• Nie Znaleziono Wyników

Theory - equilibrium equations

N/A
N/A
Protected

Academic year: 2021

Share "Theory - equilibrium equations"

Copied!
7
0
0

Pełen tekst

(1)

FEM for bar structures (statics)

Jerzy Pamin

e-mail: Jerzy.Pamin@pk.edu.pl

Piotr Pluciński

e-mail: Piotr.Plucinski@pk.edu.pl

Chair for Computational Engineering

Faculty of Civil Engineering, Cracow University of Technology URL: www.CCE.pk.edu.pl

Computational Methods, 2020 J.Paminc

Theory - equilibrium equations

Virtual work principle

δWint = δWext ∀δu Disassembly into finite elements

δWint = X

e

δWinte , δWext = X

e

δWexte

Equivalent to requirement of minimum of total potential energy functional in the space of admissible displacements

le

xe, u 1

2

ue1

ue2

le xe ye, v

1 2

ve1 ϕe1

v2e ϕe2

le xe, u ye, v

1

2

ue1 ve1 ϕe1

ue2 v2e ϕe2

truss FE beam FE frame FE

Element relations written in local (element) coordinate system

(2)

Discretization

3

1

2

100kN

10kN/m 50kNm

20kN/m

4 2

3

1

x y

x1

x2

x3

d6

d4

d1 d3

d7

d2

d5

d8

d9

d12 d11

d10

Topology (element connectivities)

TOP =

" 1 2 2 3 3 4

# e = 1 e = 2 e = 3

Data:

Nodal coordinates

Cross-section stiffnesses EA and EI

Boundary conditions Loading

Discretization:

NN=4, NE=3, NDOFN=3, NDOF=NN*NDOFN=12

Vector of degees of freedom (dofs):

de=

de1 de2 de3 de4 de5 de6

, d =

d1

d2 d3

d4 d5

d6 d7

d8 d9

d10 d11

d12

Boundary conditions:

d1= d2= d3= d10 = d11 = 0

Computational Methods, 2020 J.Paminc

Theory - equilibrium equations

Variables for frame structures

u =

 u v



- displacement vector (fundamental unknown)

e =

 0 κ



- generalized strain vector (e = Lu) s =

 N M



- generalized stress vector (s = De)

p =

 px py



- distributed load intensity vector

Galerkin approximation within element

u(xe) = N(xe)de, δu = N(xe)δde

de - degrees of freedom (dofs), i.e. nodal displacements e(xe) = B(xe)de, B = LN , δe = B(xe)δde

s(xe) = D e(xe) = DB(xe)de

(3)

Theory - equilibrium equations

Virtual work principle

δWinte = Z le

0

δeTs dxe

δWexte = Z le

0

δuTp dxe+ δdeTfe

fe - nodal force vector (forces acting on the considered element, which come from elements connected to it at nodes)

le xe ye

f1e

f2e f3e

f4e f5e

f6e py px

fe =

f1e f2e f3e f4e f5e f6e

=

−N1

Q1

−M1

N2

−Q2

M2

Computational Methods, 2020 J.Paminc

Theory - equilibrium equations

Substitute approximation

δWinte = δdeT Z le

0

BTDB dxe de = δdeTKede Ke - element stiffness matrix

δWexte = δdeT Z le

0

NTp dxe + δdeTfe = δdeT(ze + fe)

ze - equivalent joint loads (substitute nodal forces) Invoke δWinte = δWexte ∀δde

Element balance equations

Kede = ze + fe

(4)

Beam element description

Bending representation

Definitions of displacement, generalized strain and generalized stress u(x) = [v(x)], e(x) = [κ(x)], s(x) = [M (x)]

Kinematic and constitutive relations at point P (x, y, z) = P (x, 0, 0) = P (x) on beam axis

κ(x)=−d2v(x)

dx2 e = Lu, L =



d2 dx2



M (x)=EI(x) κ(x) s = De, D =

EI(x) 

Computational Methods, 2020 J.Paminc

Beam element description

Approximation of deflection

le xe ye, v

1 2

v1e ϕe1

ve2 ϕe2

N DOFn = 2, N DOFe = 4 dw

[2×1]

= {vw, ϕw} de

[4×1]= {v1e, ϕe1, v2e, ϕe2} u(xe)

[1×1]

= [v(xe)] = N(xe)

[1×4]

de

[4×1] , N = [N1e N2e N3e N4e]

