However, I think it would be slightly annoying if you had to build a shipyard and then immediately retool it so this is a case of game-play over realism but with a reasonably plausible explanation behind it.

I could live with that if shipyards cost less.  Because currently, I tend to shuffle around shipyards and always keep a few freshly built new ones on expansion so I can use any new class immediately. 
It also helps to have set tonnage ranges for classes that never change (2kton corvette, 3x 6kton frigate, 3x 6kton sensor platform, 2x 10kton destroyer, 12kton C&C, 16kton missile cruiser, 20kton battleship, 24kton independent patrol, 50kton commercial jumpship) so you already know ahead of time when you want to stop expanding a yard. 

Idle ones (classes I'm not building atm or fully expanded yards) get "mothballed" by the fleet of space stations that are nothing more than a tractor beam. 

Matt was the one who didn't seem to understand the question.  I caught your answer just fine.

The first one is ALWAYS free.....

Must...build...newest...design....

John

The theory (behind the very first retool being free)

The first one is ALWAYS free.....

Matt

so I guess the reason is that the initial tooling cost is paid using industry capacity while the shipyard retooling is paid by shipyard modification rate?

I'm a little rusty, but if I remember correctly retooling is only based on the cost of the ship.

The retooling cost/time to go from class A to class B is based on the refit cost from A to B IIRC.  So if you're simply swapping out a few small sensors it will be trivial, even for a BB, but if you're changing big/expensive systems (engines, armor, jump engines) or increasing the tonnage significantly (which costs extra in the retool cost) then it can be very expensive.

That being said, unlike an actual refit (where the cost can exceed that of new construction) it seems plausible to me that the retool cost shouldn't exceed the cost for an "initial" retool (if it weren't assumed to be absorbed in the cost of constructing the SY ).  I don't know if Steve already has this limit in, but it might be worth putting a suggestion in the main suggestions thread.

John

Could I get a verify that, as implied, retooling to build a 'slightly better' version of a ship ("We swapped out the old Mark II missile launchers for the new Mark IIIs") will take a lot less time than retooling to build a completely different ship of the same, or close, tonnage? That is, if Ship A uses 3 type-4 ion drives, and Ship B used 3 type-4 ion drives, the part of the shipyard that assembles and installs type-4 ion drives doesn't need to be retooled and doesn't count, or counts less, against the cost of retooling? Or, in other other words, is retooling 100% based on the hull size of the ship, or is it based on how many components the ship shares with the prior ship?

This is pretty important, as it could effect which yards I retool, and why.  I was going on the assumption that it took just as long to retool from the Exeter Destroyer to the Exeter-A Destroyer as it did from the Exeter Destroyer to the Franklin Geo Surveyor, if both ships were the same overall tonnage, so it didn't matter what a shipyard was previously making.

I'm a little rusty, but if I remember correctly retooling is only based on the cost of the ship. However a shipyard tooled to build class a particular class of ship can also build similar ships without modification. Although I'm not sure precisely how this level of "similar" is calculated, it's based on the cost of refitting a ship from one class to another, which is based on how many components are different and how expensive those components are. There may also be some factors for how similar the ships are in tonnage.

This produces some unrealistic constraints, but as long as you know what they are the system works alright. For example if your Exeter-A Destroyer is the Exeter Destroyer with a new sensor and a few more Magazines you will probably be able to build the Exeter-A Destroyer at Shipyards tooled for Exeters, and vice-versa.

On the other hand if the Exeter-A replaced all of the Exeter's engines and turrets with new ones of updated technology but the same sizes, then they would become too different to share shipyards because the cost would be so different, even though in "real life" it seems like that wouldn't happen. If you had the same number and same sizes of components then it should be completely straightforward to continue using the same installation processes for the ship, only the production of those components would change.

Unless you are building 1k ton ships, you are neglecting the expansion time. Expanding a shipyard to a usable size (10k tons) takes quite a while.

it's in the quoted text, this case is with a simple commercial shipyard. they comes in 10k ton apiece.