Currently, break nodes looks at the 2 nodes that a part of an element and puts a node directly in the center.

For the most part, this is ok, until you end up breaking an element at the junction of one section of the model with another, and you have a huge gap between the 2 nodes.

I recommend a little more intelligence here.

Step 1: if node 2 - node 1 <= node increment, then continue as normal
(node increment is set up in the settings file, and is typically 10)
Step 2: Check all elements attached to the element.
Case 1: No elements share the same vector.
Case 2: One element shares the same vector.
Case 3: Multiple elements share the same vector.

In Case 1 and 3:
Check to see if the smaller of the 2 nodes + default increment is available. If so, use this.
If not, if the larger of the 2 nodes - default increment is available. If so, use this.
If not, repeat the steps above, except with 1/2 the default increment.
Again, if not for both, choose the smaller node + 1/2 the default increment, and keep adding 1 until you get to an available node number.

In case 2: Look at the node number beyond on the colinear element. If it's smaller than the node number, then check to see if node number + default increment is available. If not, then try half. Repeat until the increment is 1. Then keep adding 1 until an available node number is obtained. Go in the opposite direction if it's larger. Add an additional check to ensure that you don't input a negative node number.

We can look at this idea.

I'll try to summarize more succinctly, and act less backseat programmer:

Make broken elements' new node more intelligent, based off the node increment value when available, or some portion of it, when it's not.

Consideration is provided for a few cases:

1) Continuous straight pipe in one direction.
2) Two straight runs meeting in the center.
3) A typical tee configuration.
4) 2 tees put together.
5) N-number connections placed together (e.g. for multiple nozzles projecting from a vessel).