That said, I have no idea where he got that number, whether 40 m/s or 40 m/s^2; both are incorrect.
The best estimate available for the flight dynamics can be found at flightclub.io, specifically for JCSAT-14: https://flightclub.io/results/?id=eeaf889e-f3d8-4f9b-be98-b7...
This shows the landing burn starting at a velocity of 269 m/s and experiencing a peak deceleration of 12.3g = 121 m/s^2
I don't know how accurate these numbers are, but some fans put quite a bit of effort into getting the flightclub.io profiles as close to reality as possible, so they tend to be pretty good.
The approximation I had is almost certainly less reliable than flightclubs, not sure why it was so low, will have to look at the math I did and figure out what I screwed up.
The overall structure actually goes through all sorts of stresses, not all of which are measured in G. The combination of push from the rocket and air resistance compresses the rocket from both ends, something that isn't reflected in G numbers.
I think this is only sort of true. Yes, the fueled rocket is much, much heavier. But not all parts of the rocket are load bearing as it sits on the pad waiting for launch. All parts of the rocket experience the 12g (or whatever it was) deceleration.
Also, rocket skins are really, really thin (relative to their size, the forces they endure, etc.) I've read that without the fuel inside to reinforce the rocket, it is quite fragile - think of trying to crush an empty soda can vs a full one. This is part of why rockets which deviate too far from prograde on launch are ripped apart mid-air.
The Falcon 9 has a sort of partial balloon tank. It's strong enough for ground handling without pressurization, but it requires pressurization to handle flight loads.
Note however that it's pressure, not fuel, that provides the extra strength. These tanks are still pressurized even when almost empty.
The actual force would have been closer to 5.1G
[1] https://www.reddit.com/r/spacex/comments/4htenu/rspacex_jcsa...
Ugh sorry, there's a bug of some sort in Flight Club there. There's no way it was that large. A constant deceleration down from terminal velocity over 6 seconds is
a = (300 - 0)/6 = 50m/s2 = 5.1g
I'll need to look into that
https://www.reddit.com/r/spacex/comments/4htenu/rspacex_jcsa...
So it looks like that peak is due to the sudden increase in drag during my landing legs deployment. I deploy them at 500m altitude which is a time when the stage is probably still moving a bit too fast. It's a really large increase in cross-sectional area, but I'm probably not treating it correctly. I'll revise before Thaicom-8