forked from pvossos/felt
-
Notifications
You must be signed in to change notification settings - Fork 1
/
TODO
52 lines (49 loc) · 2.81 KB
/
TODO
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
Due to our own lack of experience/hardware/software, we definitely need
volunteers (we'd be willing to help where we could of course) for this stuff:
o some sort of velvet-like interface for Windows
o all of the above kinds of things (particularly just binaries of the
command-line applications) for the Mac.
Things that maybe we'll get to someday, but if you want to try them first
then please feel free. In no particular order:
o make use of the Layout widget for geometry management of the main GUI
o more intelligent use of the command line ... command completion
and argument specification ... type node hit space,
get a "list of options" type delete hit space, get a prompt
for a number, type number hit return, etc. In other
words make it something along the lines of some of the
popular CAD systems.
o improve plotting of displaced shapes. This probably means that
element writers need to provide a pointer to a function which
knows about the shape functions for an element and can fill in
a vector over the area or length of the element based on
nodal displacements which, when plotted, will reflect the true
displaced shape of the element ...
o add snap to tool capability
o robust quadrilateral mesh generation, 3-d mesh generation, etc, etc.
o additional higher-order elements -- particularly an LST and maybe
a quadratic 6-node axisymm element. The current linear axisymm
element could also be made more accurate given an actual
stiffness integration scheme. These last would all be fairly easy.
o fold Timoshenko beam theory into the standard beam, i.e., if
kappa or nu is given then expect a G and use Timoshenko
theory instead of Euler-Bernoulli.
o put some real effort into internationalization. The current
functionality is there because it was easy. More robust
implementation would mean maintaining a table, in code probably,
that mapped all possible input and as much output as possible
to a language determined by the user.
o plug in something more universal than PPM to replace GIF -
PNG probably?
o fix the object apply code in velvet so that when an object
gets applied to a node or element its color changes as appropriate
(rather than waiting for a recolor request from the color
control dialog).
o an OpenGL 3D stress viewer
o consolidation of much of the isoparametric code: brick, CST, LST,
quad, axisymm.
o uniform stresses, nodally averaged stresses as appropriate,
principle stresses, von Mises stress
o better, more uniform handling of distributed loads
o better handling of thermal analysis for the two elements that
we currently have, convection over the ctg surface for example,
rather than just the end