Numerous membrane models have been developed and tested for the simulation of membrane processes. However, these models are often either simplified or only validated with a narrow range of experimental data. For the model-based process design of membrane systems it is necessary to have a validated and accurate model which is accurate for the range of possible operating conditions under consideration. Hence, in this study a modeling framework is developed for hollow fiber membranes which can be adjusted systematically to accurately predict the performance of a given membrane. Mixed-gas (CO2/O-2/N-2 and CO2/N-2) separation experiments are carried out over a range of different feed conditions to evaluate membrane performance and to provide reliable measurements of gas permeance. In particular the feed pressure (1-4 bar), permeate pressure (0.1-0.5 bar) and feed flow rates (0.096-0.4 N m(3)/h) are varied in these experiments (the ranges specified in brackets). Interpolation of these measured permeance allows for the accurate prediction of membrane performance at any conditions inside the measured range. A tanks-in-series modeling approach is employed here where the number of tanks (used to represent the membrane behavior in a numerical formulation) can be adjusted to calibrate and fit the membrane model to experimental results. For the membranes tested in this study it is found that using a relatively small number of tanks both minimizes the difference between model and experimental results and reduces the numerical complexity in the membrane model.