I'd say it was more a show of strength sending a full Galaxy-class starship into the Gamma Quadrant than a true 'prepare for battle' scenario. At that point Starfleet felt a diplomatic solution could be found to the Jem'Hadar threat and Captain Keogh seemed like the kind of guy who'd shy away from sending only half his ship on a mission...
But the idea was supposed to be the opposite of that -- separating the saucer shows that you mean business, because you're freeing the stripped-down battleship to operate without being hampered by the dead weight of the non-combat section. It's like drawing your sword from its sheath. It makes the ship more intimidating, not less.
At least, that's how it was intended to work. But the show gave up on the idea so quickly and got so much into the habit of showing the saucer attached during battle that we came to think of the battle section alone as "half a ship" rather than a whole battleship undocked from the massive research platform it usually carried around.
I assume the in-universe reason matches the real world one - it seemed like a good idea in development, but then it was realised that it had relatively little practical application.
Routinely separating the saucer rarely made it ‘safer’ as it was a sitting duck without the stardrive section.
People keep saying that, but it doesn't make sense. The "sitting duck" situation people imagine is one that was
never actually shown in TNG. The first time, in "Encounter at Farpoint," the saucer was left behind during a high-warp chase while the enemy pursued the battle section, so the saucer was instantly left far behind the battle and out of danger. And in "The Arsenal of Freedom," the ship
left the danger zone, separated the saucer someplace safe, and then took the battle section back into danger. That's how it was supposed to work. The "sitting duck" objection is a straw man. The only time we ever saw the separated saucer actually
in the combat zone was in BOBW, where they did it on purpose to have a two-pronged attack, which was not at all what separation was meant to be used for.
As I keep saying, the situation in "The Jem'Hadar" was exactly the kind of situation that saucer separation was meant for -- the ship is at a safe port and leaving its non-combatants behind before it proceeds into a known hazard situation. In TNG, they could easily have done the same thing. Say, if it's an episode where the ship has been ordered to investigate Romulan attacks along the border, you open it with a scene of the saucer being left behind at the nearest starbase before the stardrive section proceeds on the military mission. And then you could have a B-plot with the characters on the saucer while the A-plot confrontation with the Romulans occurs elsewhere.
It also took up time in battle situations, made it difficult to escape quickly, and may have made the ship less effective in combat anyway - by removing the main phaser banks and the saucer impulse reactors, which Riker seemed to think would be pretty useful in BoBW.
As I've said before, that's a retcon that precisely reverses the original intent of the show's creators. The stardrive section was meant to be specifically
designed for combat, while the saucer was the non-combat portion of the ship with only defensive capability. It makes no sense to assume that the designers would have collectively been so incompetent that they failed to realize over 22 years of design and testing (per the
TNG Tech Manual) that the ship didn't work the way they meant it to.
After all, the saucer is
huge. It's like hooking a heavy trailer to the back of a race car -- it inhibits the car's performance rather than enhancing it. The saucer demands a lot of power from the main engines, and freeing up that power for maneuvering and combat should offset any loss from the saucer's reactors. And the "cobra hood" has its own phaser banks to take the place of the saucer's. The strips are shorter, sure, but the idea is that the entire phaser strip gets its power from a single source, so adding more length doesn't add more power, just more directions to point in. The reason the saucer needs so much length of phaser strips is because it's so damn huge and needs the coverage. The battle section is a smaller target, so its smaller phaser strips provide proportional coverage, and the total power it can bring to bear should be on a par with what the complete ship could deliver, because it was specifically designed to work that way.
I would guess because the writers knew they would rarely use that functionality, and it would make the model more expensive and complicated to build and use, which would have negated the point of building the smaller model anyway.
That doesn't follow. The problem with the 6-footer wasn't its separation ability, it was its unwieldy size and lack of surface detail. As Dan Curry
explained:
"We had the six-footer and we also had the two-footer, and something that must have intellectually interesting to Andy was the idea of this perfectly smooth ship that was absolutely huge, and if you saw it in person the scale would be rapidly apparent. But when we photographed the big ship very smooth, we couldn't tell the difference between it and a very small ship very smooth, because photographically, particularly on television and in order to create that sense of scale and hugeness, we needed three-dimensional relief, something to cast shadows. That's why we built the four-footer. And also because the six-footer was really too big to be manageable, particularly when we shot matte passes, where you have enough lit [card] behind the ship to obtain a silhouette. (...) So if you had a big move going around it, it would sometimes take a day to get the matte pass because the ship was so huge. And the four-footer turned out to be a more manageable size." (Star Trek: The Next Generation USS Enterprise NCC-1701-D Blueprints,booklet, p. 9)
Ronald B. Moore added that it was hard to take the saucer off, but that was also due to its size and weight. It should've been easier with a smaller, lighter model; plus, they could've learned from their experience with the first model and designed a better separation system if necessary.
Or they could've just built two more models of the separated saucer and stardrive section, and used stock ILM footage to show the separation when necessary. Sure, it would've cost more, but it would've been a one-time expenditure that would've given them more options and flexibility going forward --
if the producers hadn't already lost interest in saucer separation well before the 4-footer was commissioned.