Nastran#

.bdf .nas .fem .dat read + write eager

Summary of the specification#

A Nastran bulk data file is a deck of fixed-column cards. Each card starts with a name in the first field and continues across eight further fields. Three field widths coexist: small field (8 columns), large field (16 columns, signalled by a * on the card name), and free field (comma-separated). GRID cards define points; CTRIA3, CQUAD4, CTETRA, CHEXA and friends define elements by referencing grid ids and a property id, and continuation lines carry any fields that overflow the first card.

Specification at a glance#

field widths	small (8 col), large (16 col, name*), free (comma-separated)
points	GRID id cp x1 x2 x3
elements	CTRIA3, CQUAD4, CTETRA, CHEXA, CPENTA, CBAR …
grouping	property id (PID) on each element card
continuation	/ * continuation markers on overflowing cards
other suffixes	.dat resolves by content (GRID / CEND / BEGIN BULK / SOL)

Read the full Nastran specification ↗

Reading#

import polyxios as px

mesh = px.read("model.bdf")
mesh.vertices          # (n, 3)
mesh.element_types     # element groups found in the file

Writing#

px.write(mesh, "out.bdf")

Format-specific options:

Option	Default	Effect
`large_field_grid`	`False`	Emit GRID cards in large (16-column) field form for full float precision.

Quirks worth knowing#

All three field widths are read; writing emits free field, with large-field GRID cards on request.
.dat is shared with Tecplot, LS-DYNA and plain ASCII tables, so it is resolved by looking inside the file: a $ comment banner is stepped over and a GRID card, CEND, BEGIN BULK, NASTRAN or a SOL naming a solution number or name lands the deck here - a numeric table whose first row happens to read SOL 1.0 2.0 does not. px.read("model.dat", fmt=".bdf") still forces the issue, and writing to .dat needs fmt=".bdf" because an output file has no content to inspect.
A real field is written in whichever legal spelling fits it: the plain form first, then the implicit-exponent shorthand the format allows (1.234-10 for 1.234E-10, 1.2346+7 for 1.2346E+07), which buys two or three significant digits in an eight-column field. A value below one drops its leading zero when that column is a significant digit, since bulk data reads .5 as 0.5. A spelling that reads back as the value itself is always preferred to a longer one that does not, so 1e7 goes out as 1.E+07 rather than as seven nines. A value that would round up past the largest double is stepped one digit toward zero instead of being refused, so no finite coordinate fails to write.
Property ids become element tags, so the deck’s grouping is preserved.
A card name does not say the element’s order, so the shape comes from the grid points the card carries: CTETRA with ten grid points is a quadratic_tetra, with four a tetra. Counting stops at the first blank field, because Nastran lets a mid-side grid point be omitted and that card is a linear element.
CPENTA runs the prism’s three vertical edges before its top ring where VTK runs them after, so those mid-side nodes are permuted on read and back on write. Every other card already agrees with VTK.
The axisymmetric cards read as their planar counterparts: CTRAX3, CTRAX6, CTRIAX6, CQUADX, CQUADX4 and CQUADX8, alongside CSHEAR.
One-dimensional cards (CBAR, CBEAM, CROD, CONROD, CTUBE, CBUSH, CGAP, CBEND, CVISC) read as lines rather than being skipped; CONROD names a material where the others name a property, so its grid points sit one field earlier.
A shell’s ZOFFS becomes element_attrs["zoffs"] when any card carries one, and is written back on the shell cards that have a field for it: CTRIA3, CTRIAR, CQUAD4, CQUADR, CTRIA6 and CQUAD8.
Nastran and VTK number the corner grid points alike, but not every mid-side one. CPENTA runs the prism’s vertical edges before its top ring and CHEXA does the same with the brick’s, where VTK runs the verticals last; CTRIAX6 interleaves corners and mid-side nodes rather than listing the corners first. All three are permuted on the way in and back on the way out.
CTRIAX6 names a material in the field where the shells name a property, so its second field is not read as a property id. A CBUSH grounding its second end names one grid point and no element, and is skipped with a warning of its own rather than being counted among the cards this codec has no shape for.