0 le

1

xe N1e(xe) = 1 − 3 xlee

2

+ 2 xlee

3

0 le

1

xe N3e(xe) = 3 xlee

2

− 2 xlee

3

N2e(xe) = xe

1 − xlee

2

N4e(xe) = xe

h xe le

2

xleei

(5)

Beam element description

Approximation of curvature and bending moment, stiffness matrix

e(xe)

[1×1]

= [κ(xe)] = LN(xe) · de = B(xe)

[1×4]

· de

[4×1]

s(xe)

[1×1]

= [M (xe)] = D

[1×1]

· B(xe)

[1×4]

· de

[4×1]

Ke

[4×4]

=

le

Z

0

BTDB dxe

Ke = EeIe le3

12 6le −12 6le 6le 4le2 −6le 2le2

−12 −6le 12 −6le 6le 2le2 −6le 4le2

Computational Methods, 2020 J.Paminc

Beam element description

Computation of substitute nodal forces for constant distributed loading

le xe ye

z1e

z2e z3e

z4e py

pyle 2 pyle2

12

pyle 2

pyle2 12

ze =

le

Z

0

NT

py  dxe

ze =

pyle py2le2

p12yle 2

−pyle2 12

(6)

Global balance

Transformation T

e

: global → local

Matrices referred to local element axes will be marked by overbar d¯e = Tede, ze = TeT¯ze, Ke = TeTK¯eTe

Element balance equations in local coordinates

¯fe = ¯Ked¯e − ¯ze Element balance equations in global coordinates

fe = Kede − ze

Assembly

K =X

e

Ke, d = X

e

de, z = X

e

ze, f = X

e

fe

f = K d − z = w + r w - external point load vector

r - support reaction vector

Computational Methods, 2020 J.Paminc

Computation algorithm

3

1

2

100kN

10kN/m 50kNm

20kN/m

4 2

3

1

x y

x1

x2

x3

d6 d4

r3

d7

d5

d8

d9

r11 r2

r1 d12 r10

System equilibrium

Global vectors:

d =

d1 d2

d3 d4 d5 d6 d7 d8 d9 d10 d11 d12

, r =

r1 r2

r3 r4 r5 r6 r7 r8 r9 r10 r11 r12

Boundary conditions:

d1= d2= d3= d10 = d11 = 0 Hence:

r4= r5= r6= r7= r8= r9= r12 = 0 Element equilibrium

(7)

Algorithm of FE computations for a bar structure

Equilibrium of discretized system (of nodes) K d = w + z + r

plus essential boundary conditions

Statics

1. Discretize (set numbers, axes, topology), prepare input data 2. Compute element matrices ¯Ke, Ke, assemble global matrix K 3. Compute element vectors ¯ze, ze, assemble global vector z,

set up point load vector w

4. Solve equation set Kd = w + z + r taking into account kinematic boundary conditions, i.e. compute unknown nodal displacements in d and reactions in r

Computational Methods, 2020 J.Paminc

Algorithm of FE computations for a bar structure

Statics (cont’d)

Divide matrices into blocks

 K11 K12 K21 K22

  d1 d2



=

 w1+ z1 w2+ z2

 +

 r1 r2



Known displacements d2 = ˆd, hence r1 = 0 K11d1 = w1+ z1− K12d → dˆ 1 → d r = Kd − z − w

5. Compute nodal forces in elements d → de → ¯de → ¯fe = ¯Ked¯e − ¯ze or d → de → fe = Kede− ze → ¯fe

6. Plot diagrams of section forces, check equilibrium

Cytaty

Powiązane dokumenty

The overall theory is complete in two important respects: conventional elementary set theory axiomatizes the class of algebraic models, and the axioms provided for the

Theorem 3.24 (The Least Fixed Point) Let A be a set.. Note: In this proof, it is essential that A is a set, or at least: that some pre-fixed point is a set: cf. the definition of I

The most famous recent result is in the area of extending P t -sets and is due to Baker and Davenport [1], who used Diophan- tine approximation to show that the P 1 -set {1, 3, 8,

In this paper we study the complexity of the sets of all Cauchy problems for ODEs that have a unique solution, and the ones that have a global solution. defined on all of R) solution

[r]

Confidentiality An editor and any editorial staff must not disclose any information about a submitted manuscript to anyone other than the corresponding author,

Discretize (set numbers, axes, topology), prepare input data 2. compute unknown nodal displacements in Q and reactions

1) Total world energy use of all types in subsequent years [table]. 2) Energy use in the world economy in subsequent years (in columns), and by sectors (in rows) [table]. 3)