I think there needs to be some understanding of the failure modes being modeled, and whether or not they are achievable through shooter effort, or if they are random. In your example above, hitting cable will cause a flap to get stuck. But flaps already get stuck. If you are hoping the damage model will include internals depicting the cables so that flaps will get stuck only when cables get hit, then I propose that would only be worth it if an attacker could effectively aim for a cable with a reasonable probability of hitting it. Otherwise making the cable damage random is effectively the same thing.
So you'd need to break this down into two lists:
1) what failure modes are missing from the current modeling
2) would require discrete affectation or random?
Well, my pie in the sky ideal would be having a down-to-the-rivet solidworks model of the plane that deforms according to the bullets that hit it.
Out here in the real world, I think it might be neat and somewhat attainable if there were some vague outlines inside the body of the airframe that would be damageable, so with the cable example, where the cable runs there'd be a hitbox that if it were struck would result in cable severage. For the engine, have the cooling, oil and maybe electrical systems outlined on the engine block, same deal.
For the control surfaces, maybe have them broken into 3 or 4 segments that would have x number of hitpoints and have the surface's effectiveness scale with the amount of damage on it until it's blown off.
edit: No random stuff at all, other than possibly amount of damage based on distance from the cannon explosion. Whatever gets struck gets damaged.
Wiley.