Monthly Archives: September 2013



The analysis software of choice at the University is ANSYS and whilst it is a very powerful piece of software it can be a little overwhelming at times for students who are starting to learn how to analyse structures.  Whilst we start them on hand techniques such as moment distribution and then through to matrix methods combined with using LinPRO there's a bit of a jump from those into full blown finite element packages.  One of the good things about being an academic again is that I get access to free copies to all of the AutoDesk software and so I've been giving ROBOT Structural Analysis a whirl to see how I get on.

360:365 - Robot...

I've not really used this software in anger yet, but it's comparable to many of the alternative offerings out there that I've used over the years such as QSE, Masterseries, GSA, and SFrame to name a few.  Whilst LinPRO is only 2D, ROBOT brings in the extra dimension to allow 3D frames to be analysed, and importantly for me and the dissertation topics that I offer, it also has the capacity to deal with cables and tension only elements and its this aspect rather than the building modelling that I've been really interested in for the last couple of days.  I started by modelling a simple guyed mast just to see how well it was behaved under dead and live loads.  A nice simple model like this has let me mess about with the different ways of controlling the cable tensions and seeing how the various settings in the analysis engines and element releases work.

Screen Shot 2013-09-06 at 14.32.18

After spending a little time getting my head around the releases, for some reason a local x axis restraint doesn't stop the mast spinning with a fixed base, so I had to throw a Global Z restraint to get it to analyse with no warnings...  as a newbie to the software, I'm sure I was probably doing something wrong, but it's all about learning and so there's nothing to beat myself up about.  I was able to get a pretty simple model up and running that seems to be fairly well behaved after about 10 minutes of tinkering which gives an idea of the learning curve involved.

There's a couple of forum posts knocking about that discuss where people have had errors with getting their models to converge, but this appears to be mainly through the application of very large prestress or large change in lengths that are incompatible with each other, meaning that the forces can never balance out based on the criteria set out.  The easiest way I've found to get round this initially is to set the lengths to have a zero change in relative length (as shown below), essentially taking their original starting length with zero prestress or by assigning a very low prestress of say 1kN or 5kN depending on the scale of the system.

Screen Shot 2013-09-07 at 13.52.12

After a bit of jigging round, I've managed to get a few different types of cable supported structures behaving themselves, from simple gallow brackets, to cable trusses, and even a simple icosahedron tensegrity prism.... although I need to spend a little more time rotating the geometry to reduce the number of supports needed for the prism.

Screen Shot 2013-09-06 at 16.35.37

Now I've got the models behaving and analysing efficiently, I can start to experiment with what difference the various analysis engines make and how they can affect the behaviour of the structure, particularly as I start to increase the complexity into cable nets and stacked tensegrity prisms.  One thing that I'm finding difficult is tracking down a decent ROBOT community, particularly as I think there are quite a few things I could be doing with the API to connect it with Mathematica.

If anyone knows of one, I'd appreciate a heads up or a link....