midas Meshfree Analysis Manual

Table of Contents

1. Introduction
2. Basic formulation
3. Analysis
4. Load and constraints
1. Introduction

This manual describes the concepts and theories employed in midas Meshfree. The material
presented in the manual is aimed at users with basic knowledge in computational mechanics
and related engineering fields. As many users are familiar with finite element method (FEM),
descriptions are made extensively utilizing notations and algorithm that originate from
classical FEM.

In chapter 2, key concepts employed and formulations for structural and heat transfer
analyses are explained.

In chapter 3…blah blah

Midas Meshfree is a general-purpose analysis solution capable of solving static and dynamic
problems in structural mechanics. Also analysis of heat transfer and sequentially coupled
structural thermal expansion problems can be carried out. Table 1.1 summarizes the
analysis features.

Table 1.1. Analyses feature of midas Meshfree

Category Supported analyses
Structural mechanics  Linear structural statics
 Nonlinear structural statics with elasto-plastic materials
 Normal mode analysis
 Linear direct dynamics
 Linear modal dynamics
 Pre-stressed static and normal mode
Heat transfer and coupled  Steady-state heat transfer
 Transient heat transfer
 Sequentially coupled thermal stress

1.1 Introduction to Meshfree concept

Although finite element technology and commercial applications have matured and proven in
industrial usage to boost productivity, small businesses still lag behind in adopting simulation
technologies in product designs. In this regard, the immediate obstacle is converting the
design CAD model into finite element discretized analysis model.

The “meshless” and “meshfree” terminologies in computational mechanics have been

around with abundant researches conducted. Some methods are targeted at handling
extreme nonlinearities while others are adequate for multiscale modeling. The “meshfree”
concept in midas Meshfree emphasizes the ease of use and is directly targeted at product
designers; being free from laborious process of mesh generation and related cleanup.

1.2 Notations and sign conventions

Midas Meshfree uses the following sign conventions for stress for consistency and to avoid
confusion. Note that positive normal stress refers to tension state while negative normal
stress the compression state.
 z
 zz

 zx  zy
 yz
 xz  yy
 xy  yx
 xx

x
Figure 1.1 Sign convention for stress in midas Meshfree
2. Basic concepts and formulation
This chapter describes the basic concepts and formulation of midas Meshfree. The key
concept that is unique compared to traditional FEM can be summarized as follows.

- The mesh or grids do not conform to the geometry

The domain of computation is immersed inside structured grids. Size of grids can be
controlled independently for individual CAD part. Similar concepts can be found in finite cell
method (ref. 2.1), immersed boundary method in the computational fluid dynamics (ref 2.2).

- Grids are automatically generated by the solver

For given three-dimensional CAD model, the grids are generated automatically without user
intervention. Unlike automatic mesh generation in FEM which fail more often than not
without geometry simplifications and clean ups, success rate is nearly one hundred per cent.

2.1 Compute domain immersed grid element formulation

For linear elasticity, Hu-Washizu variational principle (ref 2.3, 2.4) can be regarded as the
most fundamental equation encompassing equations of equilibrium, constitutive and
compatibility:

 Gext   (  u)T σ   εT (Dε  σ)   σT ( u  ε)d  [2-1]



 Gext : virtual work from external loads

u : displacement vector
σ : stress vector
ε : strain vector
D : elasticity matrix
 : strain-displacement operator

This generic form can be reduced to Hellinger-Reissner variational principle (ref 2.5, 2.6)
under the assumption that strain and stress satisfy the constitutive equation.

 Gext   (  u)T σ   σT ( u  D1σ)d  [2-2]



This can further be transformed to the principle of virtual work when compatibility relations
between ε and  u are satisfied.

 Gext   (  u)T D ud  [2-3]



For the sake of simplicity, further derivation assumes linear elastic structural analysis based
on assumed displacement, single field formulation.
 p  x
Sf

eD
 u  x  uˆ  x

Interior grid element

Boundary grid element (Neumann)

Su
Ce
Boundary grid element (Dirichlet)

Figure 2.1 Concept of grid elements with immersed computational domain

Figure 2.1 shows spatial grid with computational domain,  immersed inside. The
computational domain is discretized by grid elements in a nonconforming manner. Each grid
element can be classified into interior and boundary grid elements. Boundary grid elements
can be further divided into those with Dirichlet boundary condition applied and those without,
i.e. with Neumann boundary condition. For convenience, interior and Neumann boundary
grid element domain is designated as Ce and Dirichlet boundary grid element domain is
designated as  eD .

