Showing results for 
Search instead for 
Do you mean 
Reply

SYSTEM FATAL MESSAGE 7340 (LNNHERR)


I have a relatively large model 200000 elements which gives the following error after
8 hours of running.
Does any one know how I can track down the cause of this error?
*** SYSTEM FATAL MESSAGE 7340 (LNNHERR)
PROCESS ERROR REPORTED BY SUBROUTINE LNNP1ED (IER= -729)
USER INFORMATION: REPEATED SINGULAR MATRIX -- PROBABLY ILL-POSED PROBLEM.
*** SYSTEM FATAL MESSAGE 7340 (LNNHERR)
PROCESS ERROR REPORTED BY SUBROUTINE LNNDRVD (IER= -729)
USER INFORMATION: REPEATED SINGULAR MATRIX -- PROBABLY ILL-POSED PROBLEM.

TABLE OF SHIFTS: (LNNRIGL)
SHIFT # SHIFT VALUE FREQUENCY, CYCLES # EIGENVALUES BELOW
# NEW EIGENVALUES FOUND
1. 3.9478416E+01 9.9999994E-01 FACTOR ERROR
0
2. 2.1262569E+02 2.3207481E+00 FACTOR ERROR
0
3. 1.9440984E+03 7.0174484E+00 FACTOR ERROR
0

*** SYSTEM FATAL MESSAGE 7340 (LNNRIGL)
REPEATED SINGULAR MATRIX -- PROBABLY ILL-POSED PROBLEM.
1 MODAL FREE-FREE JUNE
24, 2009 NX NASTRAN 1/17/09 PAGE 12411
Thanks in advance,
Carlton.
2 REPLIES

Re: SYSTEM FATAL MESSAGE 7340 (LNNHERR)


Hi Carlton,
if you already performed all model check I suggest you divide your model in some
groups (sub systems) and perform the free free analysis on these groups separately.
Then using an exclusion process you will find out where the problem is. As for the
cause I happen to have the same problems and it was a rotation axis (modelled with
RBE and aligned nodes) which was not perfectly aligned 0.01mm error. But it could
be anything (wrong material props, CELAS on not perfectly coincident grids etc..)
Hope this helps,
regards
Francesco




"Carlton Matthew" wrote:
>
>I have a relatively large model 200000 elements which gives the following
>error after
>8 hours of running.
>Does any one know how I can track down the cause of this error?
>
>*** SYSTEM FATAL MESSAGE 7340 (LNNHERR)
>
> PROCESS ERROR REPORTED BY SUBROUTINE LNNP1ED (IER= -729)
>
> USER INFORMATION: REPEATED SINGULAR MATRIX -- PROBABLY ILL-POSED PROBLEM.
>
> *** SYSTEM FATAL MESSAGE 7340 (LNNHERR)
>
> PROCESS ERROR REPORTED BY SUBROUTINE LNNDRVD (IER= -729)
>
> USER INFORMATION: REPEATED SINGULAR MATRIX -- PROBABLY ILL-POSED PROBLEM.
>
>
>
> TABLE OF SHIFTS: (LNNRIGL)
>
> SHIFT # SHIFT VALUE FREQUENCY, CYCLES # EIGENVALUES BELOW
>
> # NEW EIGENVALUES FOUND
>
> 1. 3.9478416E+01 9.9999994E-01 FACTOR ERROR
>
> 0
>
> 2. 2.1262569E+02 2.3207481E+00 FACTOR ERROR
>
> 0
>
> 3. 1.9440984E+03 7.0174484E+00 FACTOR ERROR
>
> 0
>
>
>
> *** SYSTEM FATAL MESSAGE 7340 (LNNRIGL)
>
> REPEATED SINGULAR MATRIX -- PROBABLY ILL-POSED PROBLEM.
>
>1 MODAL FREE-FREE
> JUNE
> 24, 2009 NX NASTRAN 1/17/09 PAGE 12411
>Thanks in advance,
>Carlton.

Re: SYSTEM FATAL MESSAGE 7340 (LNNHERR)

