Regarding Tension cable design for Skylight (Horizontal cable)

Your model is using a NONLINEAR ANALYSIS, which is a geometric nonlinear analysis, primarily used for large deformations. This may be what you are intending, but you mentioned a second order analysis (i.e., PDelta). 

For cable members, you might also want to use the PERFORM CABLE ANALYSIS. For cable members, if you don't specify a tension, STAAD will use a minimum tension automatically (which is probably causing some additional warnings for you). You can simply specify a low tension (for example, 1 lb or 1N) initially to prevent those with likely no change to the results.

Also, I noted in the attached model many warnings regarding the design of round bars. The round bars from the Dutch section catalog are not member shapes that can be designed per BS5950-2000 (ref. Member Property Specifications topic in the code details), so that will also result in numerous warning messages.

Dear Jason Coleman,

thank you for your reply. 

Now I modeled all tension rod as cables and given a nominal tension of 1 kN. But I am getting Instability warning and Deflection is more that 1075 mm. Please check the attached model. My doubt is can this system of tension rod will work for this much big span without any pretension (except nominal tightening  of tension rod).

  As cable (11).std

I think the instability warnings are coming from having a beam member that is attached at both ends to only Tension-only member, which cannot carry any moment. The beam has nothing to resist rotation about its longitudinal axis. I noted the first instability at node 3 which gives an instability in MY, which is the longitudinal axis of beam 1202.

In construction of such a structure, it is unlikely it would be detailed in such a way as to truly let the beam member "spin" about its long axis. So it is not unreasonable to assign a small spring resistance to the end node (here, node 3) to numerically represent that stability. This is what the program is trying to do internally by adding a very weak spring stiffness value to the matrix. But you can suppress the warnings by doing this in your model at these locations.

Conceptually, I think of it this way: a steel cable has a very tiny bending stiffness in comparison to the amount of tension it can carry. It is enough to prevent something like free-twisting of a connected member perhaps, but not significant enough to be considered as stiffening the overall structure. Therefore, the weak springs account for this physical differences in the cable's properties.

I think the instability warnings are coming from having a beam member that is attached at both ends to only Tension-only member, which cannot carry any moment. The beam has nothing to resist rotation about its longitudinal axis. I noted the first instability at node 3 which gives an instability in MY, which is the longitudinal axis of beam 1202.

In construction of such a structure, it is unlikely it would be detailed in such a way as to truly let the beam member "spin" about its long axis. So it is not unreasonable to assign a small spring resistance to the end node (here, node 3) to numerically represent that stability. This is what the program is trying to do internally by adding a very weak spring stiffness value to the matrix. But you can suppress the warnings by doing this in your model at these locations.

Conceptually, I think of it this way: a steel cable has a very tiny bending stiffness in comparison to the amount of tension it can carry. It is enough to prevent something like free-twisting of a connected member perhaps, but not significant enough to be considered as stiffening the overall structure. Therefore, the weak springs account for this physical differences in the cable's properties.