Well, I've spent the last few days being reminded of how poorly I understand aerodynamics. It doesn't help that I'm using an aerodynamics characteristics report rather than an aero database, either. Lots of stability derivatives, not as many coefficients as I would like. In any case, I've got something plausible enough for ascent now. I completely bypassed Orbiter's airfoil system and just implemented it myself - it's slightly more work but means I don't don't have to trick the airfoils into doing what I want. Thank martins for AddForce().
With ascent aerodynamics out of the way, the last "major" task before the first release is done! I've accumulated around a dozen smaller cleanup tasks I want to complete first, but they should all fall quickly. Hopefully you'll have the ascent prototype to play with soon!
---------- Post added 09-16-14 at 08:44 AM ---------- Previous post was 09-15-14 at 12:46 PM ----------
I started digging into my backlog of little stuff. I started with the drop tank, which now has some rudimentary aerodynamics after separation and improved physics if it recontacts the nose while hinging. I had also forgotten one of the components of force during separation, so the dynamics changed somewhat. Now the aft end of the orbiter is "carried up" by the swinging drop tank if you separate in vacuum. If you separate at high dynamic pressure (don't do that), the aft end is forced downwards with great violence.
More importantly, I implemented anti-torque steering for the gimbals. This attempts to negate all aerodynamic torque in the pitch plane during ascent, so you can keep a constant pitch rate for those nice gravity turns rather than fighting the controls all the way up. The yaw plane is still free to stabilize aerodynamically.
Flying nearly horizontal at low altitude (don't do this). The cross indicates severe negative engine gimbal (almost -5 degrees, near the hard stop) trying to fight aero torques and keep the nose high.
Next on my list is to change out the gimbal control system. Instead of the pilot providing direct engine gimbal inputs, I'm going to treat them as commanded angular accelerations and have the engines figure out how to achieve that. That will help avoid the nose getting a bit squirrely at the end of burns when the vehicle is light and control authority is high.