I am not familiar with ISO 15189 standards--at a minimum you obviously have to follow them, along with manufacturer's requirements--but it seems to me...
...like you are being advised to validate a particular analyzer by comparing it to another exactly the same, which presumably was compared to another exactly the same, which was...
With any new method, you must show correlation with a previous method, in this case, your manual method. Comparing it to an analyzer at another facility seems rather pointless. (The exception being if the other facility is part of your system, and you use the same procedures and reference ranges).
You should also be doing a reference range study to validate your reference range, especially if there is significant variance in the correlation. It has been our experience that automated methods correlate pretty well with the old manual ESRs.
Scott