cable roof structure

huangyhg

cable roof structure
hello,
i'm currently analysing a cable roof structure very similar to the north carolina state faire arena at raleigh.
anybody have some info on how to modelize the structure, or how to obtain the tensions in the cables?
please see picture
thanks
simon
to compute tension in the cables, download "uss wire rope engineering hand book" from this page of my website:
tensile architecture, i love fabric structures.
there is not alot of information out there that deal with this form of structural engineering. alot of the specialised software is generally proprietary of the company that developed it, usually through the work of individuals who have a very thorough understanding of nonlinear structural analysis.
most commercial finite element packages will have a cable element (essentially a tension-only bar element) that you can use to determine the cable tensions. you will need to perform a nonlinear analysis, and you must have a reasonable estimate of what the intial geometry (the form) of the structure is in order for your analysis to run.  these structures rely on stiffness that is generated from tension and curvature, this is called the geometric stiffness and is summed with the elastic stiffness to give you what is called the tangential stiffness.
if you don't model adequate curvate or pretension, an instability will form because there will be insufficient stiffness out-of-plane.
the elastic modulus of steel cables differs from structural steel, so i would have a look at what slideruleera has to offer on his website with regard to that.
alot of specialised software that is used to analysis fabric structures work on the dynamic relaxation algorithm.  it is a neat little manipulation of the dyanamic equation of motion, works similar to the direct stiffness approach only you do not have to invert your stiffness matrix, so you don't have to worry about stiffness matrix singularity, there is published data out there that has shown this method to converge significantly faster than the direct stiffness method.
a nifty little trick you can programme yourself is the force density method, this approach linearizes the nonlinear force equilibruim equations, performs one single matrix inversion and will pop the mesh into equilibrium like an umbrella (literally).  downside is you do not have control over the tension field of the cable network.
good reading on the subject include buchholts, "cable roof structures" and lewis "tension structures". keep us informed on how you go with this because like i said, i love this form of engineering, it's just a pity i have never had a chance to work in this field.
best tip, leave this form of engineering to the specialist.
if my engineering history is correct, the raleigh arena in north carolina is what insipred a young frei otto, who later went on to mastermind the cable roof structure for the montral olympics, albeit by soap film modeling that proved to be a very ineffective method to determine cable forces.
frei otto was the director of the institution of lightweight structures at the university of stuggart until he retired in the mid 1990's and was suceeded by strobek, who is giving a keynote lecture at the australian structural engineering conference next weekend.
that reminds me, another good book on the topic is "the european design guide for tensile structures".
levy and spillers have a good text "analysis of geometrically nonlinear structures".
slideruleear and asixth thanks for your help. will keep you posted.
here is the picture of the structure we are looking at in montreal.
asixth, very astute detective work by the way. although from what i undertand, it was built before the olympics.
see picture attached.
thanks again!
asce also has a guide on tension structures but it doesnt give any real useful design methods.
one thing we do in cable supported bridges is to perform an initial linear dead load analysis with pin supports at each cable point.  this provides an idea of what the vertical component of the dead load cable force will be.  it is fairly accurate because you generally do not want any deflection of the cable points under deadload.  each cable is installed and shortened with a jack until the structure is on profile.  the resulting tension in the cable should be approximately the same as what came out of the initial analysis.  from the initial analysis, you can figure out the approximate dead load force in the cables and input them as pretensions.  after a few iterations and tweaking of your counterweights or tiedowns, you should have a good dead load model.