The microscopic description of heavy-ion reactions at low beam energies is achieved within hadronic transport approaches. In this article a new approach SMASH (Simulating Many Accelerated Strongly-interacting Hadrons) is introduced and applied to study the production of non-strange particles in heavy-ion reactions at E kin = 0.4 − 2A GeV. First, the model is described including details about the collision criterion, the initial conditions and the resonance formation and decays. To validate the approach, equilibrium properties such as detailed balance are presented and the results are compared to experimental data for elementary cross sections. Finally results for pion and proton production in C+C and Au+Au collisions is confronted with HADES and FOPI data. Predictions for particle production in π + A collisions are made.

The sign change of the slope of the directed flow of baryons has been predicted as a signal for a first order phase transition within fluid dynamical calculations. Recently, the directed flow of identified particles has been measured by the STAR collaboration in the beam energy scan (BES) program. In this article, we examine the collision energy dependence of directed flow v1 in fluid dynamical model descriptions of heavy ion collisions for √ sNN = 3 − 20 GeV. The first step is to reproduce the existing predictions within pure fluid dynamical calculations. As a second step we investigate the influence of the order of the phase transition on the anisotropic flow within a state-ofthe-art hybrid approach that describes other global observables reasonably well. We find that, in the hybrid approach, there seems to be no sensitivity of the directed flow on the equation of state and in particular on the existence of a first order phase transition. In addition, we explore more subtle sensitivities like e.g. the Cooper-Frye transition criterion and discuss how momentum conservation and the definition of the event plane affects the results. At this point, none of our calculations matches qualitatively the behavior of the STAR data, the values of the slopes are always larger than in the data.