I think it can be helpful to think not of your simulation as trying to match your animation - making it more realistic - but instead think of your animation as a guide for your simulation.
The difference is that you aren’t looking to your animation for a final result, but as effectively the “rig” to your simulation. So, if the dragon needs to reach a target quicker, tune the animation until it does. What the actual animation looks like is not important and likely won’t look very good at all.
Nobody demonstrates this better than Jason, such as these.
The trick is to let your simulation do what simulations do, and steer it using your animation as your controller.