Similar to finite element procedure, we need to restrict the integral domain to single grid
element for applying either of the variational principles. For interior grid elements and
boundary elements with Neumann boundary condition, the displacement vector, u is
interpolated by shape functions and associated nodal degrees of freedom values.

uh  Nde [2-4]

N : shape function matrix

de : element degrees of freedom vector

For boundary grid elements where Dirichlet boundary condition of u  uˆ is prescribed,

equation [2-4] is modified in a way that the boundary condition is naturally satisfied at Su .

uh  ΠNde  uˆ  ΠNde  Ndˆ e [2-5]

û : prescribed displacement

Π : solution structure modifier

where Π is a diagonal matrix of spatial functions that modifies solution structure. In

essence together with prescribed solution vector, this function enables the solution to satisfy
the Dirichlet boundary condition at nonconforming surface:
 1 X   Su
  X   [2-6]
0
 X  Su

Numerically, the special function is adopted such that the function value changes spatially
from one to zero near the vicinity of the Dirichlet boundary surface, Su .

Based on equations [2-4] and [2-5], the corresponding grid element strains can be
represented as,

εh   uh  Bde [2-7]

B : strain interpolation matrix

εh   uh   Π  Nde  ΠBde  Bdˆ e [2-8]

Principle of virtual work in equation [2-3] can be restated in terms of internal and external
force vector in the following form.

 Gext   dT Fext   dT FInt [2-9]

Employing strain-displacement relations expressed equations [2-7] and [2-8], the global
internal force vector is expressed in terms of summation of contributions from each grid
element as,

FInt    K C e   K 

D 
e d   F ˆ   F ˆ
C D
[2-10]
 C D d d

where each grid element stiffness matrices and force vectors are expressed in the following
form.

K Ce   BT DBd 
C C
e
[2-11]
BT DB  NT   Π  DB  BD   Π  N  NT  Π  D  Π  Nd 
T T
e 
D
K 
D eD

FdCˆ   BT DBd  dˆ e
C
e
[2-12]
NT  Π  DB  BT DBd  dˆ e
T
FdDˆ  
eD

It is worth noticing that integration of each grid element contributions is conducted on the
actual computation domain only (   Ce or   eD ). To this end, special integration
techniques are utilized (see section 2.4). For linear elasticity problems, the solution to the
equation [2-4] can be obtained by solving sparse system of linear equations.

The global solution vector is utilized to recover strain using equations [2-7] and [2-8].
Computation of stress follows trivially for elastic materials.

2.2 Other kinematic constraints

In finite element method, kinematic constraints are usually described in terms of nodal
variables and incorporated into the global system matrix as multi-point constraints. However,
in midas Meshfree, kinematic constraints are defined over geometries where nodes are not
present. Thus, special displacement fields that inherently satisfy the kinematic constraints
are devised and incorporated. Among the kinematic constraints available, rigid body and
welded contact constraints are introduced herein.

Rigid body
rotation

Rigid body
translation

us
Xs  Xm
R  θm 

um

Figure 2.2 Rigid body kinematics

Points in certain surface, X s can be constrained such that its displacements satisfies rigid
body motion with respect to an independent point in space, Xm . The geometric nonlinear
kinematic relations can be incorporated in the surface displacement in the following form.


 
uhs  ΠNdes   I  Π  um  R  θm   I  Xs  Xm 

[2-13]

um : translational displacement vector of master point, X m

θm , R  θm  : rotation vector and corresponding rotation matrix of

master point, X m
 R(X1  X2 )

u1

u2

X1  X2

Figure 2.3 Welded contact kinematics with finite rotation

Considering two surfaces separated by a distance constrained under welded contact

condition, with additional rotational motion considered, the geometric nonlinear kinematic
relations between two corresponding points ( X1 and X2 ) in contact surfaces are
represented as,

u1h  ΠNd1e   I  Π  u2h   R  I   X1  X2  [2-14]

 

u1h , uh2 : translational displacement vectors of contact points

X1 and X2
R : rotation matrix

Effect of finite rotation comes into play only if the points are separated by a finite distance. If
corresponding points are located at the identical spatial position, the above equation
reduces to a simpler and familiar form that satisfies u1h  uh2 at contact points.

u1h  ΠNd1e   I  Π  uh2 [2-15]

In equations [2-13] to [2-15], Π takes the form of equation [2-6], vanishing at the surface
where the special constraints are applied. Kinematic relations for sliding contact condition
can be similarly derived by releasing tangential components of displacement vector.

For linear structural analysis, linearized form of equations [2-13] and [2-14] is utilized to
derive strain and its variation which are included in the variational formulation.

