A high speed collision in low orbit can change a circular low orbit into an elliptical eccentric orbit that intersects a higher circular orbit, but unless there is an additional accelerating event at that higher altitude, it cannot recircularize its orbit at that higher altitude.
There are thus two takeaways:
1. By definition, this means that part of the orbit will always be at low altitude, regardless of the collision dynamics. So this means that it will continue to decay over time, albeit perhaps at a slower rate (decay being proportional to the time spent at lower altitude).
2. While that eccentric orbit will intersect with a higher circular orbital plane, it does so in a predictable fashion that can be routed around. The higher orbits are also much sparser, so the chance of this intersecting with a satellite that is already present is very, very small.
Thank you, others pointed this out too.
Proves that "celestial mechanics" is trickier than I thought.
I guess if a collision is messy enough, there would be secondary collisions between debris pieces, and it sounds like these in principle can push some junk into higher orbits. But I think the probability is really low; this should not be a concern.
There are thus two takeaways:
1. By definition, this means that part of the orbit will always be at low altitude, regardless of the collision dynamics. So this means that it will continue to decay over time, albeit perhaps at a slower rate (decay being proportional to the time spent at lower altitude).
2. While that eccentric orbit will intersect with a higher circular orbital plane, it does so in a predictable fashion that can be routed around. The higher orbits are also much sparser, so the chance of this intersecting with a satellite that is already present is very, very small.