| blob | 4b660a6078fb016c17186666209faea44fc74aaa |
1 This is the README file for the alpha release of Felt - a package for
2 introductory finite element analysis.
4 FElt (Finite ELemenT) is a mathematical package intended for introductory
5 finite element work. It is not intended as a replacement for cutting
6 edge or real-world research code. At this stage FElt consists of
7 of three components. The first is the basic mathematical engine (the actual
8 felt command-line application) which includes an intelligent parser and
9 straightforward inputfile syntax. Because FElt is intended to be fairly simple
10 (at this stage of development at least) and because the authors' backgrounds
11 are largely in structural engineering, the element library is fairly small at
12 this point. Truss (bar), two- and three-dimensional beam, CST (plane stress
13 and plane strain) and iso-parametric (four to nine node) two-dimensional
14 elements are all currently supported. Adding additional elements to the
15 library is intended to be a straightforward exercise, if for example, such
16 a project were to be given as an exercise in an introductory finite element
17 course. As of this release, only linear static analysis is supported
18 (sorry ... linear dynamic analysis is in the works but we wanted to get a
19 version out the door and stabile before the fall term started).
21 In addition to the command line application version, there is xfelt,
22 an X11 based encapsulator for felt which provides a graphical text-editor
23 and two- and three-dimensional viewing of the structure. You can edit
24 the actual text file, solve the problem (xfelt actually execs the felt
25 application), graphically view the structure and save and print results
26 from within the xfelt gui.
28 Velvet is a full-featured X11 interface to the two-dimensional functionality
29 of felt and corduroy. Unlike xfelt, velvet is fully self-contained - it is
30 intended as the primary graphical user interface and along with the
31 mathematical engine is where most future development will occur. Within
32 velvet a user can interactively set-up, model, and solve a FElt problem
33 through a menu driven, CAD-like interface with no reliance on traditional
34 text based input files (as in felt, corduroy and even xfelt). Velvet does
35 not currently provide any real access to the three-dimensional aspects of
36 FElt.
38 Features:
40 o Easy to use. Intelligent parsers mean that there are no files
41 of comma delimited lists of numbers like you see in a lot of
42 other FE codes used in introductory courses. Velvet provides
43 a full-featured, interactive problem editor.
45 o Portable. Felt and corduroy should compile on any reasonable
46 Un*x system. There is also a DOS version of the felt program
47 only (including a simple utility for structural visualization,
48 drop one of us a note if you're interested). Velvet is built
49 around the public domain Athena widget set using
50 X11R5 ... R4 should work as well. Either regular or 3d Athena
51 widgets work (we prefer the 3d). The package was
52 developed on 386's running Linux; it has also been tested on 386's
53 running SysV R3.2, HP 700's, and Suns. We can provide binaries
54 for any of the above systems if there is an interest.
56 Caveats:
58 o Remember, this is meant largely as a teaching tool. In most
59 cases in the code we have sacrificed efficiency in favor of
60 readability. It is not our intention to be solving anyone's
61 cutting edge research problems ... the project was born out of
62 a fit of "we can do better than that" based on some antiquated
63 Fortran code we had been using.
65 o On that same note (and this is only a caveat to devoted
66 believers in the Fortran way of life) the entire package
67 is coded in C (with the exception of the Barry Joe's Geompk
68 routines ... they are in Fortran and we have not modified
69 them except where the interface is concerned). This really
70 shouldn't be a problem; we think the code is very readable and C's
71 data structures lend themselves quite readily to keeping
72 track of nodes and elements and forces and constraints and such.
73 Also, we can do dynamic memory allocation without a lot of
74 hocus pocus and have long variable names, so the code is
75 usually pretty easy to follow
77 The file felt.ps included in this distribution is a postscript document
78 providing a user's manual for the entire package, including mathematical
79 and algorithmic details and programming notes. Man pages are available
80 which briefly describe each individual application.
82 For installation instructions see the file INSTALL. The best way to get
83 a feel for the package is to use it. Several examples are provided
84 in the etc/Tests/ directory.
86 The FElt package is free software. This means that you can redistribute it
87 and/or modify it under the terms of the GNU General Public License, a copy of
88 which is provided in the file COPYING. (Excluding the Geompk routines which
89 are covered under their own terms, see the README file in lib/Generate/Geompk
90 for details.) Given the way in which it was developed, we wouldn't
91 want it any other way ... Linux is a Unix-like OS, based on a free kernel,
92 free development tools (gcc) and hundreds of free utilities (GNU, etc.).
