In extrusion operations, material solid state welding takes place thanks to the very high pressure and temperature at which the material undergoes between the several welding criteria developed, the attention was particularly focused on the Piwnik and Plata one. In this criterion a suitable parameter w, calculated as the interface pressure and the effective stress ratio integrated along the time, has been defined. According to this criterion the material should start to weld as this parameter exceed a limit value wlim. In the present paper a new procedure for the wlim identification as function of the temperature, based on coupled experimental-simulative strategy, is proposed. In particular, flat rolling experimental tests of sandwiches made of two rectangular specimens in AA6082 alloy were performed. The specimens were characterized by different heights in order to consider different compression ratio, that means different interface pressure and effective stress distributions. All the tests were repeated at different temperature. Once the experimental tests were performed, FEM simulations of the rolling process were run for the very same conditions. Thanks to a suitable user routine developed and implemented in the calculus code, it was possible to evaluate the steady state value of the w parameter for all the different conditions at the steady state conditions. By verifying if the actual experimental test demonstrated the presence or less of material bonding, it was possible to identify the wlim values as function of the temperature. Particular attention must be paid to the study of the macrostructure of the welded material in order to identify the influence of the process parameter on the weld quality. This means that it is possible to identify not only if the weld will take place, but also if it will be qualitatively adequate.