2.4 Numerical integration

The key to the accuracy of numerical methods based on geometrically non-conforming grids
is the domain integration method where the domain is provided as arbitrary three-
dimensional CAD data. Robustness, accuracy as well as computational efficiency highly
depend on volume and surface integration techniques.

e e
 

(a) Recursive octree subdivision (b) Tetrahedral subdivision

Figure 2.4 Two-dimensional depiction of volumetric integration of CAD geometry

For volume integration, several methods are available in midas Meshfree:

1) Gauss quadrature method with recursive octree domain division.
Figure 2.4 (a) shows the grid element division concept in 2D where the domain is divided
along the domain boundary recursively. The recursion continues until integration error is
within acceptable tolerance.

2) Gauss quadrature method with domain division by tetrahedrons.

As shown in Figure 2.4 (b), the domain defined by intersection of computational domain and
of grid element domain   e is subdivided into tetrahedrons. These tetrahedrons are
used as base geometry for integration purpose only.

3) Proprietary method based on adaptive generation of integration points aided by ray-

tracing algorithm with automatic geometry repair. The adaptive procedure is driven by
minimizing integration error as well as reducing the overall number of integration points.

By default, volume integration is conducted using the third method. It can repair the flaws in
CAD models which in many cases exist. Other methods are utilized to make the whole
integration procedure fail-safe.

 ,

Ti
 Figure 2.5 Integration over CAD surface

With regards to surface integration method, the CAD surface is divided into triangular
patches, and integration is conducted using Gauss quadrature on triangles.

S fdA   T i 
f x  ,   x  x d d [2-16]

Ti : sub-triangle domain

, : parametric coordinates over sub-triangle

2.5 Hybrid formulation

Finite elements especially low-order types based on assumed displacement formulation

generally suffer from poor accuracy due to locking phenomenon. This fact extends to the
immersed grid method adopted in midas Meshfree. To enhance accuracy mixed hybrid
formulation is developed based on Hellinger-Reissner variational principle (see equation [2-
2]).

In addition to the displacement assumption in equations [2-4] and [2.5], stress is

independently assumed in the grid element level (ref 2.7):

σ  Pβe [2-17]

For interior grid elements and boundary elements with Neumann boundary condition, the
right hand side of equation [2-2] can be expressed as,


 deT Q1T βe   βeT Q1de  Pβe  [2-18]

For boundary grid element with Dirichlet boundary condition prescribed, the same terms
takes the following form:

 
 deT Q1T  QT2 βe   βeT Q1  Q2  de  Pβe 
 
[2-19]

where, Q1 , Q2 and P are expressed as,

Q1   PT Bd 
e

Q2   PT  ΠNd  [2-20]
eD

P PT D1Pd 
e

By eliminating stress interpolation coefficients, β e at grid element level, element stiffness

matrices corresponding to equations [2-18] and [2-19] can be respectively, expressed as
equations [2-21] and [2-22] as follows.

K e  Q1T P1Q1 [2-21]

 Ke  Q1T P1Q1  Q1T P1Q2  Q2T P1Q1  Q2T P1Q2 [2-22]

2.6 Error estimation

Error is introduced during the discretization process of the solution field into finite number of
grid elements. To assess the discretization error, a posteriori error estimation can be utilized.
To this end, strain energy error is calculated based on a stress error definition. The stress
error can represented as error between exact stress that satisfies the governing equation
and approximate grid stress which includes errors from discretization.

e  σexact  σh [2-23]

σexact : exact stress

σh : approximate grid element stress

Similar to the methods used in the finite element method community, stress obtained by a
separate recovery procedure is employed as an estimate to the exact stress to define stress
error.

e  σ R  σh [2-24]

σR : recovered stress

From strain energy expression,

1
2  Ce h
U σ T D1σ h d  [2-25]

strain energy error and non-dimensional strain energy error are computed in the following
form:

1
2  Ce 
eU  eT D1e d 
[2-26]
e
eU  U U

Meshfree provides error measures as well as strain energy for individual part in an assembly,
to assess discretization error. The information can be useful for adjusting overall grid size or
part level grid size. Table 2.1 summarizes the output information generated to assess
discretization error.

Table 2.1 Output information to assess discretization error

Output item Equation
Average strain energy density for part P: 1
2 e Ce σ h D σ hd 
T 1

UP  P
VP
Subscript P indicate a specific part, V P is the
volume of part.
Average energy error density for part P: 1
2 e Ce e D e d 
T 1

eU , P  P
VP
Part strain energy ratio UP
100%
U
Part energy error ratio eU , P
100%
eU
Total strain energy U
Total energy error eU
Non-dimensional error energy eU

2.7 Heat transfer analysis using immersed grid elements

The heat transfer formulation is based on a variational form of the energy balance equation
with Fourier law of heat conduction.

