0% found this document useful (0 votes)
109 views

New Trinitas in A Nutshell

TRINITAS is a Graphical Environment for Conceptual Design, Optimization and Finite Element Analysis. Major Objectives to obtain an overall speed-up of the entire engineering-cycle to increase control to reduce sources for errors.

Uploaded by

gorot1
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
109 views

New Trinitas in A Nutshell

TRINITAS is a Graphical Environment for Conceptual Design, Optimization and Finite Element Analysis. Major Objectives to obtain an overall speed-up of the entire engineering-cycle to increase control to reduce sources for errors.

Uploaded by

gorot1
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 29

TRINITAS

a Graphical Environment for Conceptual Design, Optimization and Finite Element Analysis

Major Objectives
to obtain

an overall speed-up of the entire engineering-cycle


to increase

control
to reduce

sources for errors


4/4/2013 TRINITAS Research & Developement 2

Scope

Geometry, Domain property and Boundary

condition modeling

Mesh generation Analysis Evaluation Optimization


TRINITAS Research & Developement 3

4/4/2013

Geometry modeling

Points

Surfaces
3-, 4- and N - edged general 3-D faces

Lines
Straight Line Parabola Cubic Bezier Circular Arc

Volumes
2D, 3D and Axisymmetric volumes as 4 -, 5 -, 6 - or N faced regions

4/4/2013

TRINITAS Research & Developement

Boundary Conditions
Essential Boundary Conditions
Fixed and Prescribed Displacement or Temperatures

Natural Boundary Conditions


General volume, surface, line and point loads Pressures

Contact Interfaces in 2D and 3D


4/4/2013 TRINITAS Research & Developement 5

Boundary Conditions (continued)


All boundary conditions can be made dependent of both space, time and temperature

Defined as symbolic functions


Or by User-subroutines

4/4/2013

TRINITAS Research & Developement

Element library
Bar and Beam elements in 2D and 3D 2D, Axi-sym. and 3D Solid elements from 3 to 27

nodes
Mindlin-Reissner 3D 3- to 9-node Shell elements Full or selective Gauss integration technique

Iso- or Orthotropic material behavior

4/4/2013

TRINITAS Research & Developement

Mesh Generation
Mapped and free meshed sub-domains
Advancing Front Technique capable of

adaptive analysis (only 2D)


A fast topology based bandwidth minization algorithm
4/4/2013 TRINITAS Research & Developement 8

Constraints between:
(master and slave)

Beam to Shells Beam to Solids

Solid to Bar Solid to Beam

Shell to Shells
Shell to Solids Solid to Solids Rigid Links

Solid to Shell
Shell to Beam Shell to Bar Groups of nodes

4/4/2013

TRINITAS Research & Developement

Analysis Overview
Linear static heat

transfer analysis
Linear buckling analysis Linear static elasticity analysis including

Dynamic eigenvalue analysis Transient heat transfer analysis

Transient linear elasticity analysis

optimization
4/4/2013 TRINITAS Research & Developement 10

Linear static elasticity analysis


Adaptivity (only 2D currently)

Contact Mechanics
Fracture Mechanics Quasi-static Load Sequences Topology, Shape & Size Optimization MPI Conjugate gradient solver on the element

level is available
4/4/2013 TRINITAS Research & Developement 11

Adaptivity (2D)

H-refinement technique

Local Super-convergent Patch Recovery Technique


3- or 6-node plane or axi-symmetric triangular elements are available A User-chosen relative energy error tolerance level have to be defined
TRINITAS Research & Developement 12

4/4/2013

Contact Mechanics
Automatic contact interface generation along shared lines and surfaces in the geometry a Gap function defines overlap or clearance as function of space and time Node to Node contact without friction Lagrangian solution or direct transformation

technique
4/4/2013 TRINITAS Research & Developement 13

Fracture Mechanics
Virtual crack extension technique Automatic crack growth direction calculation
(currently only in 2D)

Automatic integration of Paris law


(currently only in 2D)

Crack Closure
Semi-manual 3D crack growth analysis can been carried out
4/4/2013 TRINITAS Research & Developement 14