93 (And not only is it free, but it runs circles around most commercial Un*ces
94 available for the 386). The entire graphical environment (X11 and the
95 Athena widget set) is also free. What a concept ...
97 Lastly, we would like to stress that feedback would be greatly appreciated ...
98 this is an alpha release and we welcome all comments as to usability and
99 functionality. Mail bug reports, comments, hints, suggestion, complaints,
100 etc. to one of us at the email addresses below. Our one wish in life is
101 that people find the package useful (well, maybe not, but you get
102 the point ...)
104 Jason Gobat Darren Atkinson
105 jgobat@ucsd.edu atkinson@ucsd.edu
107 -------------------------------------------------------------------------------
109 This is the README file for version 1.3 of FElt - a package for
110 introductory level finite element analysis.
112 See the file README.alpha for a basic description of the package. See
113 the file CHANGELOG for a detailed description of what the changes are
114 from the public release of 1.0a and this release of 1.3. This file will
115 simply bring you up to speed on a few things.
117 The most visible changes are to the velvet GUI. Things should be a lot
118 easier to use now. Most of the dialogs have keyboard interfaces (tab,
119 space, return keys etc.) which is nice and you can keep them up while
120 you work, which is very nice. There is also a real file selection
121 mechanism. For a nice gee whiz, from FElt/bin/Velvet execute "velvet
122 +numbers ../../etc/Tests/wrench.flt", solve it and then select plot
123 stresses from the problem menu ... wait a while and see the new color
124 stress output (assuming you have an 8-bit screen of course). The other
125 post-processing addition is basic displaced structure plotting but this
126 still needs some work. Other important, but not so visible, changes
127 mostly relate to distributed loads on beams and beam3ds and several bug
128 fixes to beam3d elements (they might even produce the right answers
129 now!) and constraints (displacement boundary conditions and hinges).
130 Some of the changes to distributed loads are not backward compatible.
131 If you have an input file from 1.0a that used direction=x y or z, you'll
132 have to change it to work with the new version. Sorry. The only new
133 element is a beam element that uses Timoshenko beam theory, but it is
134 pretty much just for instructional purposes at this stage, see Chapter 9
135 of the User's Guide and Reference manual on how to add elements to see
136 what we mean.
138 The file felt.ps ("FElt User's Guide and Reference Manual") in this
139 distribution bears little resemblance to the document in the alpha
140 release. The manual has tripled in size and includes much better
141 discussions of elements, algorithms, adding elements and using velvet ...
142 lots of cool new figures even. Most importantly, we did this one in
143 LaTeX and produced postscript with dvips so everyone should be able to
144 print it :-). Updated man pages are also available which briefly describe
145 each individual application.
147 For installation instructions see the file INSTALL. The best way to get
148 a feel for the package is to use it. Several examples are provided
149 in the etc/Tests/ directory.
151 Finally, as you are probably aware, FElt has found a permanent home via
152 anonymous ftp at cs.ucsd.edu in /pub/felt. Given this, we are going to
153 go to a system of smaller releases (e.g., 1.31, 1.32). In all
154 probability we will not announce each such revision very broadly. If
155 you want to be made aware of such updates, please let us know and we
156 will put your name on a mailing list whereby we will notify you every
157 time a new release becomes available. The latest version (including bug
158 fix "patchlevel" changes) will always be available on cs.ucsd.edu. We
159 will only be uploading major new versions (this one, the one that
160 finally has dynamic analysis in it, etc.) to sunsite.unc.edu.
162 As usual, we would like to stress that feedback is always appreciated ...
163 we welcome all comments as to usability and functionality. If you're using
164 the package please drop us a note and let us know what you think, even if
165 you've got no complaints whatsoever. Mail bug reports, comments, hints,
166 suggestion, complaints, etc. to one of us at the email addresses below.
168 Jason Gobat Darren Atkinson
169 jgobat@ucsd.edu atkinson@ucsd.edu
171 -------------------------------------------------------------------------------
173 This is the README file for version 2.0 of FElt - a package for
174 introductory level finite element analysis.
176 See the file README.alpha for a basic description of the package. See the
177 file CHANGELOG for a detailed description of what the changes are from the
178 public release of 1.3 to this release. This file will simply bring you up to
179 speed on a few things.
