To model molecules as hard ellipsoids seems of course unrealistic, even if the molecules are elongated and have a belly. After all, the real interactions (electrostatic, magnetic, ...) are not “hard”.
But if you want to extract the effects that are induced by the shape of the molecules (e.g. elongated, belly), then hard particle systems tell the story: here shape is the only player. And if you later study a new system where you know the shape of the particles, you already have an idea of what phenomena can occur, even if you don’t know the real interactions. For example, you can predict the emergence of a nematic phase* in elongated molecules: the fact that they are elongated already brings a tendency to form a nematic phase. It’s a matter of geometry.
The simplicity of such models also makes computer simulations easier and faster.
I simulate crystal structures (i.e. everything ordered) and the dynamics of ellipsoids.
* In a nematic phase, all particles point in almost the same direction, but the positions show now ordering. LCDs (displays of laptops, calculators, etc.) make use of such phases of matter.