

(a) (b) (c)

Figure 1: (a) Quadrilateral linear element in natural coordinate system. (b) undeformed element
in global coordinates. (c) deformed element in global coordinates.

shown in Figure 1b in global coordinates as well as in its deformed configuration in Figure 1c.
Your task is to calculate the global coordinates as well as global displacements of the point rip
in the element.

Given quantities:
1 T
rip = 1 1
2
 T
x = x1,x x1,y x2,x x2,y x3,x x3,y x4,x x4,y
 T
= 1 1 3 2 4 4 2 3
 T
u = u1,x u1,y u2,x u2,y u3,x u3,y u4,x u4,y
 T
= 0 0 1 1 1 1 1 1

3.1.a: Calculate the shape function matrix N.

3.1.b: Calculate the global coordinates x of the point rip .
3.1.c: Calculate the global displacements u of the point rip .

18
3.1.a: (Result)

 
1 1 0 3 0 9 0 3 0
N (rip ) =
16 0 1 0 3 0 9 0 3

3.1.b: (Result)

 
1 52
x=
16 52

3.1.c: (Result)

 
1 15
u=
16 15

19
Problem 3.2: Strain-Displacement Matrix

(a) Quadrilateral linear element (b) Quadrilateral linear element

in natural coordinate system. in global coordinate system.

You are given the isoparametric quadrilateral linear element shown above once in the global
coordinates in Figure 2b and once in the local coordinates in Figure 2a. You task is to calculate
the strain-displacement-matrix B for this element for the shown integration point rip .

Given quantities:
1 T
rip = 1 1
2
 T
x = x1,x x1,y x2,x x2,y x3,x x3,y x4,x x4,y
 T
= 1 1 3 2 4 4 2 3

3.2.a: Calculate the derivatives ∂x

∂r of the approximated global coordinates x w.r.t. the
natural coordinates r at the integration point.
3.2.b: Calculate the shape function Jacobian J, its determinant det (J) and its inverse J−1
at the integration point.
3.2.c: Calculate the shape function derivatives w.r.t. the global coordinates x.
3.2.d: Assemble the B matrix for the integration point.

20
3.2.a: (Result)

∂x
(rip ) = 1
∂r
∂x 1
(rip ) =
∂s 2
∂y 1
(rip ) =
∂r 2
∂y
(rip ) = 1
∂s

3.2.b: (Result)

1
 
1 2
J (rip ) = 1
2 1

3
det (J (rip )) =
4

− 21
 
−1 4 1
J =
3 − 21 1

3.2.c: (Result)

 ∂h   
1 −1
1
∂x
 =
∂h1 12 −1
∂y
 ∂h   
2


∂x
= 1 +5
∂h2 12 −7
∂y
 ∂h   
3


∂x
= 1 +3
∂h3 12 +3
∂y
 ∂h   
1 −7
4
∂x
 =
∂h4 12 +5
∂y

3.2.d: (Result)

 
    
−1 0 +5 0 +3 0 −7 0 
1 

B=  0 −1  0 −7  0 +3  0 +5

12 
 −1 −1 −7 +5 +3 +3 +5 −7 

| {z }| {z }| {z }| {z }
B1 B2 B3 B4

21
Problem 3.3: Body Force Vector

You are given the above shown triangular linear isoparametric element. The element is subjected
to the body force field f . The element has the global nodal coordinates x1 , x2 and x3 and a
thickness of t. The element has a single integration point having the natural coordinates r1 and
weight w1 .
Given quantities:
 2         
x −y 2 3 1 1 1 1
f (x) = , x1 = , x2 = , x3 = , t = 2, r1 = , w1 =
2x + y 1 2 3 3 1 2

3.3.a: Evaluate the shape function matrix N at the integration point.

3.3.b: Calculate the shape function Jacobian J and its determinant det (J).
3.3.c: Calculate the nodal body force vector fbody for the element.

22
3.3.a: (Result)

 
1 1 0 1 0 1 0
N=
3 0 1 0 1 0 1

3.3.b: (Result)

 
1 1
J=
−1 2

det (J) = (1 · 2) − (−1 · 1) = 3

3.3.c: (Result)

 
2
6
 
2
fbody,el = 
6
 
2
6

23
Problem 3.4: Strain and Stress Calculation

(a) Undeformed finite element model. (b) Deformed finite element model.

(c) Quadrilateral linear isopara-

metric element in natural coor-
dinate system.

