Hallo!
Ich habe derzeit folgendes Problem: Ich versuche eine Simulation von einem Temperaturfeld durchzuführen, wobei aus dem Bauteil immerwieder Elemente in einzelnen Steps gelöscht werden und das Bauteil lokal aufgewärmt wird. Insgesamt werden 80 kleine Elemente gelöscht.
Jedoch bricht mir dir Simulation nach bereits 3 gelöschten Elemten ab. Im msg-File steht, dass das Fixed Increment zu groß ist, jedoch habe ich es bereits halbiert und er kommt dennoch nicht weiter. Außerdem beschwert er sich immer wieder über Knoten aus bereits gelöschten Elementen und erhält somit Singularitäten. (siehe Auszug aus dem msg-File)
Ich vermute, dass das eigentliche Problem die Singularitäten sind und frage mich daher, warum prüft er überhaupt noch Knoten von gelöschten Elementen? Da die zu löschenden Elemente sehr kleine Quader sind (a=0.5mm) und der Rest des Bauteils relativ groß, konnte ich sie nur mit einem Netz belegen, das halb so groß ist, damit die Rechenzeit erträglich bleibt. Könnte das ein Problem sein? Hat da jemande Erfahrung mit?
Danke und Gruss,
Killroy
(Im Step-2 wurde das (erste) Element Span.1 gelöscht und in Step-4 soll das nächste gelöscht werden)
msg-File:
[...]
S T E P 4 T R A N S I E N T H E A T T R A N S F E R
FIXED TIME INCREMENTS
TIME INCREMENT IS 1.000E-04
TIME PERIOD IS 1.000E-04
CONVERGENCE TOLERANCE PARAMETERS FOR HEAT FLUX
CRITERION FOR RESIDUAL HEAT FLUX FOR A NONLINEAR PROBLEM 5.000E-03
CRITERION FOR TEMP. CORRECTION IN A NONLINEAR PROBLEM 1.000E-02
INIT. VALUE OF TIME AVG. HEAT FLUX IS TIME AVG. HEAT FLUX IN PREVIOUS STEP
AVERAGE HEAT FLUX IS TIME AVERAGE HEAT FLUX
ALTERNATE CRIT. FOR RESIDUAL HEAT FLUX FOR A NONLINEAR PROBLEM 2.000E-02
CRITERION FOR ZERO HEAT FLUX RELATIVE TO TIME AVRG. HEAT FLUX 1.000E-05
CRITERION FOR RESIDUAL HEAT FLUX WHEN THERE IS ZERO FLUX 1.000E-05
CRITERION FOR TEMP. CORRECTION WHEN THERE IS ZERO FLUX 1.000E-03
CRITERION FOR RESIDUAL HEAT FLUX FOR A LINEAR INCREMENT 1.000E-08
FIELD CONVERSION RATIO 1.00
CRITERION FOR ZERO HEAT FLUX REL. TO TIME AVRG. MAX. HEAT FLUX 1.000E-05
VOLUMETRIC STRAIN COMPATIBILITY TOLERANCE FOR HYBRID SOLIDS 1.000E-05
AXIAL STRAIN COMPATIBILITY TOLERANCE FOR HYBRID BEAMS 1.000E-05
TRANS. SHEAR STRAIN COMPATIBILITY TOLERANCE FOR HYBRID BEAMS 1.000E-05
SOFT CONTACT CONSTRAINT COMPATIBILITY TOLERANCE FOR P>P0 5.000E-03
SOFT CONTACT CONSTRAINT COMPATIBILITY TOLERANCE FOR P=0.0 0.100
DISPLACEMENT COMPATIBILITY TOLERANCE FOR DCOUP ELEMENTS 1.000E-05
ROTATION COMPATIBILITY TOLERANCE FOR DCOUP ELEMENTS 1.000E-05
ITERATION CONTROL PARAMETERS:
FIRST EQUILIBRIUM ITERATION FOR CONSECUTIVE DIVERGENCE CHECK 4
EQUILIBRIUM ITERATION AFTER WHICH ALTERNATE RESIDUAL IS USED 9
MAXIMUM EQUILIBRIUM ITERATIONS ALLOWED 16
MAXIMUM ITERATIONS FOR SEVERE DISCONTINUITIES 12
AUTOMATIC TOLERANCES FOR OVERCLOSURE AND SEPARATION
PRESSURE ARE SUPPRESSED
GLOBAL STABILIZATION CONTROL IS NOT USED
FRICTION IS INCLUDED IN INCREMENT THAT THE CONTACT POINT CLOSES
PRINT OF INCREMENT NUMBER, TIME, ETC., EVERY 1 INCREMENTS
THE MAXIMUM NUMBER OF INCREMENTS IN THIS STEP IS 1
LINEAR EXTRAPOLATION WILL BE USED
CHARACTERISTIC ELEMENT LENGTH 2.500E-04
DETAILED OUTPUT OF DIAGNOSTICS TO DATABASE REQUESTED
PRINT OF INCREMENT NUMBER, TIME, ETC., TO THE MESSAGE FILE EVERY 1 INCREMENTS
COLLECTING STEP CONSTRAINT INFORMATION FOR OVERCONSTRAINT CHECKS
INCREMENT 1 STARTS. ATTEMPT NUMBER 1, TIME INCREMENT 1.000E-04
USING THE DIRECT SOLVER WITH 2 PROCESSORS
***WARNING: SOLVER PROBLEM. NUMERICAL SINGULARITY WHEN PROCESSING NODE
SPAN-1.23 D.O.F. 11 RATIO = 1.21694E+32.
