As Uzair mentioned up there that is one way for the modelling of sag rods, but as per my experience it do not give satisfactory results. several reasons behind this one as it depends upon your building geometry like size of purlins or beams you want to bridge and there length as well. The sag rods behave as pin elements at the ends and they are most of the time lose tight with the main member which is a bit difficult to model, they come to action only in case of severe loading (Normally they are just holding the alignment of main beams). if you model them as per above method you have to break the main member at the connection point to send through the force to them so they act as tension member but as same time you have to make them pin assign due to which you will get very strange deflection shapes for the model, even the force distribution changes alot in some cases again depending upon geometry. Some time you hit the right nodes but most of the time results are a bit strange and some compromise is done in design of members, the best way i figured out is that you model line elements (with none material or frame property just null element) as sag rods and apply loading keep the main member as one piece without break and take out the member force generated in the null element. Design the tension rod or sag rod for that force manually as per required code.
The above method is more suitable for building braces for lateral actions or tension rods for hanging cantilevers where high loads are expected. Thanks