Abstract: Establishing the productivity model aimed at the percolation mechanisms and development characteristics of shale gas is the key means to realize a reliable prediction of shale gas productivity and guides its reservoir development efficiently. Shale gas has multiple transport mechanisms and multi-scale flow effects in the matrix-fracture network system,which lead to complex mathematical descriptions and difficult coupling solving processes. In order to deal with these problems,considering the characteristics of desorption,diffusion and seepage of shale gas,the coupled model of shale gas seepage in the matrix-fracture system is revised and improved by means of high-pressure isothermal adsorption model and generalized permeability model derived from a series of experiments and combining matrix-fracture fluid channeling model. Then,the PKN model of principal fracture propagation is embedded in the dual-porosity fracturebedrock model to establish an equivalent fracture network model of triple-porosity media,which consists of bedrock,secondary fracture system and principal fracture. By coupling the equivalent fracture network model with the improved seepage model,the productivity prediction considering the multiple flow mechanisms and multi-scale transportation of shale gas has been realized. The case study for the new productivity prediction model shows that the fitting effect of the gas well production data of 1 435 days using the integration curve is satisfactory,revealing the reliability of the application of the new model. Equivalent bedrock-fracture network parameters of the well are also regressed using the proposed model. The new model is used to predict the well productivity for past 20 years. Compared with the contrast model,the predicted production results in the first 10 years are basically the same,and those from 10 to 20 years are gradually deviated. The EUR (estimated ultimate recovery) by the new model is 133 million cubic meters,which is 5% higher than the 127 million cubic meters of the comparative model. The study shows that the new model,which considers the comprehensive seepage mechanisms and coupled fracture network,fully reflects the motion of adsorbed gas and the increasing diffusion capacity for gas wells at low pressures in the later stage of production,and is consistent with the actual production situation of gas wells. The proposed model is,therefore,more reliable for the mid-long term productivity prediction of gas wells. And for shale gas horizontal well production capacity prediction,the new model closely relies on on-site construction and production data,and is characterized by being solved and applied easily.