I performed a nonlinear regression using a simplified model of longitudinal dynamics:

P = m*v*a + 0.5*CxA*ρ^v^3


So:

P = (m*a+k*v2)*v

P = m*v*a + k^v^3


Engine power P and car mass m don’t change with DRS
I estimated k twice: first using instants without DRS, then with DRS (in quali→no tow).

Results are robust, as both teammates showed similar values (e.g. 21.4% for LEC and 21.9% for HAM).

Repost this if you found it interesting!