2006-08-20 New version 0.7.1

Bugs and features:
Please submit any new bugs on the wiki.

Impact will attend the SIMDI06 conference in Gothenburg, Sweden. Meet us there!

2006-07-10 New version 0.7.0

Several bugs fixed in this release:
The Wiki has been updated by Al'be:do and Jonas. There is a new chapter on time step control which has been requested by several users since this is the most common source of errors.

The bugfix parade will continue for another release so keep sending in bugs. You can report them here. It is really easy and requires no log in or registration.

2006-06-25 New version 0.6.9

Several bugs fixed in this release:

A new wiki has been set up where you Impact users can add your tips and documentation.

Some very nice tutorials have been designed by Kumar and al'be:do. Many thanks gentlemen!

http://www.caelinux.com hosts a complete distribution with open source finite element software. Impact is included and there is a forum for discussing the Impact program there.

In addition, you are most welcome to join in the discussion of file format standardisation there. Impact will add interfaces to read and export to this format when the discussion concludes.

Bugfix will now contunue for the next release. Let us know if you find any.

Enjoy!

2006-05-27 New version 0.6.8

This release took some time, but it involves a new stage for Impact.
By using hardware bindings from the JoGL project (http://jogl.dev.java.net), the graphics will now be much faster providing you have a graphics card which supports OpenGL.

The release is not complete, there is more to be done and some functions are not implemented yet, but it would be appreciated if you try it out and give us some feedback.

Start the new graphics by running the ImpactGL.sh file (for Unix and Apple) or ImpactGL.bat (for Windows). The release includes support for these three platforms but more is available. Contact us if you have special needs.

The documentation has been updated for the English version.

Finally, Yuriy is working on the optimizer. We have disabled that part of the solver in this release and will enable it again when his new code arrives.

Enjoy!

2006-04-02 Reshuffling of version contents

First of all, thanks for the feedback on the UNA solver!

We are approaching the end of 0.6 branch and it is clear that some focus is needed on stabilization and bug fixing of the code rather than adding more features at this point. Therefore some changes including speed optimization has been moved to another branch.

I had the opportunity of using Impact for a real world case and although the modelling, solving and analysis went well, several bugs revealed themselves in the GUI which was quite annoying.


/Jonas

2006-03-28 Finding the UNA solver

I recieved a mail regarding another FE code (UNA). If there is anyone of you Impact users who know about this code and where to obtain a version of it or how to contact any of the authors, please mail me (Jonas) and I will forward this information to the sender of the mail.

Thanks

2006-03-25 Bugs in ImpactOpt

Excellent user feedback by Jens Bomholt has highlighted issues with the ImpactOpt feature.
We are on it. Please be patient. There will be a bugfix release.

/Jonas


2006-03-21 Impact 0.6.7 released

This release is dedicated to a new module that Yuriy has written. ImpactOpt will enable you to optimize the geometry of your design. The picture below says it all:

Optimization

You can run the optimization from the GUI (new module added to the solver part) or from the command line as follows:

java -Xmx1000m -cp .;doc;bin;Impact.jar run.ImpactOpt steps=20 ignore.stress(i)=35 examples/optimization.in

Some of the icons are not operational at this stage but will come later.

Enjoy!

2006-02-28 Impact 0.6.6 released

Several new commands and goodies in this release.
Remeber to give us some feedback if you find any bugs.
While you are here, why not give us a rating?

Keep well!

2006-02-11 Impact 0.6.5.mca released

This is not a full release due to a number of reasons, but still we release it for your pleasure :-)
The recent changed revealed a weakness in the build process for the cluster version meaning that this release does not contain a working cluster version. For this, we recommend you stay with the 0.6.5 release.
Finally, there is a possibility that Impact will be included in Software Developer's Journal CD, so keep your eyes open and buy a copy if you get a chance.

Enjoy!

2006-01-20 Impact 0.6.5 released

This release contains mainy contributions from Yuriy and his team.
Enjoy!

2006-01-05 Impact 0.6.4.mca released

Not at 0.6.5, but I thought a mid cycle action was needed.

This release contains many bugfixes and speedups:
There is a video (3MB) of how to use the preprocessor available for download. It shows some of the features as well as the principle on how to create geometry and mesh. For some reason the quality is really bad, so please overlook this.

Should anyone be interested in creating a video for Impact users, let me know (or just send one in).

Finally, a team member has reported susipcios results using the Shell_C0_3 element. Be careful when using it in your models until we have this sorted out!

Keep well!

2005-12-20 Impact 0.6.4 released

I had a chat with Santa and he urged me to get a new release out before christmas. I guess it could be concidered a present to someone :-)

It is far from perfect and I suspect you will find quite a few bugs in it but please - let me know and I'll have a look at it. The release includes:
Please don't blame me for the new ugly Icons. I don't have the skill on this like Yuriy. Hopefully we'll get this sorted out when Yuriy returns in the game.

A really - really merry christmas to you all Impact users across the globe!!

/Jonas

2005-10-18 Slow progress

Impact development is currently very slow. In effect, Jonas is the only active developer at this time and his available time for Impact is at a minimum.

