In recent years, the membrane technology has emerged as a competitive solution for CO2 removal from natural gas. We present a detailed theoretical model to describe heat and mass transport across a cellulose acetate membrane, accounting for the most important physical phenomena. The model makes it possible to gain deeper insight into the permeation process and allows us to give guidelines for operation and development of membranes for CO2 separation. An equation of state is used in the description of the gas phases to account for the Joule-Thomson effect. We find that this effect causes a small temperature drop across the membrane (similar to 0.005 K), but a considerable temperature drop in the feed stream, amounting to 15 K for a 40 m membrane module. In agreement with the literature, we show that the presence of a support layer does not represent a significant resistance to mass transport, and neglecting it introduces errors below 1% in the mass flux prediction. However, because of its thickness, the support constitutes the largest resistance to heat transport. We use nonequilibrium thermodynamic to account for coupling between heat and mass fluxes in the system. In conventional membranes, the support limits the possibility of using a temperature difference to enhance the performance. With thinner or better conductive support layers, we estimate that it is possible to exploit coupling effects to enhance both permeation and selectivity by 14% and 8% respectively, with a transmembrane temperature difference of 20 K. The possibility of using a thermal driving force to enhance the membrane performance is appealing, as a large amount of waste heat is typically available at the natural gas extraction site. (C) 2016 Elsevier B.V. All rights reserved.