You are given the global displacement vector u for the system shown in Figure 3a and Figure 3b.
The system consists of tow quadrilateral linear elements like shown in Figure 3c. Furthermore,
you are provided the B matrix evaluated at global node 2 for both element I (BI ) and element
II (BII ). The system is modeled using a plane stress assumption, a Young’s modulus of E and
a Poisson ratio of ν.

Given quantities:
1
E = 1, ν =
4
 T
u = u1,x u1,y u2,x u2,y u3,x u3,y u4,x u4,y u5,x u5,y u6,x u6,y
 T
= 0 0 0 0.5 0 0 0 0 0 −0.5 0 0

 
−1 0 1 0 0 0 0 0
BI =  0 0 0 −1 0 1 0 0
0 −1 −1 1 1 0 0 0
 
−1 0 1 0 0 0 0 0
BII = 0 −1 0 0
 0 0 0 1
−1 −1 0 1 0 0 1 0
3.4.a: Calculate the strains at node 2 for element 1 and element 2 by directly evaluating the
stains at the point.
3.4.b: Calculate the stresses at node 2 for element 1 and element 2 using the calculated
strains.
3.4.c: Calcualte the the average stress for node 2.

24
3.4.a: (Result)

 
0
εI = −1
0.5
 
0
εII =  −1 
−0.5

3.4.b: (Result)

 
−2
16  
σI = −8
120
1.5
 
−2
16 
σII = −8 
120
−1.5

3.4.c: (Result)

 
−2
σ = −8
0

25
Problem 3.5: Hourglass Stabilization

(a) (b) (c) (d)

Figure 4: (a)Quadrilateral linear isoparametric element in natural coordinate system. (b)Zero

energy mode 1. (c)Zero energy mode 2. (d) Quadrilateral linear isoparametric element in global
coordinate system (deformed and undeformed).

You are given a quadrilateral linear element applying reduced integration as shown in Figure 4a.
Due to the under-integration, the element has two zero energy modes shown in Figure 4b and
Figure 4c. Your task is to calculate the hourglass stiffness matrix KS to stabilize this ele-
ment using the stabilization stiffness ES . You are then given a displacement vector u, which is
the result of the analysis applying the physical stiffness matrix of the element K and the cal-
culated hourglass stiffness matrix KS . You should then calculate the physical strain energy as
well as the stabilization strain energy of the element and conclude if these results can be trusted.

Given quantities:

−1 −1 −3 −1
 
3 1 1 1
1
 3 1 1 −1 −3 −1 −1

−1 1 3 −1 1 −1 −3 1
 
−1 1 −1 3 1 −1 1 −3
K=
−3

 −1 1 1 3 1 −1 −1

−1 −3 −1 −1 1 3 1 1
 
1 −1 −3 1 −1 1 3 −1
1 −1 1 −3 −1 1 −1 3
 T
u= 0 0 0 1 0 0 0 0
ES = 12

3.5.a: Assemble the shown zero energy modes. Normalize the modes to a length of the modes
to ∥φi ∥ = 1
3.5.b: Assemble the Hourglass stabilization stiffness matrix using the hourglass stiffness ES .
3.5.c: Calculate the physical strain energy of the system.
3.5.d: Calculate the stabilization strain energy of the system.
3.5.e: Can the results be trusted? Explain why?

26
3.5.a: (Result)

1  T
φ1 = −1 0 1 0 −1 0 1 0
2 |
|{z} {z }
1 φ̃1
∥φ̃1 ∥

1  T
φ2 = 0 1 0 −1 0 1 0 −1
2 |
|{z} {z }
1 φ̃2
∥φ̃2 ∥

3.5.b: (Result)

0 −1 0 0 −1 0
 
1 1
0
 1 0 −1 0 1 0 −1

−1 0 1 0 −1 0 1 0
 
 0 −1 0 1 0 −1 0 1 
KS,el =3· 
1
 0 −1 0 1 0 −1 0 

0
 1 0 −1 0 1 0 −1

−1 0 1 0 −1 0 1 0
0 −1 0 1 0 −1 0 1

3.5.c: (Result)

Ephys = 3

3.5.d: (Result)

ES = 3

3.5.e: (Result)
No. Stabilization energy to high in relation to total system energy.

27
4 Structural Elements
Problem 4.1: No Hand Calculations Exercises for this Topic

28
5 Transient Analysis
Problem 5.1: Mass Matrix

You are given the above shown quadrilateral linear element having the nodes 1, 2, 3 an 4 with
the corresponding shape functions h1 , h2 , h3 and h4 . The element has a single integration point
having the natural coordinates rip and weight wip . The element has a density ρ and an out of
plane thickness t.

