1- The rocket's weight is its Mass times the Gravitational pull (-9.8m/s/s at Earth's surface). This makes the acceleration of -9.8m/s/s correct even when fixed on the ground in the pure calculations of Physics.
Acceleration is dv/dt or F/m (the two are defined to be the same in classical physics). In the frame of reference we're using (the launchpad), velocity is 0 and unchanging before liftoff. Therefore acceleration is 0. Prior to liftoff, the net force on the rocket is 0, therefor acceleration is 0. Claiming that acceleration (within a given frame of reference) is non-zero when velocity is unchanging (within that frame of reference) is wrong.
The Velocity is constrained to not be negative.
Prior to liftoff, acceleration is zero because the launch pad exerts an upward force equal to the weight of the rocket minus the engine's thrust. Before liftoff, net force is zero, therefor acceleration is zero, and and velocity is unchanging. The OR simulation engine appears to ignore that upward force and applies constraints to position/velocity to try to hide that omission
without also constraining acceleration.
2- If you have an Accerometer chip in your rocket then it will measure an acceration of -9.8m/s/s when the rocket is sitting on the pad. Another proof this thinking is correct.
That is measuring the force of gravity, and the sign of that value is positive (the same as the acceleration produced by the engine thrust). The "extra" acceleration value shown by OR has a
negative sign. In the frame of reference we are using (the launchpad), what is measured by the accelerometer chip when at rest is not acceleration — though it is indistinguishable from acceleration from the accelerometer's frame of reference. That value should be zeroed out before flight if you want to know actual acceleration in the launchpad frame of reference.
3- Calculating the rockets motion one starts with the Thrust as applied to the Weight (Mass times Gravity) of the rocket. Once Thrust exceeds weight then there is motion and Velocity can be calculated. NOT the other way around (Acc = dv/dt is not done although mathematically correct).
Your simulation appears to be ignoring the upward force exerted by the launchpad before liftoff. If you do that, you end up with pre-liftoff downward acceleration as shown by OR. Displaying non-zero acceleration that doesn't produce any ∆v makes no sense. We know why it does that, and we know it doesn't affect post-liftoff results, so I'm happy to ignore it. But I'm not going to admit that it's "right".
Yes, I have written rocket simulation code and after finding Open Rocket I compare the results are the same in both.
Agreement is not the same thing as correctness.
If in your definition of things, F ≠ ma, a ≠ dv/dt, and ∆v ≠ ∫a dt,
then this discussion is rather pointless.