If you want faster progress, please concider joining the team as a developer. There are many areas open for improvement.

Welcome!

2005-09-23 Impact 0.6.3 released

Lots of bugfixes and additional features in this one including:
Enjoy!

2005-09-04 Impact goes public

Impact will participate in SIMDI - Simulation and Data Intensive Fair, Gothenburg, Sweden

We will be present with a poster and personnel will be on place on Wednesday afternoon. Poster material available here and here.

2005-08-25 Version 0.6.2 released

Work on the preprocessor continues but also a sleigh of bugfixes in this release. Much of it based on input from you users. Thanks - keep the tips coming in!
The GUI was tested by users and got very bad reviews. Therefore a significant effort has been made in trying to improve user friendliness. Windows standard has been implemented:
The GUI should now have better user friendliness. Please give us feedback on this new approach!

In addition - a few more tweaks are made on the GUI. You figure out where :-)

al'be:do is planning to contribute a useful tool to Impact:
"
I'm working on a python script for fast and easy commandline tool with textfile output of periodic properties (forces, accelerations, movements etc.). 
The goal is that you can simply define the properties and the mode of periodicity (sinusodial, sawtooth, impulse etc.) at a given precision (time and space and the program produces a plain text file with the result. Now you don't have to hack in complicated movements or force changes manually any more. Simply cut and paste the text string into your *.in-file ans start the calculation. 
 
If you have any suggestions, just tell me. 
 
Planned waveform cycle modes:  sinus/cosinus, saw rising, saw falling, square with definable ratio, triangle, white noise 
 
Planned one shot modes:  square pulse, sin^2, soft pulse"


A first version of the program is available for download here.

Impact team has been extended with further personnel

Ahmet Alper Parker who will act as a scientific advicer. Ahmet is currently in a thesis stage of an Earthquake engineering master.

Richard Horton has joined the team as an official tester. This brings important feedback on "sharp case" use of Impact.

Welcome Richard and Ahmet!

2005-07-21 Version 0.6.1 released

Releasing this version due to vacation - i.e. no work for a while. Some bugs remain in the works with the most noticeable being that extrusion of arcs are not 100% correct. It is however quite usable apart from this. The release includes:
Overall, the preprocessor is now starting to feel good. Much more to come in the near future.

Have a nice summer!

2005-07-12 Version 0.6.0 released

First version of the 0.6 branch is a final cleanup of the 0.5 branch. The stage is now set for further optimization and creating a simple but robust preprocessor. This release includes:
The team has been joined by Tjepke Heeringa who will contribute as a developer. Welcome Thepke!

Have a nice summer!

2005-07-05 Version 0.5.9 released

More bug fixing in this release but also some new features. The pre-processor is now restored to it's previous state before the new nurbs-engine and simple models can now be created. Here are the highlights:
An upgrade is well recommended.

2005-06-16 Version 0.5.8 released

This release has concentrated on bug fixing. The following has been updated:
A requested prevention of tree reset when using modification commands in the preprocessor has been noted but proved more difficult to implement than initally assumed. Concider this work in progress for now :-)

An upgrade is recommended. Let us know if you find more bugs.
Keep well

2005-06-06 Version 0.5.7 now released

A redesign was made of the pre-processor internals. This will enable more complex CAD and mesh features in the future. At the moment, not all commands have been re-implemented but this will be fixed before 0.6 branch starts. A new Arc command has also been added to the preprocessor.

Documentation has been updated regarding the cluster version of the solver.
No other changes in this release.

Work now continues with several bugfixes supplied by the users. Thanks! You are helping in creating a better Impact.

Keep well!

2005-05-09 A brief note

Summer is approaching. Spending more time outside is a good thing for all of us but with the disadvantage of slower development pace for Impact.

Just a brief note to let you know that develoment of an Arc command is in progress. The prototype showed that the current NURBS engine was not enough for this, so a complete rewrite is underway. This explains the delay but will enable a very powerful graphical preprocessor when finished.

Keep well!

2005-04-24 Version 0.5.6 now released

This was a cleanup release. The cluster version can now be run from the GUI. Just start the GUI with the ClusterGUI.sh (or .bat) file and it should work. The printout will appear in the terminal window (to be sorted out later).

In the prevous few releases, a bug caused the solutions to be unstoppable from the GUI. This is now fixed and runs should terminate properly. Startup script bugs has also been fixed.

The Postprocessor now has a possibility of adjusting the max and min levels of the colour plot.

The Grapher has a range of new functions for result plotting.

The Impact team has been joined by a new member. Paul Ipe will help out with the optimizations for 0.6 branch. Welcome Paul.

2005-03-30 A little bug

A user submitted a bug report on 0.5.5. The ImpactGUI cannot be started under Windows XP. The fix is quite easy. Just open the ImpactGUI.bat file and replace the : with a ;

The bug will be fixed in the next release.

2005-03-20 Version 0.5.5 available

Improvements on the cluster solver continues. The updates include:cluster_2 scalability

Some benchmarking for this new version is provided in the picture. The scaling behaviour has clearly improved enabling up to 10 clients to give positive contributions. Models with heavy contact situation should be even better in this respect.

