Ok I will read the article above mentioned by you.

In the end, Once tha VAL is established in a correct position the Caesar model should provide a correct value of displacements. I understood that maybe, truncate the model at the exact length of VAL is not a good modeling procedure. Then is ever necessary to model a length of pipeline greater than effective VAL, and this length of pipeline is reached after have performed few trials (with different lengths) where we can demonstrate that
the pipeline length does not affect the displacements at bends/transition AG-UG zone,etc.