(Abridged version) We explore whether a scenario that combines an origin by mergers at $zsim$1.8-1.5 with a subsequent passive evolution of the resulting S0 remnants since $z sim$0.8-1 is compatible with observational data of S0s in the Tully-Fisher relation (TFR). We studied a set of major and minor merger experiments from the GalMer database that generate massive S0 remnants. We analysed the location of these remnants in the photometric and stellar TFRs assuming that they correspond to $zsim0.8$ galaxies. We then estimated their evolution in these planes over the last 7 Gyr. The results were compared with data of real S0s and spirals at different redshifts. We also tested how the use of Vcirc or Vrot,max affects the results. We found that just after $sim$1-2 Gyr of coalescence, major mergers generate S0 remnants that are outliers of the local photometric and stellar TFRs at $zsim0.8$. After $sim$4-7 Gyr of passive evolution in isolation, the S0 remnants move towards the local TFR, although the initial scatter among them persists. This scatter is sensitive to the indicator used for the rotation velocity: Vcirc values yield a lower scatter than when Vrot,max values are considered instead. In the planes involving Vrot,max, a clear segregation of the S0 remnants in terms of the spin-orbit coupling of the model is observed, in which the remnants of retrograde encounters overlap with local S0s hosting counter-rotating discs. The location of the S0 remnants at $zsim 0$ agrees well with the observed distribution of local S0 galaxies in the $sigma_0$-$M_K$, Vcirc-$sigma_0$ and Vrot,max-$sigma_0$ planes. Thus, massive S0 galaxies may have been formed through major mergers that occurred at high redshift and have later evolved towards the local TFR through passive evolution in relative isolation, a mechanism that would also contribute to the scatter observed in this relation.