Like CraigB says try to split at anchor. If you do not have any you have a good change that your piping will not behave as you expect. I use a simple rule. If there is a piece of pipe that is not "forced" to move into specific direction then no good. Any movement depending on friction or spring rate is not "forced". One day pipe moves one direction and next some other direction.

I have used sometimes splitting in a way that my two models overlap, e.g. I have same section of the pipe in both models. Calculation results near the ends in both models are wrong but I make this clear in the report. Typical place to use this is when I calculate an existing system, which doesn't have ANC and there are few kilometers of piping.