Highly interesting quotes if you ask me. Makes a lot of sense.
These issues wouldn't have shown in CFD or the wind tunnel because the models aren't pulling any G's. At pre-season testing in Barcelona the cars did, but then maybe not in combination with the track falling away and those kerbs. They might've had full correlation aerodynamically, simulation data matching reality in perfect conditions. But now they've found conditions that ruin that, and the car stalls aerodynamically. Possibly they could've prevented this had they used the kerb data and historical data to extract the exact frequency range put through the car and done a modal analysis, but yeah easier said than done if you want to get it right. They then would've needed to put the deformed shape of the surfaces through CFD and wind tunnel to fully know the outcome and its effect on the car, and wouldn've seen the stalling occure.
Actually Red Bull have been historically pushing the boundaries with this.. Last year they tested running the car without tension bars connection the rear section of the floor to the gearbox at pre-season testing. Everyone has had those for years, out of necessity to keep the floor from deflecting too much. Eventually Red Bull showed up at the first race with these tension bars. Their attempt hadn't worked.
Anyway the impressive thing is that they clearly identified the stalling problem, the parts that are causing it and the corners in which the stall occurs. They probably did that over the weekend, possibly after qualifying but more likely after the race once they had all the data, with different fuel loads and therefor different cornering speeds. They then had about four days to introduce counter meassures. So that probably means first a bunch of iterations between CAD work and FE-analysis to settle on a sufficiently stiff geometry which ultimately will mean thicker bodies. This will then have to go into CFD for analysis, the thicker bodies will likely slightly decrease performance in optimal conditions. Further alternations to the geometry might be needed. Once satisfied it needs to go into the wind tunnel for validation, for that there needs to be drawings produced and 60%-scale parts manufactured. Probably using some form of rapid prototyping. Then the real parts needs to be manufactured, possibly carbon items that need to be in the autoclave for some hours. They then have to go through validation and NDT testing. If all OK, ship to the track asap for testing on one of the cars in FP1. They likely send another batch to arrive on saturday morning.
Once it's friday and FP1 begins, the driver hit a massive yellow kerb on his outlap and the parts are all ruined