FatigueToolbox just got a major overhaul. In addition to the Fatlab project, it now contains several other fatigue resources such as publications and fatigue data. In the future, the site will be updated more regularly and with more broad fatigue topics.
As mentioned, the standalone version will be discontinued and only the source will be provided for running under Matlab. This is because the deployed version behaved slightly different than the source version, and ran much slower. So, going forward, development will not be halted by limitations of the deploytool.
One of the limiting issues of the deployed version was in parallel execution of the code. It ran fine in Matlab, but not when deployed. So now this feature has been re-implemented, as of Fatlab 2.022. It uses the Parallel Computing Toolbox. Fatlab will detect whether the toolbox is installed and enable it accordingly. The user can then select a number of cores for the execution under Run Analysis.
Special thanks to Martin Dahl Kilt for helping with this feature and performance issues.
As of version 2.008 Fatlab will support also support the following ANSYS shell elements; SHELL63, SHELL181 and SHELL281. Additionally, the wireframe display now works for both plane and shell elements as well as solid elements of course.
Fatlab 2.006 is now available (source only, the standalone version will not be updated for the time being). The source code has been moved to SourceForge and is now under SVN version control, thus the change in version numbering. The commit log is now public.
The new version includes support for 2D ANSYS elements; PLANE42, PLANE82, PLANE182 and PLANE183. Like all other elements, only the corner nodes are read, so the will be limited difference between the Q4 and Q8 elements in Fatlab.
In some cases, the extraction of peaks/valleys from the stress signal were faulty. The extraction routine could not detect a peak consisting of two identical values, but it can now. The problem would occur e.g. for a sine load with an unfortunate number of sample points
For very long lives Fatlab would report infinite utilization under some circumstances. This was due to the use of single precision format in some of the calculations, which maxes out at approximately 10^38. Now using double precision, this is no longer an issue.
There were some problems using Matlab 2016a & b (graphics only) which have now been fixed, so Fatlab will run equally well in any Matlab version from 2015aSP1 and forward.
A new example have been added, which shows how to setup a model with a bolted joint including pretension. The model is also subjected to “bi-linear loading” due to two hinged actuators which are either pulling or pushing, leading to very different stress responses depending on the loading direction.
Fatlab 1.119 and onwards now support the Matlab Handle Graphics 2 engine (HG2) which was implemented in Matlab 2014b.
This means significantly improved performance for 3D visualization, e.g. during rotation of the model, and also includes antialiasing of the edges, so things look a lot nicer.
Programmatically, the main difference is that handles are now objects rather than numbers. This change means that when exploiting the advantages of HG2, previous Matlab versions will no longer be supported, i.e. future versions of Fatlab will require Matlab 2014b as a minimum.