I've found it tricky to estimate the thrust load on a wind turbine, too. I did try working it out, but found a lot of variables that need to be understood, and they aren't easy to quantify. If you go too conservative on each of them, you end up with big numbers like the OP has.
No, a "flat plate" estimate doesn't work, even though it's a simple way to estimate, the result is way too high. Except - surprise - a wind turbine in a runaway condition applies huge thrust loads on the tower that are much HIGHER than the flat-plate estimate. Several counter-intuitive things like that.
To get myself out of this indecision feedback-loop, I found some load-test data where a WT was instrumented by the NREL and run for a long time in real-world conditions. The report included thrust load measurements with some correlation to wind speed. You may be able to find reports like these on the NREL website (it's been reorganized so I think you need to use Sandia's servers now).
[EDIT: Found it:
https://research-hub.nrel.gov/en/publications/small-wind-research-turbine-final-report]
Below is a chart from report TP-500-38550 (October 2005) the test of a Bergey Excel-S. It's rated at 10kW at 13 m/s, having a 5.6m (18.4 feet) diameter rotor. You're looking at data with a lot of scatter, so I wouldn't just rely on the average through the middle, but for design of the tower account for the maximums. If your WT has a diameter of 2 meters, then I'd scale it down by swept area. Square of 2/5.6 is a factor of 0.127. If the Excel's peak is 5500 Newtons, then your thrust goes up to 700 Newtons. Maybe use 1400 N to have a safety factor of 2. I have nothing to tell me if that's enough to address a potential runaway, but it's better than nothing.
Don't use "kg" for forces - misuse of the unit of mass not force can easily confuse.
