"BEND on element xxx to xxx not properly specified"

I have a model where I entered an element and indicated a bend. I then went back and inserted a branch element at a previous tee location. When I go back to the first element and continue the model does not show a bend at the intersection and when the model is run reports the error in the subject line, which continues:
"The following element should define the DOWNSTREAM side of the bend".

How do I recover from this without rebuilding the model?

The only coding (model building) restriction in c2 is that once you start a bend, you have to finish it. This means the two elements that comprise the bend (in-coming and out-going tangents) must be sequential back-to-back elements.

For example;

Element 1: 10 to 20
Element 2: 20 to 30 (bend)
Element 3: 30 to 40 (NOTE: this element must follow 20-30)
Element 4: can be anything 29-410 for example

Element 2 cannot be separated from element 3 in "spreadsheet order".

Richard, I think your example is not completely correct (when piping model is concerned).
Element 4 can not be 29-410, because 29 will be c2 generated for the bend mid-point.

What is important is: Element 2 cannot be separated from element 3 in "spreadsheet order".

I think that it can be like Richard said when you have a trunnion for example.
Other thing is that if you specify the bend after you construct the element 29 to 410, when caesar "see" that 29 is assigned somewhere else in the model it will assign another number for the mid-point (e.g.31).

Regards,

Yes, I understand why this is happening, but the big question is: What can I do to recover from this?

Do I need to use the "insert" function to add an element?

This is already a big model and I don't want a momemtary lapse to force me to reconstruct a large portion.

Thanks.

Go to element 20 to 30 (bend) then use "insert", "after", then insert again an element 30 to 40. Then go forward to the other element 30 to 40 and delete it.

it should work.

@danb and regarding my previous post. My wording saying that "Richard has example is not completely correct" was not the best choice of words.
I ment to say that in a standard piping input (node increment = +10, no trunnions and other difficult stuff) it is not obvious.