Question-and-Answer Resource for the Building Energy Modeling Community
 | 1 | initial version |
OpenStudio's behavior is based on a mix of superstition and (perhaps outdated) guidance that E+ has fewer convergence errors if there is a demand bypass pipe, even if that bypass has no flow or negligible flow moving through it. I admit I don't have data handy to back this up though. I am more than happy to reconsider this behavior if we have some evidence to motivate a change. I have always been a little reluctant to insert magic stuff into the topology like this, but I think this is fairly unobtrusive. It doesn't muck with nodes surrounding components you put in your model and it has been in there for a long time. Maybe if we make a change we should just beef up the logic to not add the demand bypass pipe if there is already one there.
As for the inlet and outlet pipes, the deal is that the way E+ input syntax is structured, there are necessarily "Branch" objects before and after the splitter mixer pairs, and branches necessarily have at least one component on them or the input syntax is not valid. In the OS -> E+ translator software, we are checking to make sure there is at least one component on the inlet/outlet branches and adding a pipe if nothing else is there to keep the simulation from failing for a trivial reason.