Given quantities:
1 3 9 3
h1 (rip ) = h2 (rip ) = h3 (rip ) = h4 (rip ) =
16 16 16 16
det (J (rip )) = 16
wip = 2
ρ=2
t=4

Tasks:

5.1.a: Calculate the consistent element mass matrix Mel

5.1.b: Calculated the diagonalized element mass matrix Mel,rs using the row-sum method
5.1.c: Calculate the element mass mel
5.1.d: Calculate the diagonalized mass matrix Mel,ds using the diagonal scaling method

29
5.1.a: (Result)

 
1 0 3 0 9 0 3 0
0 1 0 3 0 9 0 3
 
3 0 9 0 27 0 9 0 
 
0 3 0 9 0 27 0 9 
Mel = 
9

 0 27 0 81 0 27 0 

0 9 0 27 0 81 0 27
 
3 0 9 0 27 0 9 0 
0 3 0 9 0 27 0 9

5.1.b: (Result)

 
16 0 0 0 0 0 0 0
 0 16 0 0 0 0 0 0
 
 0 0 48 0 0 0 0 0
 
 0 0 0 48 0 0 0 0 
Mel,rs 
 0 0 0 0 144 0

 0 0 
0 0 0 0 0 144 0 0 
 
0 0 0 0 0 0 48 0 
0 0 0 0 0 0 0 48

5.1.c: (Result)

mel = 256

5.1.d: (Result)

 
1 0 0 0 0 0 0 0
0 1 0 0 0 0 0 0
 
0 0 9 0 0 0 0 0
 
256 
0 0 0 9 0 0 0 0
Mel,ds = 
0
100  0 0 0 81 0 0 0

0 0 0 0 0 81 0 0
 
0 0 0 0 0 0 9 0
0 0 0 0 0 0 0 9

30
Problem 5.2: Eigenfrequency Analysis

You are given the above shown 2D truss system having the global stiffness matrix K and mass
matrix M. Assume that the degrees of freedom are sorted by global node number and alternating
x- and y-coordinates of the respective node.
Given quantities:

2 0 −2 0 0 0
   
0 0 1 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0
   
−2 0 3 1 0 0 −1 −1 0 0 3 0 0 0 0 0
   
0 0 1 3 0 −2 −1 −1 0 0 0 3 0 0 0 0
K= 
,
 M=  
 0 0 0 0 0 0 0 0   0 0 0 0 1 0 0 0 

 0 0 0 −2 0 2 0 0 0 0 0 0 0 1 0 0
  

 0 0 −1 −1 0 0 1 1 0 0 0 0 0 0 3 0
0 0 −1 −1 0 0 1 1 0 0 0 0 0 0 0 3

5.2.a: Calculate the constrained systems eigenfrequencies.

31
5.2.a: (Result)

r
1
ω1,2 = 1∓
3

32
Problem 5.3: Modal Decomposition
You are given a system consisting of a consistent mass matrix M and a stiffness matrix K.
Furthermore, you are given the eigenvector matrix Q.

Given quantities:
       
9 −2 17 −6 1 1 2 T1 1 2
M= , K= , Q= √ , Q =√
−2 6 −6 8 5 2 −1 5 2 −1
5.3.a: Calculate the generalized mass matrix Mg and generalized stiffness matrix Kg .
5.3.b: Calculate the mass normalized generalized mass matrix M̂g and mass normalized
generalized stiffness matrix K̂g .
5.3.c: Calculate the eigenfrequencies φi of the system.
5.3.d: Apply modal reduction using only the eigenmode having the first lowest eigenfrequency
and calculate the generalized mass matrix and generalized stiffness matrix using the modally
reduced model.

33
5.3.a: (Result)

 
5 0
Mg =
0 10

 
5 0
Kg =
0 20

5.3.b: (Result)

 
1 0
M̂g =
0 1

 
1 0
K̂g =
0 2

5.3.c: (Result)

ω1 = 1
√
ω2 = 2

5.3.d: (Result)

Mg,red = 5

Kg,red = 5

34
Problem 5.4: Explicit Time Integration
   
1 0 0 9 0 0
0 1 0 p̈(t) + 0 16 0 p(t) = 0
0 0 1 0 0 1

You are given the above shown dynamic modally decomposed system shown above.

