In the analysis of using a large, cheap, expendable, not-so-reliable booster to put water into low Earth orbit, we addressed the suggestion that we use flexible tanks to contain the water. The water would be frozen at launch, so that in this scenario we would hope it retains its shape until it reaches orbit.
This story demonstrates why we don't want to use a flexible tank for handling fuel.
God's Eye View: A big, flexible bag of liquid water floats in orbit. It weighs in at a quarter million pounds. The water has so much thermal mass that the 92-minute day-night cycles don't have much effect on its temperature; it stabilizes at about 50 degrees F. The water remains in its liquid state, neither freezing into ice nor, thankfully, boiling into vapor.
Undisturbed, the bag is very well-behaved, with just a stately spin or tumble that gives it a bit of a dumbbell shape. At one end of the bag, a large prong from a grapple fixture glints in the sun as it rotates in and out of shadow. Large white spots, each surrounded by a black halo for contrast, decorate the bag in an apparently random pattern.
Pan the camera to: Jabberwocky, our orbital switch engine. He approaches from out of the sun, coming to pick up the cargo. Water, so common on Earth, so precious in space.
Jabber's little electronic brain patiently observes the white spots. An image builds in his mind. Guided by those spots, he models the topography of the bag. Finally, he figures out that which moving light is his target. He zeroes in on the prong. A few million floating-point operations later, he has a model of the grapple fixture's trajectory. Time to make his move.
>Clank!< (There's no sound in space, but Jabber hears vibration.) His capture latch contacts the prong. >snik< Wires quickly close around the end of the prong. Jabber yanks on the prong and rigidizes the connection.
Waho-o-o-o! Now he's in for the ride of his mechanical life! His prize wobbles and undulates, swinging Jabber every which way. The universe spins and jerks back and forth, in and out. Fortunately, none of Jabber's delicate appendages shake loose.
His star trackers are worthless; getting a fix on the stars is hopeless when he can't even get a clear image. His accelerometers collect millions of bits of interesting, but useless, data on the dynamics of the floppy water bag.
Patiently, Jabberwocky waits for the undulations to dampen out. A few hours later, all is once again sedate. Clinging to the payload, he finds himself slowly rotating about a strange axis. Images of stars begin to make sense again. He starts to build up a new model of his mass properties. He integrated data from his accelerometers, gyros, and start trackers to determine his distance from the center of mass of the water bag.
Now, to stabilize this thing. He has to stabilize it before he can orient his main propulsion system for the rocket burn that will put his prize into a rendezvous trajectory with the orbiting hotel. Jabber lays out a plan. He'll stop the rotations, one axis at a time.
Tentatively, he fires a pair of thrusters. Oops! Waho-o-o-o! Here we go again! Nothing to do but wait for the oscillations to dampen out again.
This continues for several hours, until Jabberwocky's maneuvering fuel guage hits its red line. At that point, he has to give up.
Jabberwocky reluctantly releases his prize. He's quite a distance from the center of mass of the bag, so its rotation tosses him away. He quickly fires his maneuvering thrusters to stabilize himself.
>pop< >pop< >pop< >pop< Four quick bursts from coupled thrusters orient him in the direction he wants to go; first one axis, then another.
With a disgusted belch of flame aimed precisely at the ungainly water bag, Jabberwocky heads for home, leaving the payload to its fate.
So that's why we don't use a flexible bag full of frozen water to launch our fuel to orbit. A big floppy bag of water in orbit would be uncontrollable. Off hand, I don't know how we'd be able to maneuver it to where we want it to go.
We might attach some rings to the bag and send some people out to rig it up to some kind of framework, but that would add a lot more cost to the program. It would be less expensive to use rigid tanks.