Dear Carlton,
According the error message you have a "singular matrix" or/and "ill-posed problem". I understand you are trying to compute the eigenvalues and mode shapes of a free-free model, then you will have to revise your model to see if all material properties are defined properly, also geometric properties (depending of element types used: shell thickness, bar/beam area & moment of inertia, etc..).
If the model is ill-conditioned the reasons could be many: for instance, you have slender beams the ratio of the bending terms to the stretching terms (h/l, slenderness) maybe as small as 1/20 or 1/50, so the stiffness matrix may possibly ill-conditioning. Or you have large changes in stiffness across element boundaries. For instance, the volume of one element should not be more than two to four times the volume of any immediately adjoining element.
Problems often arise in the modelling idealisation where physical discontinuities in the structure occurs, typically in the modelling of joints and other connections. If some DOF have to be interconnected between beams & shells to form the joint then this is best done by means of multi-point constraints to joint the freedoms together in some specified pattern. Some elements have rigid offset facilities built into their formulation and can be used to connect the neutral axis of beam elements with the centre line of a plate (in NX Nastran we have RBE2 & RBE3 elements that are really powerful), but somebody use fictitious very stiff elements. This is a very dangerous practice that should be avoided wherever possible because if the property value used to define the fictitious element is not high enough the constraints will not be satisfied, but if they are too high then the resulting equations will be ill-conditioning, and error could arise in the solution process.
Some care has to be taken when using the Lanczos method if the structure is not supported in some directions (it is free-free) or if it contains any mechanism. In such cases the stiffness matrix will be singular and any solution method that involves solving a set of equations defined by the stiffness matrix will be very probe to ill-conditioning. The user has little control over this situation and if un unsupported structure is to be analysed it is suggested that the eigenvalue extraction process that is to be used is tested first on a small problem where the stiffness matrix is not supported. There are two main strategies that are employed to overcome this problem: the simplest is the shift technique used in the Power method. The main disadvantage is the choise of the shift value: if too small then the equations will still be very ill-conditioned, if too large then the convergence to the eigenvalues and vectors will be very slow. The shift value should be of the order of one tenth to one half of the value of the first natural frequency.
In summary, check the quality of your model, this is the key. As a double-check, I will run the model using the Sturm modified inverse power method, it can be the best choice when the model is too large to fit into memory, only a few modes are needed, and a reasonable eigenvalue search range is specified. This method is also a backup method for the Lanczos method and is used when a check of the results is needed.
Also, in the NX Nastran Basic Dynamic Analysis User's Guide you have an overview ot the treatment of Rigid Bodies in Normal Modes Analysis:
Theoretical Considerations
------------------------------------------------------------------------------
Degrees-of-freedom defined on the SUPORT entry are placed in the r-set. When an r-set is present, static rigid-body vectors are calculated in NX Nastran by first partitioning the a-set into the r- and l-sets
../..
The rigid-body error ratio and the strain energy should be zero if a set of statically determinate SUPORT DOFs is chosen. Roundoff error may lead to computational zero values for these quantities. ("Computational zero" is a small number (10-5 , for example) that normally is 0.0 except for numerical roundoff.) The rigid-body error ratio and strain energy may be significantly nonzero for any of the following reasons:
a.. Roundoff error accumulation.
b.. The ur set is overdetermined, leading to redundant supports. The condition gives high strain energy.
c.. The ur set is underspecified, leading to a singular reduced-stiffness matrix and a MAXRATIO error. This condition gives a high rigid-body error ratio.
d.. The multipoint constraints are statically indeterminate. This condition gives high strain energy and a high rigid-body error ratio.
e.. There are too many single-point constraints. This condition gives high strain energy and a high rigid-body error ratio.
f.. is null. This condition gives a unit value for the rigid-body error but low strain energy (see "Advanced Dynamic Analysis Capabilities" ).
Hope it helps!!.
Best regards,
Blas.
--
~~~~~~~~~~~~~~~~~~~~~~
Blas Molero Hidalgo
Ingeniero Industrial
Director
IBERISA
Edificio Ercilla
Rodríguez Arias 23, 3º - Dpto. 19
48011 BILBAO (SPAIN)
Tel. (+34) 94 410 65 50
Fax. (+34) 94 470 26 34
E-mail: info@iberisa.com
WEB: http://www.iberisa.com
"Carlton Matthew" escribió en el mensaje news:4a436f41$1@bbsnotes.ugs.com...
>
> I have a relatively large model 200000 elements which gives the following error after
> 8 hours of running.
> Does any one know how I can track down the cause of this error?
>
> *** SYSTEM FATAL MESSAGE 7340 (LNNHERR)
>
> PROCESS ERROR REPORTED BY SUBROUTINE LNNP1ED (IER= -729)
>
> USER INFORMATION: REPEATED SINGULAR MATRIX -- PROBABLY ILL-POSED PROBLEM.
>
> *** SYSTEM FATAL MESSAGE 7340 (LNNHERR)
>
> PROCESS ERROR REPORTED BY SUBROUTINE LNNDRVD (IER= -729)
>
> USER INFORMATION: REPEATED SINGULAR MATRIX -- PROBABLY ILL-POSED PROBLEM.
>
>
>
> TABLE OF SHIFTS: (LNNRIGL)
>
> SHIFT # SHIFT VALUE FREQUENCY, CYCLES # EIGENVALUES BELOW
> # NEW EIGENVALUES FOUND
>
> 1. 3.9478416E+01 9.9999994E-01 FACTOR ERROR
> 0
>
> 2. 2.1262569E+02 2.3207481E+00 FACTOR ERROR
> 0
>
> 3. 1.9440984E+03 7.0174484E+00 FACTOR ERROR
> 0
>
>
>
> *** SYSTEM FATAL MESSAGE 7340 (LNNRIGL)
>
> REPEATED SINGULAR MATRIX -- PROBABLY ILL-POSED PROBLEM.
>
> 1 MODAL FREE-FREE JUNE
> 24, 2009 NX NASTRAN 1/17/09 PAGE 12411
> Thanks in advance,
> Carlton.