181 Major changes include the incorporation of transient analysis. In general
182 this means you can solve a structural dynamics problem using the Hilbert-
183 Hughes-Taylor alpha method of numerical integration (a variant of
184 Newmark integration). You can specify dynamic nodal forces with a continuous
185 function in time and they will be dynamically evaluated throughout the
186 course of the simulation. Velvet includes an animation capability to model
187 the dynamic behavior of the complete structure given this new analysis
188 capability. Old fashioned tabular and ascii based time-displacement plots
189 are also available in both felt and velvet. Gibbs-Poole-Stockmeyer/Gibbs-King
190 automatic node renumbering is also now available for badnwidth/profile
191 reduction. The matrix assembling routines now assemble directly into
192 a compact column representation. This is in itself can save major amounts
193 of memory. Coupled with renumbering, savings in both speed and memory can
194 be dramatic.
196 To keep people better informed a mailing list has been setup. We'll try to
197 make bug reports and fixes, announcements, etc. available through this
198 channel. Hopefully, it will also turn into a forum for FElt discussions in
199 general - including usage questions, how do I do xyz questions and discussions
200 of what capabilities people would like to see in the package. If you
201 are using FElt we urge you to subscribe if for no other reason than to keep
202 track of bug fixes.
204 To subscribe send a message to listserv@mecheng.fullfeed.com with the line
206 subscribe felt-l
208 as the only line in the _body_ of the message (the server ignores subject
209 headings). To send a question or comment to the list just set it up as a
210 regular email and send it to felt-l@mecheng.fullfeed.com. Thanks to
211 Greg Jackson at the ASME BBS for setting this up for us.
213 felt.ps ("FElt User's Guide and Reference Manual") has been updated to
214 reflect the new changes - as always look there for the real details. Updated
215 man pages are also available which briefly describe each individual
216 application.
218 For installation instructions see the file INSTALL. The best way to get a
219 feel for the package is to use it. Several examples are provided in the
220 etc/Tests/ directory. A simple dynamic example can be found in
221 etc/Tests/dynframe.flt.
223 As usual, we would like to stress that feedback is always appreciated ... we
224 welcome all comments as to usability and functionality. If you're using the
225 package please drop us a note and let us know what you think, even if you've
226 got no complaints whatsoever. Mail bug reports, comments, hints, suggestion,
227 complaints, etc. to one of us at the email addresses below or even better.
229 Jason Gobat Darren Atkinson
230 jgobat@mit.edu atkinson@ucsd.edu
232 -------------------------------------------------------------------------------
234 This is the README file for version 2.30 of FElt - a package for
235 introductory level finite element analysis.
237 See the file README.old for a basic descriptions of previous versions of FElt
238 and what it's all about. See the file CHANGELOG for a detailed list of
239 what's new/fixed in this release. The list of frequently asked questions
240 (and their answers) is in FAQ.
242 Major changes include the incorporation of modal analysis and the addition
243 of a brick element (8 node solid). In general modal analysis means that
244 FElt can figure out the natural frequencies and mode shapes for an arbitrary
245 structure. Velvet can actually give you a graphical rendering of mode shapes.
246 Support for the brick is still pretty weak, but at least we've officially
247 crossed the line into solid elements now.
249 For those of you who don't know there is a mailing list for FElt. We try to
250 make bug reports and fixes, announcements, etc. available through this
251 channel. We'd also like for it to turn into a forum for FElt discussions in
252 general - including usage questions, how do I do xyz questions and discussions
253 of what capabilities people would like to see in the package. If you
254 are using FElt we urge you to subscribe if for no other reason than to keep
255 track of bug fixes.
257 To subscribe send a message to listserv@mecheng.asme.org with the line
259 subscribe felt-l
261 as the only line in the _body_ of the message (the server ignores subject
262 headings). To send a question or comment to the list just set it up as a
263 regular email and send it to felt-l@mecheng.asme.org. Thanks to
264 Greg Jackson at the ASME BBS for setting this up for us.
266 felt-2.30.ps.gz ("FElt User's Guide and Reference Manual") is the updated
267 documentation for FElt. It's up to 151 pages or so. As always look
268 there for the real details. Updated man pages are also available which briefly
269 describe each individual application.
271 For installation instructions see the file INSTALL. The best way to get a
272 feel for the package is to use it. Several examples are provided in the
273 etc/Tests/ directory. A simple modal example can be found in
274 etc/Tests/modal.flt and simple example using a brick element is in
275 etc/Tests/brick.flt.