T
  c t  T   T 
T
κTd     TqdS    Trd  [2-27]
Sq 

T : temperature
 : material density
c : material capacitance
κ : material conductivity 3x3 matrix
q : surface heat flux
r : volumetric heat generation
 : gradient operator

Loading and boundary conditions include prescribed temperature, heat generation on

volume, surface flux, surface convection and radiation.


q  qˆ  hc T  Tamb   Ar T 4  Tamb
4
 [2-28]

q̂ : specified heat flux

hc : surface heat convection coefficient
Ar : surface radiation coefficient
Tamb : ambient temperature
 q  x   qˆ  hc T  Tamb   Ar T 4  Tamb
4 

   
Sq

eD
 T  x  Tˆ  x

Interior grid element

Boundary grid element (Neumann)

Su
Ce
Boundary grid element (Dirichlet)

Figure 2.4 Heat transfer grid elements with immersed computational domain

The concepts used to obtain discretized form of equations for structural analysis in section
2.2 is adopted in a similar manner. For interior grid elements and boundary grid elements
with Neumann boundary condition, temperature, T is interpolated by shape functions and
associated nodal degrees of freedom values.

T h  Nhe [2-29]

he : element temperature degrees of freedom vector

For boundary grid elements where boundary condition of T  Tˆ is prescribed, equation [2-29]
is modified in a way that the boundary condition is naturally satisfied at Su .

T h   Nhe  Tˆ   Nhe  Nhˆ e [2-30]

Tˆ : prescribed temperature

For transient heat transfer analysis, backward difference algorithm is adopted. The final
integral equation is expressed as:

 ht 



c e
C
 ce 
D
 ht t
 t
  k Ce  k 
 C
D 
e  ht t   f q  f r  f ˆ
D  T   [2-31]

where the grid element contributions are represented as follows (equations [2-32] and [2-33]).

ce    cNT Nd 
e

 N 
T
k Ce   κNd  [2-32]
C C
e

 N  κN  NT    κN   N κ    N  NT    κ    Nd 

T T T T
e 
D
k 
C C
e
 fq  
Sq e    
NT qˆ  hc Ttht  Tamb  Ar Ttht 4  Tamb
4
dS

fr   NT rdS [2-33]
e

hˆ e  hˆ te  D
fTˆ  ce  t t k ˆe
eD ht t
D  t 
 

If material property is dependent on temperature or in the presence of radiation, the equation

is nonlinear. The nonlinear equation is solved using Newton-Raphson method, where
linearized residual equation is iteratively solved until convergence.

2.7 Reference

2.1 A. Duster, J Parvizian, Z Yang, and E Rank, “The finite cell method for three-dimensional
problems of solid mechanics”, The finite cell method for three-dimensional problems of solid
mechanics, Vo. 197, Issues 45-48, 2008

2.2 R Mittal and G Iaccarino, “Immersed boundary methods”, Annual Review of Fluid
Mechanics, Vo. 37, 2005

2.3 Hu, H.C., “On some variational principles in the theory of elasticity and the theory of
plasticity,” Scintia Sinica, Vol. 4, 1955

2.4 Washizu, K., On the Variational Principles of Elasticity, Aeroelastic and Structural
Research Laboratory, MIT, Technical Report, 1955

2.5 Hellinger, E., “Der allgemeine Ansatz der Mechanik der Kontinua,” Encyclopadie der
Mathernafischen Wissenschaften, Vol. 4, 1914

2.6 Reissner, E., “On a variational theorem in elasticity,” Journal of Mathematical Physics,
Vol. 29, 1950

2.7 Pian, T.H.H. and Sumihara, K., “Rational approach for assumed stress finite elements,”
International Journal for Numerical Methods in Engineering, Vol. 20, 1984
3. Analysis algorithm

3.1 Sparse matrix solver

3.2 Eigenvalue extraction
3.2 Mode superposition
3.3 Dynamic response
3.3.1 Implicit time integration
3.3.2 Frequency response
3.3.3 Random response
3.3.4 Response spectrum

3.4 Nonlinear solution

3.5 Large deformation
3.6 Nonlinear contact condition
3.7 Fatigue
3.7 Topology optimization?
4. Mechanical constitutive relations
4.1 Elastic materials
4.2 Elastoplastic materials
4.3 Hyperelastic materials (--)