Sidebar
This is an old revision of the document!
Table of Contents
Appendix 20: Cycle jump procedure
The cycle jump procedure can be used to significantly lower computation times in cases of cyclic loading (typically for fatigue). Instead of performing a complete finite-element simulation over all the cycles, groups of cycles can be extrapolated to speed up calculations:
- a number of $N_i$ cycles is computed using the finite-element model;
- using the results from the last 2 computed cycles, the stresses, deformations, state-variables, reactions, etc. are exrapolated for the next $N_j$ cycles.
These 2 steps are then repeated until the calculation is completed. The number of cycles to be fully calculated $N_i$ and the number of “jumped” cycles $N_j$ can vary during the simulation. One can also pick the results to be extrapolated during the jump.
The cycle jump can be performed with Lagamine usign the Fortran EXTRAPOL program along with the attached Python script.
User guide
The cycle jump procedure can be used on any cyclic simulation. It is launched using the Python script and requires an additional info file.
The following tables describe the structure of the info file.
Note: With the exception of character strings, the file is not formatted. This means numbers don't need to be put in a specific number of columns, but simply separated by spaces.
1. General information
| Line 1 (A255) | |
|---|---|
| LAGname | Name of the *.lag file (without the extension) |
| Line 2 (A255) | |
| DATname | Name of the data *.dat file (without the extension) |
| Line 3 (integer) | |
| ISAVE | Units: = 1 - saving of the .F03 file after every full calculation of $N_i$ cycles = 2 - saving of all the .OTO file (NOT IMPLEMENTED) |
| Tens: = 1 - saving of the ipx (ipe, ipn, ipr, ipc, and/or ips) result file(s) defined on the next line. The results are concatenated in a file named LAGname_all.ipx. The 1st line of the file containing the columns title (see Appendix 9) is deleted. |
|
| If ISAVE≥10 - Line 3b (A255) | |
| List of the file extensions (ipe, ipn, ipr, ipc, and/or ips) the user wishes to record, separated by commas. | |
| Line 4 (2 integers) | |
| T | Period of the loading [s] |
| Tmax | Final/Maximum value of the time step (this corresponds to ALAMBF in a classic simulation) |
2. Definition of $N_i$ and $N_j$
| Line 1 | |
|---|---|
| Title | Anything can be written here, (for instance “Cycle definition”), this is just to make the file easier to read for the user |
| Line 2 | |
| IDENT | = 0 - $N_i$ and $N_j$ are kept constant over the whole simulation and defined at the next line = 1 - $N_i$ and $N_j$ are predefined by blocks. This allows for instance to have low values of $N_j$ at the begining of the simulation and higher values later once the beaviour of the structure is stabilized. |
| 10 ≤ IDENT < 1000 - User-defined calculation of $N_j$ and constant $N_i$. This must be implemented by the user in the EXTRAPOL program. 1000 ≤ IDENT < 2000 - User-defined calculation of $N_j$ and $N_i$ defined by blocks. The calculation of $N_j$ must be implemented by the user in the EXTRAPOL program. |
|