The variable distribution of elements is quite useful in a home or office network where computers of different strength are connected. The algorithm uses the initialization process duration to benchmark each client speed and then distributes the model accordingly. If your network fits this description, a significant improvement in performance can be expected with this version.

When trackers (special commands which measure things in the model) is not used, the amount of data that has to be shipped across the network can be much smaller. This version takes this into account to reduce traffic which in turn improves performance when many computers are involved in solving the same problem.

Note that Impact evolves in parallel with JavaParty. This version requires an update of the JP release. It has been updated and is available from the links below.

Enjoy!

2005-03-15 Why are the simple solutions the most difficult to find?

With oodles of help from Bernhard and Claus, the cluster solver has been rewritten 4 times now. The current solution is much simpler than the prevous designs, memory efficient and consequently much faster as well. All contact models can be used and the work remaining now is to get fractured element capability included. This is currently switched off for cluster solving.

The new design  has been bechmarked on up to 13 CPU:s (thanks to Bernhard for access to the cluster) with poor scaling results. Work is in progress to improve this but since the current design has no partitioning scheme, the results are  understandablecluster scaline

Still, if you have an extra computer or two in your office network, solution speed can at least be doubled on the cluster version.

A new element has been added, called solid_iso_4 which is a tetrahedron element. The interface to GiD and Gmsh has been updated to include this element. Since this is a collapsed hex element, the post processor cannot yet accurately show the element and it may look a bit odd. Note also that the autostep algorithm has not been updated to handle this element yet. Therefore, solutions involving this element will only work with a defined timestep!

The beam_spring_2 element contained a range of bugs and has been simplified and rewritten. It is now a simple 6 degree of freedom spring element and works in the global coordinate system at all times (ie. no local coordinate systems to be concerned about).

If you are interested in trying the cluster version, you will need the latest 1.09 version of JavaParty. Version is alpha and inofficial. Installation instructions are the same as for the standard version. Read more below under Download&Installation

Finally, dont forget to join the new mailing list. Get updated on the latest Impact discussion!

Enjoy!


2005-02-18 Version 0.5.3 available

This release certainly took its time. Reason being that the solver engine was re-written about three times trying to find the simplest - most efficient way of expanding impact into cluster computing. Turns out that the work done by Claus Wonnemann and Bernard Haumacher was the best way to go in the first place. Thanks to JavaParty, only one additional class was required as well as a recompile. Very elegant - hats off to the JavaParty team for their excellent API!

A first version of the cluster solver is now included in Impact. At the moment, it can only be run from the command line and it is painfully slow since the internal structure is not optimized. The cluster version should not be used to anything other than testing your setup and running a few examples at this stage.

Should you want to use Impact on a network cluster you have to set up a JavaParty environment first. To do this, you should visit the JavaParty webpage and download the latest version from here

This release contains a pre-build cluster version but in the coming releases in order to keep the download size down on Impact, the cluster version .class files will not be included. You will need to compile the version yourself from scratch. Thankfully, a build file is written to help you with this. Just use ANT for this and type ant compile_cluster. This should be enough. You can then start impact from the Impact directory with bash Cluster.sh ../examples/Ver_12.in

As usual, Yuriy has contributed with a new version of the GUI where the search&replace functions now are implemented and some more details have been worked on.

On the administrative side, the users forums have been closed and replaced by a mailing list instead. Just join the discussion or ask a question by clicking one of the links to the left on this page.

As usual, enjoy!

2004-11-17 Version 0.5.2 available

A few bugfixes starts off this release with the Shell_BT_4 element now determining its surface normal more accurate (thanks Rodrigo); especially for large deformations. The Nastran importer has also been fixed. Please note that if you are trying to import nastran geometry, read the users manual first because there might be some editing required to get the stuff through the parser.

The solver has been slightly rewritten to enable a larger part of the cycle to run in parallel including nodal sort although this is yet to be implemented. Yuriy and team has updated the preprocessor to read .in files.

As usual, enjoy!

2004-10-17 Version 0.5.1 available

Large updates on the preprocessor in this release. Yuiriy and his team are storming ahead at frantic pace, getting all of the bits together. This is the first release where the complete chain of development (making a model, applying boundary conditions, generating indatafile, solving and postprocessing) can be done all within Impact!

The updates include:

In working on the parallel solver (SMP), I bought a book about programming threads in Java. It is an easy read and is recommended for those interested in this subject. Title: "Java Thread Programming" Author: "Paul Hyde" It is top on the list at Amazon.

One reason that the results from earlier test runs showed none or small improvements could be that any second processor will automatically be assigned to handle the garbage collection in Java. Since Impact is generating garbage at a tremendous rate, the second processor on my machine is normally fully occupied with this, which means a default speed increase compared to a single processor machine. I expect that the real benefit with the SMP solver would show up on a machine with more than two processors. Should any of you faithful users have tried Impact on such a machine, please share the results with us. It would be very interesting feedback indeed!

Finally, check out the still incomplete but impressive tutorial in the russian documentation (under the help tab).

Let us know what you think!