Context. The Planck Catalogue of Galactic Cold Clumps (PGCC) contains over 13 000 sources that are detected based on their cold dust signature. They are believed to consist of a mixture of quiescent, pre-stellar, and already star-forming objects within molecular clouds. Aims. We extracted PGCC-type objects from cloud simulations and examined their physical and polarisation properties. The comparison with the PGCC catalogue helps to characterise the properties of this large sample of Galactic objects and, conversely, provides valuable tests for numerical simulations of large volumes of the interstellar medium and the evolution towards pre-stellar cores. Methods. We used several magnetohydrodynamical simulation snapshots to define the density field of our model clouds. Sub-millimetre images of the surface brightness and polarised signal were obtained with radiative transfer calculations. We examined the statistics of synthetic cold clump catalogues extracted with methods similar to the PGCC. We also examined the variations of the polarisation fraction p in the clumps. Results. The clump sizes, aspect ratios, and temperatures in the synthetic catalogue are similar to the PGCC. The fluxes and column densities of synthetic clumps are smaller by a factor of a few. Rather than with an increased dust opacity, this could be explained by increasing the average column density of the model by a factor of two to three, close to N(H₂) = 10²² cm⁻². When the line of sight is parallel to the mean magnetic field, the polarisation fraction tends to increase towards the clump centres, which is contrary to observations. When the field is perpendicular, the polarisation fraction tends to decrease towards the clumps, but the drop in p is small (e.g. from p ~8% to p ~7%). Conclusions. Magnetic field geometry reduces the polarisation fraction in the simulated clumps by only Δp ~1% on average. The larger drop seen towards the actual PGCC clumps therefore suggests some loss of grain alignment in the dense medium, such as predicted by the radiative torque mechanism. The statistical study is not able to quantify dust opacity changes at the scale of the PGCC clumps.