Context. Solar coronal mass ejections (CMEs) can catch up and interact with preceding CMEs and solar wind structures to undergo rotation and deflection during their propagation. Aim. We aim to show how the interactions of a CME in the corona and heliosphere can play a significant role in altering its geoeffectiveness as predicted at the time of its eruption. To do so, we consider a case study of two successive CMEs launched from the active region NOAA 12158 in early September 2014. The second CME was predicted to be extensively geoeffective based on the remote-sensing observations of the source region. However, in situ measurements at 1 au recorded only a short-lasting, weak negative B z component followed by a prolonged positive B z component. Methods. We used the EUropean Heliosphere FORecasting Information Asset (EUHFORIA) to perform a self-consistent 3D magnetohydrodynamical (MHD) data-driven simulation of the two CMEs in the heliosphere. First, the ambient solar wind is modelled, followed by the time-dependent injection of CME1 with the LFF spheromak and CME2 with the Flux Rope in 3D (FRi3D) model. The initial conditions of the CMEs are determined by combining observational insights near the Sun, which are fine-tuned to match the in situ observations near 1 au, with additional numerical experiments related to each individual CME. Results. By introducing CME1 before CME2 in the EUHFORIA simulation, we modelled the negative B z component in the sheath region ahead of CME2 whose formation can be attributed to the interaction between CME1 and CME2. To reproduce the positive B z component in the magnetic ejecta of CME2, we had to initialise CME2 with an orientation determined at 0.1 au and consistent with the orientation interpreted at 1 au instead of the orientation observed during its eruption. Conclusions. EUHFORIA simulations suggest the possibility of a significant rotation of CME2 in the low corona in order to explain the in situ observations at 1 au. Coherent magnetic field rotations with enhanced strength (potentially geoeffective) can be formed in the sheath region as a result of interactions between two CMEs in the heliosphere even if the individual CMEs are not geoeffective.