Abstract. In continuation of recent EMC3-Eirene simulations of a series of 'snowflake plus' (SF + ) configurations in TCV [1] we report on simulations with the same code applied to 'snowflake minus' (SF − ) configurations, where the secondary X-point (x 2 ) is located in the common flux region of the primary separatrix on the low-field side of the primary one. While for the SF + the power flux to the secondary strike points (SPs) was only of the order of 1 % in the simulation and ∼ 10 % experimentally, a much higher flux was found for the SF − simulations, consistent with recent measurements [2]. The ratio of total power fluxes P SP 2 /P SP 4 , as well as that of the peak values r 2/4 = q ||,max,SP 2 /q ||,max,SP 4 to the primary and secondary outer SPs (labeled by SP 2 and SP 4), can be tuned by varying the radial magnetic position ρ x2 of x 2 . A reduction of a factor of two is found for q ||,max,SP 2 for the equal power load, r 2/4 = 1, considered as the optimum at ρ x2 ∼ 1.013, which corresponds to a fraction of the power fall-off length λ q at the outboard mid-plane. In addition to these pure deuterium simulations discharges with nitrogen-and neon impurities radiating 20 % of the input power are simulated. Due to an impurity accumulation effect between SP 2 and x 2 more power is radiated on the LFS reducing the power load in particular for the outer SPs. Due to the increase of the outboard mid-plane to target connection length of a factor of two the outer target is expected to detach at lower line-integrated densities compared to the single null (SN) configuration. For all these reasons the maximum tolerable P sep /R is expected to be significantly larger in a LFS SF − compared to a SN.