Accurate calibration of ground-based, mid-infrared observations is challenging due to the strong and rapidly varying thermal background emission. The classical solution is the chopping/nodding technique where the secondary mirror and the telescope are being moved by several tens of arcseconds on the sky. However, chopping is generally inefficient and limited in accuracy and frequency by the mass and size of the secondary mirror. A more elegant solution is a drift scan where the telescope slowly drifts across or around the region of interest; the source moves on the detector by at least one FWHM of the PSF within the time over which the detector performance and the background emission can be considered stable. The final image of a drift scan is mathematically reconstructed from a series of adjacent short exposures. The drift scan approach has recently received a lot of interest, mainly for two reasons: first, some of the new, large-format mid-IR Si:As detectors (AQUARIUS) suffer from excess low frequency noise (ELFN). To reach the nominal performance limit of the detectors, chopping would have to be performed at a high frequency, faster than what most telescopes can handle; second, the next generation of extremely large telescopes will not offer chopping/nodding, and alternative methods need to be developed and tested. In this paper we present the results from simulated drift scan data. We use drift scanning to simultaneously obtain an accurate detector flat field and the sky background. The results are relevant for the future operation and calibration of VISIR at the VLT as well as for METIS, the thermal infrared instrument for the E-ELT.