Given quantities:
 T
p(0) = 0 2 32
 T
ṗ(0) = 8 4 4
1
∆t = .
4
5.4.a: Calculate the critical step size tcrit for explicit central differences time integration for
the given system.
5.4.b: Calculate the initial condition q̈(0) and calculate q(−∆t) using the Taylor-expansion
approach.
5.4.c: Perform 2 time steps using ∆t and the explicit central differences scheme.
5.4.d: Calculate the velocity and acceleration at t = ∆t.

35
5.4.a: (Result)

tcrit = 0.5

5.4.b: (Result)

 
0
q̈(0) = −32
−32

 
−2
ut0 −∆t = 0 
30

5.4.c: (Result)

 
2
u∆t = 2

32

 46 
16
u2∆t =  0 
30

5.4.d: (Result)

ut+∆t − ut−∆t
u̇∆t =
 2∆t 
23
1
== −16
4
−16
ut+∆t − 2ut + ut−∆t
ü∆t = 2
  ∆t
−18
= −32

−32

36
6 Formulary
Discrete Systems
Local stiffness matrix for 1D Spring Element
 
1 −1
Kel = ke
−1 1

Local stiffness matrix for 1D Truss Element

 
EA 1 −1
Kel =
L −1 1
|{z}
ke

Transformation Matrix for Rotated 1D Elements

 
cos(α) sin(α) 0 0
T (α) =
0 0 cos α) sin(α)

Coordinate Transformation between local coordinates uL and global coordinates uG

for rotated 1D Spring/Truss Elements
 
    u1,x
u1 cos(α) sin(α) 0 0 u1,y 
=  
u2 0 0 cos α) sin(α) u2,x 
| {z } | {z } u
uL T 2,y
| {z }
uG

Stiffness Matrix of rotated 1D Spring/Truss Element

Ke,G = TT Ke,L T

37
Numerical Integration
Newton-Cotes quadrature formula for 1D
Z b n
X
f (x)dx ≈ wi f (xi )
a i=1

Newton-Cotes quadrature formula for 2D

Z bx Z by ny
nx X
X
f (x, y)dydx ≈ wx,j wy,i f (xi , yj )
ax ay i=1 j=1

Newton-Cotes quadrature order of accuracy p expressed in terms of number of steps

n (applicable per dimension)
(
n , if n is odd
p=
n + 1 , if n is even

Newton-Cotes quadrature integration point formula (equidistant n+1 points in [a, b])

xi = a + (i − 1) · h, i ∈ 1, ..., n

Newton-Cotes quadrature weights table

Steps n Step Size h Order p Formula

b−a 1
1 1 1 2 h (f (x1 ) + f (x2 ))
b−a 1
2 2 3 3 h (f (x1 ) + 4 · f (x2 ) + f (x3 ))
b−a 3
3 3 3 8 h (f (x1 ) + 3f (x2 ) + 3f (x3 ) + f (x4 ))
b−a 2
4 4 5 45 h (f (x1 ) + 32f (x2 ) + 12f (x3 ) + 32f (x4 ) + 7f (x5 ))

Gauss(-Legendre) quadrature formula for 1D

Z b n  
b−aX b−a a+b
f (x)dx ≈ wi f ri +
a 2 2 2
i=1

Gauss(-Legendre) quadrature formula for 2D

bx by ny
nx X  
bx − ax by − ay X bx − ax ax + bx by − ay
Z Z
ay + by
f (x, y)dydx ≈ wx,i wy,j f rx,i + , ry,i +
ax ay 2 2 2 2 2 2
i=1 j=1

Gauss(-Legendre) quadrature integral transformation

x(r) : [−1, 1] → [a, b]

b−a a+b dx b−a
x(r) := r+ ⇒ :=
2 2 dr 2
Z b Z 1 Z 1  
dx b−a b−a a+b
f (x)dx = f (x(r)) dr = f r+ dr
a −1 dr 2 −1 2 2

Gauss(-Legendre) quadrature order of accuracy p expressed in terms of evaluation

points n (applicable per dimension)

p = 2n − 1

38
Gauss(-Legendre) quadrature points and weights table

Points n Points ri Weights wi Order

1 0 2 1
2 ± √13 1 3
8
0 9
3 q 5
3 5
± 5 9
r q √
3 2 6 18+ 30
± 7 − 7 5 36
4 r 7
q √
18− 30
± 37 + 27 65 36