***WARNING: SOLVER PROBLEM. ZERO PIVOT WHEN PROCESSING NODE 23 INSTANCE
SPAN-1 D.O.F. 11
***NOTE: A ZERO PIVOT WARNING (IF THE *MODEL CHANGE OPTION WAS USED) MIGHT
INDICATE EITHER AN OVERCONSTRAINT OR THAT AN ELEMENT REMOVAL HAS
REMOVED ALL OF THE STIFFNESS ASSOCIATED WITH ONE OR MORE
DEGREES-OF-FREEDOM FROM AN ACTIVE NODE.
***NOTE: FOR EXAMPLE, IF A SHELL ELEMENT AND A BRICK ELEMENT SHARE A COMMON
NODE, AND THE SHELL ELEMENT IS REMOVED, THE BENDING STIFFNESS AT THE
NODE WOULD BE SET TO ZERO BUT THE NODE WOULD STILL BE ACTIVE SINCE
THE BRICK ELEMENT IS STILL PART OF THE MODEL. ONE POSSIBLE FIX IN
THIS CASE WOULD BE TO CONSTRAIN THE INACTIVE ROTATION
DEGREES-OF-FREEDOM USING A BOUNDARY CONDITION.
***NOTE: IF AN OVERCONSTRAINT OCCURS, THE CHAINS OF CONSTRAINTS CAUSING THE
OVERCONSTRAINT ARE OUTLINED NEXT.
OVERCONSTRAINT CHECKS:
An overconstraint was detected at node 23 INSTANCE SPAN-1. There are multiple constraints
applied directly or chained constraints that are applied indirectly at this node.
The following is a list of nodes and chained constraints between these nodes
that most likely lead to the detected overconstraint.
How to read this information:
(This explanation is printed once per analysis.
More details given in the OVERCONSTRAINT CHECKS section of the User's Manual.)
A sample output is shown below:
A (or LAGRANGE MULTIPLIER)->B,E: constraint 1
..B->C: constraint 2
....C->A: constraint 3
..E->F: constraint 4
....F->G,H: constraint 5
......G->A: constraint 6
......H->I: constraint 7
........I-> *BOUNDARY
Notation:
1. A particular line,e.g ....F -> G,H: constraint 5
issues information related to one particular constraint:
* node F is involved in constraint 5 together with nodes G and H.
* after the :, a brief constraint description (such as type and nodes involved) is output.
2. The left-most node (A or LAGRANGE MULTIPLIER in this case) is the (internal) node where the zero pivot was detected.
3. Indentation levels identify chains of constraints:
A->B,E: constraint 1
..B->C: constraint 2
....C->A: constraint 3 <- chain A->B->C->A ENDS HERE
..E->F: constraint 4
....F->G,H: constraint 5
......G->A: constraint 6 <- chain A->E->F->G->A ENDS HERE
......H->I: constraint 7
........I-> *BOUNDARY <- chain A->E->F->H->I->BOUNDARY ENDS HERE
4. A chain of constraints can either
(#) form a closed loop, e.g
chain A->B->C->A or,
chain A->E->F->G->A
(#) end up in a ground, such as a boundary condition, e.g
chain A->E->F->H->I->BOUNDARY
(#) end up in a free end (no constraints) when
it has no effect on the overconstraint.
For this particular overconstraint, the following chains of constraints
have been identified:
Please analyze these constraint loops (if printed) and remove unnecessary constraints.
If:
- no constraints are printed, or
- the printed constraints do not form closed loops, or
- the printed constraint chains do not end on *BOUNDARY nodes,
then the zero pivot node is most likely ***NOT*** overconstrained.
Since the ratio of the average stiffness associated with Lagrange multiplier enforced constraints
(such as contact, connectors, coupling, etc.) and the overall average stiffness in the model
is very small, the system of equations may be poorly conditioned and convergence may be slow.
In two-dimensional contact analyses, increasing the contact thickness may help.
***WARNING: Solver problem. Zero pivot when processing D.O.F. 11 of 1 nodes.
The nodes have been identified in node set
WarnNodeSolvProbZeroPiv_11_4_1_1_1.
