Archiving Mechanica results?

[TractorGuy]

My department has a file server with limited space and we have found that most of the usage is coming from FEA results that were archived in the past. In most cases, the entire directory that was used for analysis is compressed into a 7zip file and placed on the server. The problem is that most of these files are several hundred MB in size. Because of this, we are evaluating a way to keep all of the results data, but dramatically cut down on the file sizes. Does anyone know the best way to archive the mechanica results so that file sizes are reduced? What files are necessary to keep and which ones can be deleted?

[JonathanHodgson]

In theory, if you extract the .prt / .asm files that are copied within the analysis folder, they should contain all the information to re-run the analysis. Perhaps also keep the .rpt file for a summary of the run, and some screenshots of the results for quick reference?

[TractorGuy]

Is there a way to save just the results window and the supporting models? In other words, I want to have the models already setup in case I need to re-run the analysis, but all I really want to keep is enough information to show the results window. What files are critical to showing results and which ones can I blow away?

[JonathanHodgson]

Not sure, but I believe that .rwd files use absolute paths. If you create them in one location and then move both them and the results to another location, they no longer work.

Unfortunately I suspect that "just" the results data is the majority of the hundreds of MB per result set - it has to store a matrix of stress, displacement etc for every element, at the resolution you specified in the 'Plotting Grid' setting, and this is a lot of information.

[integrisbrad]

I second Jonathan. There are several files that can be deleted (I don't remember what they are and I don't currently have a Mechanica license), but the majority of the big files are going to be the results themselves. Since space was always a premium, I encouraged people to do the same as Jonathan suggested when I was with my previous employer...just keep what you needed to re-run later. However, there is always a chance that it may not run or give you different results in later versions as software changes are made that control meshing and solving params.

Several years ago when I was in engineering software support for a big company, one of the ideas that I was trying to push was to investigate lightweight results viewing solutions that could handle FEA results from all of the different software we were using. There wasn't enough support from the community to fund the project, but we did look at tools like VCollab. Note that I'm not affiliated with that company in any way. I thought it would be beneficial to store lightweight results files with projects/parts/assemblies in our PLM database that everyone could view in a common format.

[346gnu]

The rules (law) sometimes dicates what you are allowed to throw away.

Pressure vessels, Aerospace ...

Sometimes it is necessary to be able to recalculate the exactly the same answer in the future and compare it with the information used for design decisions. (in the word of planes this can mean maintaining old computers and software for a long time)

For most general purposes, document but throw the iterations away and keep only those analyses from which numbers are taken for decision making.

[SylvainA.]

I agree with Jonathan and Brad.

I also would just keep ASM/PRT files copied into the analaysis folder in order to re-run the analysis later if needed.

FYI, in Microsoft Windows, the "compress content" option can lower the size of a result folder to less than 50% of its original size.

[TractorGuy]

I just wanted to say thanks to everyone who gave input and let everyone know what I ended up doing:

To begin with, I created a formal standard for archiving future FEA results. Here is a quick overview of what is in the standard:

1. The first step is to use best practices when creating an analysis. This means properly naming load sets, constraints, and analysis so that the next person can easily understand what is going on. Also, if several runs of a similar analysis are executed, they should be numbered or labeled with a date to avoid confusion.

2. The top level model should be renamed by the following convention: description-fea-date. For example, if I were running an FEA of a frame today I would use frame-fea-20131022.asm so that the model is easy to locate.

3. Create multiple screenshots of each analsysis that would be useful for a presentation or report. Name the screenshots acording to the analsysis which was used to produce them.

4. Delete all of the subdirectories in the root analsysis folder where the meshing and results data is stored.

5. Compress the analysis folder in 7zip format. The 7zip format provides excellent compression and fast decompress times.

This has dropped our results data from about 120GB to about 10GB.