In this work an extended kinetic analysis involving both experimental measurements and modelling procedures to describe the thermal degradation process of biomass is proposed. Three biomasses belonging to the hardwood family are investigated: ash-wood, beech-wood and hornbeam. The experiments are performed with a thermogravimetric balance working at four constant heating rate: 3, 5, 10 and 20 degrees C/min. This study is specifically dedicated to investigate the thermal behaviour of the selected biomasses when they undergo to torrefaction temperature conditions. The modelling analysis focuses on investigating the capability of consolidated models, usually applied to solids, in determining the activation energy (Ea) of the indicated biomasses within the torrefaction range. The adopted methods belong to the so-called isoconversional "model free" methods and, in this contest, both the differential ones as those of Friedman and Flynn and the integral ones as those of Kissinger-Akahira-Sunose, Doyle and Starink have been applied. The performances reached by adopting the integral methods widely satisfy the accepted accuracy level conventionally set at values lower than 10%. At the same time it is verified even that, when the methods are applied to biomasses belonging to the same family, the resulting Ea vs. a trends are very close for all the biomasses. This condition is exploited to propose a generalized predictive approach for the Ea calculation based on the knowledge of only the conversion fraction a. The presentation of the results includes also the investigation of the limits of the proposed methods in view of indicating their reliable application range when utilized for torrefaction design procedures. (C) 2016 Elsevier Ltd. All rights reserved.