EQUILIBRIUM ITERATION 1
AVERAGE HEAT FLUX 5.466E-03 TIME AVG. HEAT FLUX 5.466E-03
LARGEST RESIDUAL HEAT FLUX -3.878E-04 AT NODE 4068 DOF 11
INSTANCE: BASIS-1
LARGEST INCREMENT OF TEMP. -2.09 AT NODE 4227 DOF 11
INSTANCE: BASIS-1
LARGEST CORRECTION TO TEMP. -2.09 AT NODE 4227 DOF 11
INSTANCE: BASIS-1
HEAT FLUX EQUILIBRIUM NOT ACHIEVED WITHIN TOLERANCE.
USING THE DIRECT SOLVER WITH 2 PROCESSORS
***WARNING: SOLVER PROBLEM. NUMERICAL SINGULARITY WHEN PROCESSING NODE
SPAN-1.23 D.O.F. 11 RATIO = 1.21694E+32.
***WARNING: SOLVER PROBLEM. ZERO PIVOT WHEN PROCESSING NODE 23 INSTANCE
SPAN-1 D.O.F. 11
OVERCONSTRAINT CHECKS:
An overconstraint was detected at node 23 INSTANCE SPAN-1. There are multiple constraints
applied directly or chained constraints that are applied indirectly at this node.
The following is a list of nodes and chained constraints between these nodes
that most likely lead to the detected overconstraint.
Please analyze these constraint loops (if printed) and remove unnecessary constraints.
If:
- no constraints are printed, or
- the printed constraints do not form closed loops, or
- the printed constraint chains do not end on *BOUNDARY nodes,
then the zero pivot node is most likely ***NOT*** overconstrained.
Since the ratio of the average stiffness associated with Lagrange multiplier enforced constraints
(such as contact, connectors, coupling, etc.) and the overall average stiffness in the model
is very small, the system of equations may be poorly conditioned and convergence may be slow.
In two-dimensional contact analyses, increasing the contact thickness may help.
***WARNING: Solver problem. Zero pivot when processing D.O.F. 11 of 1 nodes.
The nodes have been identified in node set
WarnNodeSolvProbZeroPiv_11_4_1_2_1.
EQUILIBRIUM ITERATION 2
[... Step-7...]
***WARNING: Solver problem. Zero pivot when processing D.O.F. 11 of 2 nodes.
The nodes have been identified in node set
WarnNodeSolvProbZeroPiv_11_7_1_11_1.
EQUILIBRIUM ITERATION 11
AVERAGE HEAT FLUX 5.531E-03 TIME AVG. HEAT FLUX 5.531E-03
LARGEST RESIDUAL HEAT FLUX -2.396E-10 AT NODE 5 DOF 11
INSTANCE: SPAN-1-LIN-1-16
LARGEST INCREMENT OF TEMP. 2.360E+04 AT NODE 23 DOF 11
INSTANCE: SPAN-1-LIN-1-2
LARGEST CORRECTION TO TEMP. 7.936E+03 AT NODE 23 DOF 11
INSTANCE: SPAN-1-LIN-1-2
TEMP. CORRECTION TOO LARGE COMPARED TO TEMP. INCREMENT
***NOTE: THE SOLUTION APPEARS TO BE DIVERGING. CONVERGENCE IS JUDGED UNLIKELY.
***ERROR: FIXED TIME INCREMENT IS TOO LARGE
ANALYSIS SUMMARY:
TOTAL OF 21 INCREMENTS
0 CUTBACKS IN AUTOMATIC INCREMENTATION
96 ITERATIONS INCLUDING CONTACT ITERATIONS IF PRESENT
96 PASSES THROUGH THE EQUATION SOLVER OF WHICH
96 INVOLVE MATRIX DECOMPOSITION, INCLUDING
0 DECOMPOSITION(S) OF THE MASS MATRIX
1 REORDERING OF EQUATIONS TO MINIMIZE WAVEFRONT
0 ADDITIONAL RESIDUAL EVALUATIONS FOR LINE SEARCHES
0 ADDITIONAL OPERATOR EVALUATIONS FOR LINE SEARCHES
883 WARNING MESSAGES DURING USER INPUT PROCESSING
132 WARNING MESSAGES DURING ANALYSIS
88 ANALYSIS WARNINGS ARE NUMERICAL PROBLEM MESSAGES
13 ANALYSIS WARNINGS ARE NEGATIVE EIGENVALUE MESSAGES
1 ERROR MESSAGES
JOB TIME SUMMARY
USER TIME (SEC) = 12855.
SYSTEM TIME (SEC) = 1532.6
TOTAL CPU TIME (SEC) = 14388.
WALLCLOCK TIME (SEC) = 19038
-----------
Eine Antwort auf diesen Beitrag verfassen (mit Zitat/Zitat des Beitrags) IP