Topology Optimization
for Bar, Beam, Shell and 2D, Axi-sym.
and 3D Solids

Optimality criteria method


Filtering techniques

MPI-implementation under testing


4/4/2013 TRINITAS Research & Developement 15

Shape Optimization
Min Weight, Max Stiffness or MinMax stress

C1-continuity in-between Bezier splines


Analytical Brookeman derivatives Moving Asymptotes (MMA) is utilized for solving non-linear optimization problems in the design variables
4/4/2013 TRINITAS Research & Developement 16

Linear static heat transfer analysis


Fixed flux Fixed temperature Convective boundaries

4/4/2013

TRINITAS Research & Developement

17

Dynamic eigenvalue analysis


Sub-space iterations* Generalized Jacobi-method for the sub-problem

Lumped or consistent mass


Stress stiffening or not Automatic shifting of the eigenvalues
*(See Thomas J.R. Hughes The Finite Element Method )

4/4/2013

TRINITAS Research & Developement

18

Linear transient heat transfer analysis


Generalized Trapezoidal rule*
Backward or forward Euler Unconditionally or conditionally stable Automatic time step control
*(See Thomas J.R. Hughes The Finite Element Method )

4/4/2013

TRINITAS Research & Developement

19

Linear transient elasticity analysis


Generalized Newmark algorithm* Implicit and explicit Lumped or consistent mass Rayleigh damping
*(See Thomas J.R. Hughes The Finite Element Method )

4/4/2013

TRINITAS Research & Developement

20

Linear Buckling analysis


Sub-space iterations The Algorithm is shared by the Dynamic eigenvalue analysis

4/4/2013

TRINITAS Research & Developement

21

Non-linear static stress and strain analysis


Line search with automatic restart
A global non-linearized spherical arc-length control Automatic load increment control

4/4/2013

TRINITAS Research & Developement

22

Evaluation
Dynamic visualization of single numeric scalar- or vector node- or element values General tools for scalar- and vector-field

visualization
2D-graph representations of functions both in time or space. Dynamic or Static storage
4/4/2013 TRINITAS Research & Developement 23

Evaluation (continued)
Arbitrary number of independent cameras visualizing different result entities PostScript images of all camera views by a

single command
Animation

4/4/2013

TRINITAS Research & Developement

24

Implementation Details
The code consists of 8.6 Mb source code files with 4612
subroutines and 1252 functions written in FORTRAN 90 Currently used development environment is Intel Visual

Fortran
The program runs under Windows, Linux and Beowulf clusters under MPI OpenGL is utilized for 3D rendering Object-Oriented domain decomposition
4/4/2013 TRINITAS Research & Developement 25

Object Oriented Decomposition


Hierarchy: Model Interface Class Interface Matrix Interface Array Interface Data base Object Interface
Model Interface Class Interface Matrix Interface Create_a_New_Project Set_Current_Project Get_Class_Name(Object_Name, Array Interface Class Examples: Delete_Current_Project Set_Class_Name(Object_Name, Activities: Activities: Delete_Class_Name(Object_Name, Point Create, Get, Set, Exist, Create, Set, Get, Delete, Change_Name Exist_Class_Name(Object_Name) Line Exist, Delete Create_A_New_Model Surface Attributes: Set_Current_Model Support: Attributes: Object_Name Init_Current_Model Volume Get_New_Class_Object_Name Array Identifier Number_of_Rows Copy_Current_Model Internal Pointer Get_Number_of_Class_Instances ... Number_of_Columns Add_to_Current_Model Number of Bytes Matrix_Type FE-Classes Delete_Current_Model Matrix_Precision Internal pointers ... Temporary or Permanent Close_Current_Model
About 100 classes

Memory

Registers

4/4/2013

TRINITAS Research & Developement

26

Data base Object Interface


Registers Memory Internal pointer ~1 Gbyte

Registers
Logical Disk Array Virtual internal pointer Many Gbytes

Data Base File Memory Buffer

Data Base File Record Length Data Base File Record Length Number of Data Base File Records Number of Data Base File Records

Back Buffer

4/4/2013

TRINITAS Research & Developement

27

General Features
Finite element analysis in a What You See Is What You Get fashion No need for Node- or Element numbers in any situation Graphic control reduces the sources for errors to a minimum Errors is discovered as early as possible
4/4/2013 TRINITAS Research & Developement 28

Current Major Bottlenecks

3D geometry modeling

NURBS surfaces General 3D Boolean operations

An unstructured 3D mesh generator do not exist

4/4/2013

TRINITAS Research & Developement

29

You might also like