39
Isoparametric Elements
Shape functions and their derivatives for a quadrilateral linear element given in
natural coordinates r and s:
1 ∂h1 1 ∂h1 1
h1 (r, s) = (1 − r)(1 − s), = − (1 − s), = − (1 − r)
4 ∂r 4 ∂s 4
1 ∂h2 1 ∂h2 1
h2 (r, s) = (1 + r)(1 − s), = (1 − s), = − (1 + r)
4 ∂r 4 ∂s 4
1 ∂h3 1 ∂h3 1
h3 (r, s) = (1 + r)(1 + s), = (1 + s), = (1 + r)
4 ∂r 4 ∂s 4
1 ∂h4 1 ∂h4 1
h4 (r, s) = (1 − r)(1 + s), = − (1 + s), = (1 − r)
4 ∂r 4 ∂s 4
Shape functions and their derivatives for a triangular linear element given in natural
coordinates r and s:
∂h1 ∂h1
h1 (r, s) = 1 − r − s, = −1, = −1
∂r ∂s
∂h2 ∂h2
h2 (r, s) = r, = 1, =0
∂r ∂s
∂h3 ∂h3
h3 (r, s) = s, = 0, =1
∂r ∂s
Shape function matrix N for 2D:
 
h 0 ... hnnodes 0
N= 1
0 h1 ... 0 hnnodes

Isoparametric element nodal coordinates (2D):

T
x = x1 y 1 x2 y 2 ... xnnodes y nnodes ∈ R[ndofs ×1]


Isoparametric element nodal displacements (2D):

T
u = u1,x u1,y u2,x u2,y ... unnodes ,x unnodes ,y ∈ R[ndofs ×1]


Isoparametric element approximation of global coordinates x and displacements u

(2D):
nX
nodes nX
nodes

x= hI xI ux = hI uI,x
I=1 I=1
nX
nodes nX
nodes

y= hI y I uy = hI uI,y
I=1 I=1
x = Nx u = Nu

Jacobian matrix J of shape functions:

 ∂x ∂y  Pnsf ∂hi Pnsf ∂hi

∂r ∂r i=1 ∂r xi i=1 ∂r y i
J= =
Pnsf Pnsf

∂x ∂y ∂hi ∂hi
∂s ∂s i=1 ∂s xi i=1 ∂s y i

Shape function derivatives with respect to global coordinates x and y:

 ∂h   ∂h 
i i
∂x ∂r
  = J−1  
∂hi ∂hi
∂y ∂s

40
Strain-displacement matrix B in 2D for a single shape function I:
 ∂h 
∂x
I
0
 
∂h
 
 0
I
BI =  ∂y 
 
∂hI ∂hI
∂y ∂x

Strain-displacement matrix B in 2D for the entire element:


∂h1 ∂h2 ∂hnsf 
0 0 ... 0
 ∂x ∂x ∂x

 
∂h1 ∂h2 ∂h n
B= 0 sf 

∂y 0 ∂y ... 0 ∂y 


 
∂h1 ∂h1 ∂h2 ∂h2 ∂hnsf ∂hnsf
∂y ∂x ∂y ∂x ... ∂y ∂x

Constitutive matrix C for plane stress:

 
1 ν 0
E 
C= ν 1 0 
1 − ν2 1−ν
0 0 2

Constitutive matrix C for plane strain:

 
1−ν ν 0
E
C=  ν 1−ν 0 
(1 + ν)(1 − 2ν) 1−2ν
0 0 2

Elemental stiffness matrix Kel for 2D isoparametric element:

nip
X
Kel = BT CB det(J)wip t
ip

Elemental stiffness matrix entries KIJ of degrees of freedom I and J for the element
stiffness matrix for 2D isoparametric element:
  Xnip
Kel,II Kel,IJ
Kel,IJ = = BTJ CBI det(J)wip t
Kel,JI Kel,JJ
ip

Nodal body force vector fel for 2D isoparametric element:

nip
X
fbody,el = NT f (Nx) det(J)wip t
ip

Nodal body force vector entries fel,J for node J for 2D isoparametric element:
nip
X
fbody,el,J = hJ f (Nx) det(J)wip t
ip

Strain ε for integration point in isoparametric element:

ε = Bu

41
Stress σ for integration point in isoparametric element:

σ = Cε = CBu

Hourglass stiffness matrix KS,el for 2D isoparametric element:

nmodes
X
KS,el = ES φi φTi
i=1

Physical internal energy Ephys of the entire system:

Ephys = uT Ku

Hourglas stabilization energy ES of the entire system:

ES = uT KS u

42
Transient Analysis
Mass matrix Mel for 2D isoparametric element:
nip
X
Mel = ρNT N det(J)wip t
ip

Element mass mel for 2D isoparametric element:

nip
X
mel = ρ det (J) wip t
ip=1

= 1T Mel 1

Diagonalized mass matrix Mel,rs using the row-sum method:

n
X
Mel,rs,ii = Mel,ij
j=1

Diagonalized mass matrix Mel,ds using the diagonal scaling method:

mel
c=
1T diag (Mel )
Mel,ds,ii = cMel,ij

Generalized Eigenvalue problem for transient systems:

(K − λM) q = 0

Characteristic polynomial of generalized Eigenvalue problem for transient systems:

(K − λM) q = 0

Eigenvector matrix of a system (K, M): (qi : ith-eigenvector)

 
Q = q1 q2 ... qndof s

Eigenvector matrix of a system (K, M): (qi : ith-eigenvector)

 
Qred = q1 q2 ... qnmodes

Generalized mass matrix for modal decomposition:

Mg = QT MQ

Generalized stiffness matrix for modal decomposition:

Kg = QT KQ

Generalized force vector for modal decomposition:

fg = QT f

Mass normalized generalized mass matrix for modal decomposition:

M̂g = I

43
Mass normalized generalized stiffness matrix for modal decomposition:

K̂g = M−1 T
g Q KQ = Λ

Mass normalized generalized force vector for modal decomposition:

f̂g = M−1 T
g Q f

Generalized mass matrix for modal reduction:

Mg,red = QTred MQred

Generalized stiffness matrix for modal reduction:

Kg,red = QTred KQred

Generalized force vector for modal reduction:

fg,red = QTred f

Critical step size for explicit central difference time integration:

2
tcrit = , ωmax = max(ωi )
ωmax
Initial condition for displacement at t = −∆t for explicit central differences time
integration using Taylor-expansion:
1
ut0 −∆t ≈ ut0 − u̇t0 ∆t + üt0 ∆t2
2
Explicit central difference scheme:
  
2 −1 1
ut+∆t = ∆t M ft − K [ut ] − M (ut−∆t − 2ut )
∆t2
ut+∆t − ut−∆t
u̇t =
2∆t
ut+∆t − 2ut + ut−∆t
üt =
∆t2

44
Auxiliary vectors and matrix formulas
One vector to get x-direction entry for each node in a system matrix
T
1 = 1 0 1 0 ... 1 0 ∈ R[ndofs ×1]


Determinant of a [2 × 2] matrix
 
a b
A= ⇒ det (A) = a · d − b · c
c d

Inverse of a [2 × 2] matrix
   
a b −1 1 d −b
A= ⇔A =
c d det (A) −c a

45
Notation

• x, y global coordiante system

• r, s: natural coordiante system
• hi : ith shape function of an isoparametric element
• nnodes : Number of nodes/shape functions in an isoparametric element
• N: Shape function matrix
• x: nodal global coordinates of an isoparametric element
• u: nodal global displacements of an isoparametric element
• x: global coordinate approximation of an isoparametric element
• u: global displacement approximation of an isoparametric element
• J: Shape function Jacobian
• BI : Strain-displacement matrix of the Ith shape function of an isoparametric element
• B: Strain-displacement matrix of an entire isoparametric element
• C: Stress-strain matrix
• Kel : Elemental stiffness matrix
• fbody,el : Elemental nodal body force vector
• ε: strain vector in Voigt notation
• σ: stress vector in Voigt notation
• φi : ith hourglas mode of reduced integrated isoparametric element
• KS : Elemental hourglas stiffness matrix
• Ephys : Physical internal energy of the entire system
• ES : Hourglas stabilization energy of the entire system
• Mel : Elemental mass matrix
• mel : Elemental mass
• q: eigenvectors of a system (K, M)
• p: generalized coordinates
• Mel,rs : Diagonalized elemental mass matrix using the row-sum method
• Mel,ds : Diagonalized elemental mass matrix using the diagonal scaling method
• Q: Eigenvector matrix containing the eigenvectors of a system (K, M)
• Qred : Reduced eigenvector matrix ⇒ a subset of the eigenvectors of a system (K, M)
• Mg : Generalized mass matrix
• Mg,red : Generalized mass matrix applying modal reduction
• M̂g : Mass normalized generalized mass matrix (identity matrix)
• Kg : Generalized stiffness matrix
• Kg,red : Generalized stiffness matrix applying modal reduction
• K̂g : Mass normalized generalized stiffness matrix (eigenvalues of (K, M) on diagonal)

46