[szoszon]

also, to test the software you don't need to launch the rocket, so I'm not sure why you even mention that..

[foota]

Tell that to the mars climate orbiter: https://en.m.wikipedia.org/wiki/Mars_Climate_Orbiter

[kortex]

What's the test fixture for realistic vibration oscillations of the support struts or combustion instability as the rocket breaks the sound barrier?

At some point, your in silico model needs real world data to validate it.

[szoszon]

Vibration testing is a thing, you know.. I actually do a lot of work with it so..

[gridspy]

Software is full of constants you can estimate but need to measure to be sure.

A few

- How long fuel takes to load

- How much is lost due to venting, how fast we need to recycle that to keep the ship on the pad ready to launch

- Actual engine startup times when many starting at once

- Actual thrust as the vechicle leaves the pad

- Actual response to control authority (how fast the ship changes course) for vectoring, gas thrusters, flaps

- Actual vibration measurements, how much measured acceleration is noise

[baq]

Exactly. Falcon 9 took years and a couple dozen launches to optimize the countdown and flight profile to the current best in class general purpose medium weight launcher with the flight hardware remaining relatively constant (if for nothing else than certification reasons.)