277 As usual, we would like to stress that feedback is always appreciated ... we
278 welcome all comments as to usability and functionality. If you're using the
279 package please drop us a note and let us know what you think, even if you've
280 got no complaints whatsoever. Mail bug reports, comments, hints, suggestion,
281 complaints, etc. to one of us at the email addresses below or even better.
283 Jason Gobat Darren Atkinson
284 jgobat@mit.edu atkinson@ucsd.edu
286 -----------------------------------------------------------------------------
288 This is the README file for version 3.00 of FElt - a package for
289 introductory level finite element analysis.
291 See the file README.old for a basic descriptions of previous versions of FElt
292 and what it's all about. See the file CHANGELOG for a detailed list of
293 what's new/fixed in this release - there are a fairly large number of
294 substantial changes with this release. The list of frequently asked
295 questions (and their answers) is in FAQ.
297 The big jump from 2.30 to 3.00 is based on the incorporation of burlap in
298 this release. burlap is an entirely new appliation designed to combine
299 the ease of use of FElt with the power and flexibility of command-line
300 driven, scripting environments like Matlab. burlap allows the easy
301 incorporation and testing of new element types, new analysis types, and
302 new ways to use the existing analysis types. There are two brand new chapters
303 in the manual that describe it and we'd really like people to give it a try
304 and let us know what they think.
306 Based largely on the two new burlap chapters, the manual has grown to 220
307 something pages; part of the growth is also due to a new chapter near the
308 front that tries to give some of the mathematical basis for each of the
309 analysis types. We've also done some extensive proofreading for the first
310 time in several releases. The xfelt chapter is gone because xfelt is gone
311 from the source and binary distributions. You can still get it from old
312 ones if you really want it (and don't for some reason already have it).
314 Three new analysis types have been added: static and transient thermal
315 analysis and spectral analysis (transfer functions, spectral response).
316 Modal analysis has some new capabilities (like the ability to do actual
317 modal analysis as opposed to just calculating eigenvalues and vectors).
318 The only new element is a one-dimensional spring.
320 The configure process and Makefile scheme is all new. We're still sort
321 of fine tuning it so if you have any problems or complaints we'd really
322 like to hear about them.
324 For installation instructions see the file INSTALL. The best way to get a
325 feel for the package is to use it. Several examples are provided in the
326 etc/Tests/ directory. A simple modal example can be found in
327 etc/Tests/modal.flt and simple example using a brick element is in
328 etc/Tests/brick.flt.
330 As usual, we would like to stress that feedback is always appreciated ... we
331 welcome all comments as to usability and functionality. If you're using the
332 package please drop us a note and let us know what you think, even if you've
333 got no complaints whatsoever. Mail bug reports, comments, hints, suggestion,
334 complaints, etc. to one of us at the email addresses below.
336 Jason Gobat Darren Atkinson
337 jgobat@mit.edu atkinson@ucsd.edu
339 -----------------------------------------------------------------------------
341 This is the README file for version 3.02 of FElt - a package for
342 introductory level finite element analysis.
344 See the file README.old for a basic descriptions of previous versions of FElt
345 and what it's all about. See the file CHANGELOG for a detailed list of
346 what's new/fixed in this release. This is basically an interim bug
347 fix release - there are lots of new features but for the most part
348 they are only partially implemented ... The list of frequently asked
349 questions (and their answers) is in FAQ (which has not been updated
350 for this release). The documentation (felt-3.02.ps.gz) has also
351 only been marginally updated. It now includes an (incomplete)
352 chapter describing the operation of WinFElt.
354 For installation instructions see the file INSTALL. The best way to get a
355 feel for the package is to use it. Several examples are provided in the
356 examples/ directory. Nonlinear, loadcase and load range examples
357 include truss_nl.flt beam_lr.flt, beam_lc.flt, etc.
359 As usual, we would like to stress that feedback is always appreciated ... we
360 welcome all comments as to usability and functionality. If you're using the
361 package please drop us a note and let us know what you think, even if you've
362 got no complaints whatsoever. Mail bug reports, comments, hints, suggestion,
363 complaints, etc. to one of us at the email addresses below.
365 Jason Gobat Darren Atkinson
366 jgobat@mit.edu atkinson@ucsd.edu