JL5
Aerospace
- Jul 18, 2009
- 1
Hi, I'm trying to extract mass and stiffness matrices from a .full file using the HBMAT command but I'm either doing something wrong in ANSYS or I've misunderstood the Harwell-Boeing format. The problem is that I'm getting zeros in the row index vector of the Harwell-Boeing output. Does anybody know under what circumstances this can happen? Details below:
I'm starting by performing a modal analysis on the model with a Block Lanczos solver to create the .full file. Next, I'm running HBMAT from the List->Files->Binary Files menu, to produce ASCII files for the mass and stiffness matrices. I'm intending to import these into MATLAB but I haven't got that far yet.
The files created with HBMAT are supposed to be in the Harwell-Boeing format, and as I understand the format I should see three consecutive vectors in the main data block: the column pointers, the row indices, and the data values. I haven't included the RHS vectors so this should be the only data. The problem is that in the mass matrix, some of the row indices are zero, but the indices are supposed to be 1-based so zeros should be impossible.
The model is comprised of BEAM4, BEAM44, SHELL43 and MASS21 elements.
I would be really grateful for any pointers, I've been going round in circles with this for a while now!
Cheers,
Jon
I'm starting by performing a modal analysis on the model with a Block Lanczos solver to create the .full file. Next, I'm running HBMAT from the List->Files->Binary Files menu, to produce ASCII files for the mass and stiffness matrices. I'm intending to import these into MATLAB but I haven't got that far yet.
The files created with HBMAT are supposed to be in the Harwell-Boeing format, and as I understand the format I should see three consecutive vectors in the main data block: the column pointers, the row indices, and the data values. I haven't included the RHS vectors so this should be the only data. The problem is that in the mass matrix, some of the row indices are zero, but the indices are supposed to be 1-based so zeros should be impossible.
The model is comprised of BEAM4, BEAM44, SHELL43 and MASS21 elements.
I would be really grateful for any pointers, I've been going round in circles with this for a while now!
Cheers,
Jon