Is that so? I never knew about that. Thanks for explaining. That does make sense, having a control subject makes for a more standardized test without relying on prior experience of operating similar types of vessel, though it does cost more if the design doesn't perform as hoped. I guess China chose this approach due to its much shorter history of fielding modern vessels compared to say the US. Without those prior experience, designing the tests would be pretty troublesome if not for the control vessel of the pair.
That said, those direct mass production exceptions sure are bold. Or the Chinese navy just have a lesser tendency for requirement change/shift after design freeze which plagued the US navy? If they are both confident in the initial design and won't mess around too much with the requirements once the ship started building, then that approach also make sense.
I would think that the Chinese Navy is also guilty of constant requirement and shift after design in one particular area: nuclear submarines. That said they are doing this to desperately catch up with the US Navy. In the case of the Type 093, it appears they make a pair first, followed by an improved pair afterwards, then an even more improved pair and so on, but never truly making mass production of the type. Likewise with the 094 Jin class too. I surmise that each pair might be quieter than the previous, as PLAN continues to learn acoustic silencing techniques by trial, experimentation and evolution with each pair but never happy enough with the result to jump fully both feet in mass production.
Among the PLAN large production models, there is a similarity to the US Navy that there are production "blocks" or "flights" within that class. This points to changed requirements and need for improvements which is applied to the next production batch. The Type 054A for example, can be divided into three blocks. So does the Type 052D.
With the Type 075, there are two possible ways it can go down the route.
The first is that they build the pair, and after a while of evaluation, they may decide its not good enough and the experiences are collated into data to create an improved Type 075A class. Supposing they are much happier with the second try, the 075A will become the definite mass production item.
The second is that they build the pair. Then after evaluation they only need to make smaller changes and improvements to bring the class to snuff, and this results in a relatively improved Type 075 to be mass produced that is still within the same class but not enough changes to call it a new class altogether.