In this study a variotherm process with ceramic electrical heating elements and a ball filled mold (BFMold®) insert, which is alternately flooded with hot and cold water, was simulated. The injection molding simulation packages Cadmould 3D-F Version 6.0, Moldex3D R11.0, Autodesk Simulation Moldflow Insight 2013 and Sigmasoft Version 5.0 (or Version 5.1) were used to simulate the experimentally applied process. Therefore, nozzle temperature, ceramic element temperature, BFMold® temperature, and holding pressure were systematically varied in 2-level Design of Experiments plans. Their effects on filling, shrinkage and warpage of 2 plate-shaped parts made of POM (Ultraform N2320) were evaluated. The experiments showed a shrinkage of 2 % to 3 %, dominated by the holding pressure, suceeded by the nozzle temperature, the BFMold® temperature and the ceramic element temperature and interactions of those. While Cadmould and Sigmasoft predict 0.5 to 1.2 %, Moldflow calculates 1.7 to 2.2 % and Moldex3D determines 2.1 to 2.6 %. While Moldflow predicts the warpage well, Sigmasoft overestimates and Cadmould underestimates some of the 20 deflection values. Moldex3D considerably undervalues 4 of the deflection values by more than 1 to 2 mm. Cadmould and Moldflow predict the maximal injection pressures of 1200 to 1350 bar very well. While Sigmasoft calculates somehow higher values, Moldex3D determines unrealistic high injection pressures of several thousand bars. Moreover, Cadmould and Moldflow simulate the measured cavity pressure evolution rather well, however Moldex3D overestimates the cooling of the parts considerably. The Virtual and the real filling studies agree very well. However, the partial filling of thin wall areas of 0.5 mm wall thickness cannot be predicted sufficiently with the simulation softwares. The variotherm process with fluid tempering can be simulated rather well, however the simulation of the ceramic heating elements still represents a challenge in some simulation software. In Moldflow an overheating of these occured. Additionally, the BFMold® geometry had to be substituted by a less complex model. Although, injection molding software can simulate variotherm processes sufficiently, software bugs still may lead to misinterpretation of results. Therefore, a critical view on the simulation results by an engineering expert remains necessary.