[img]http://www.coade.com/ubbthreads/ubbthrea...0229561194ccdd4[/img]
Maybe this will help...
I have a buried L with a 400 ft run on one side and a variable run on the other side of a 90 degree bend. How long must that variable leg be in order to develop a "virtual anchor length" (VAL)?
The plot shows the axial deflection of the variable leg. The leg on the elbow side moves in the positive direction and the free end moves in the negative direction. When the leg collects enough friction on either side, given a fixed amount of thermal strain (a fixed internal load), it will not be able to move. Clearly, the 2000 ft and 4000 ft legs "lock up"; they are fully restrained as demonstrated by the large portion of zero growth. The 1000 ft run may or may not show a VAL and the 500 foot length is clearly too short to build up the required friction restraint.
If I look at the axial compressive load in each of these four legs I see that the 2000 & 4000 ft runs have THE SAME load of 45566; the 100 ft run has a little less - 45392 and the 500 ft has 41601. So we see that the 1000 ft run does not create a complete VAL.
I could also examine the deflection of the free end. 4000, 2000 & 1000 all achieve -0.1765 inch while the 500 ft run moves only -0.1759. These numbers would be more dramatic if I chose "better" soil properties. But we can see that the longer legs have enough length to collect the friction. In a way, you could look at this defection as the growth of the leg length that is needed to collect all that friction. Once again, the 1000 ft leg is just about long enough.
I am uncertain of what you mean by "virtual" cap. It is the pipe-soil friction in the axial direction that builds up the restraint load. If that free end was another elbow then the soil bearing around that corner would alter these results but as long as the run is long enough, that "free end" will not affect the position of the elbow included in this model. Inert pipe indicates a good location to end this model.